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 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
6 * omp-general.c (omp_max_vf): Likewise.
7 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
8 (expand_omp_simd): Handle polynomial safelen.
9 * omp-low.c (omplow_simd_context): Add a default constructor.
10 (omplow_simd_context::max_vf): Change from int to poly_uint64.
11 (lower_rec_simd_input_clauses): Update accordingly.
12 (lower_rec_input_clauses): Likewise.
14 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
15 Alan Hayward <alan.hayward@arm.com>
16 David Sherwood <david.sherwood@arm.com>
18 * tree-vectorizer.h (vect_nunits_for_cost): New function.
19 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
20 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
21 (vect_analyze_slp_cost): Likewise.
22 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
23 (vect_model_load_cost): Likewise.
25 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26 Alan Hayward <alan.hayward@arm.com>
27 David Sherwood <david.sherwood@arm.com>
29 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
30 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
31 from an unsigned int * to a poly_uint64_pod *.
32 (calculate_unrolling_factor): New function.
33 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
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 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
40 from an unsigned int to a poly_uint64.
41 (_loop_vec_info::slp_unrolling_factor): Likewise.
42 (_loop_vec_info::vectorization_factor): Change from an int
44 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
45 (vect_get_num_vectors): New function.
46 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
47 (vect_get_num_copies): Use vect_get_num_vectors.
48 (vect_analyze_data_ref_dependences): Change max_vf from an int *
50 (vect_analyze_data_refs): Change min_vf from an int * to a
52 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
53 than an unsigned HOST_WIDE_INT.
54 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
55 (vect_analyze_data_ref_dependence): Change max_vf from an int *
57 (vect_analyze_data_ref_dependences): Likewise.
58 (vect_compute_data_ref_alignment): Handle polynomial vf.
59 (vect_enhance_data_refs_alignment): Likewise.
60 (vect_prune_runtime_alias_test_list): Likewise.
61 (vect_shift_permute_load_chain): Likewise.
62 (vect_supportable_dr_alignment): Likewise.
63 (dependence_distance_ge_vf): Take the vectorization factor as a
64 poly_uint64 rather than an unsigned HOST_WIDE_INT.
65 (vect_analyze_data_refs): Change min_vf from an int * to a
67 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
68 vfm1 as a poly_uint64 rather than an int. Make the same change
69 for the returned bound_scalar.
70 (vect_gen_vector_loop_niters): Handle polynomial vf.
71 (vect_do_peeling): Likewise. Update call to
72 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
73 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
75 * tree-vect-loop.c (vect_determine_vectorization_factor)
76 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
77 (vect_get_known_peeling_cost): Likewise.
78 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
79 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
80 (vect_transform_loop): Likewise. Use the lowest possible VF when
81 updating the upper bounds of the loop.
82 (vect_min_worthwhile_factor): Make static. Return an unsigned int
84 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
85 polynomial unroll factors.
86 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
87 (vect_make_slp_decision): Likewise.
88 (vect_supported_load_permutation_p): Likewise, and polynomial
90 (vect_analyze_slp_cost): Handle polynomial vf.
91 (vect_slp_analyze_node_operations): Likewise.
92 (vect_slp_analyze_bb_1): Likewise.
93 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
94 than an unsigned HOST_WIDE_INT.
95 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
96 (vectorizable_load): Handle polynomial vf.
97 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
99 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
101 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
102 Alan Hayward <alan.hayward@arm.com>
103 David Sherwood <david.sherwood@arm.com>
105 * match.pd: Handle bit operations involving three constants
106 and try to fold one pair.
108 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
110 * tree-vect-loop-manip.c: Include gimple-fold.h.
111 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
112 niters_maybe_zero parameters. Handle other cases besides a step of 1.
113 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
114 Add a path that uses a step of VF instead of 1, but disable it
116 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
117 and niters_no_overflow parameters. Update calls to
118 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
119 Create a new SSA name if the latter choses to use a ste other
120 than zero, and return it via niters_vector_mult_vf_var.
121 * tree-vect-loop.c (vect_transform_loop): Update calls to
122 vect_do_peeling, vect_gen_vector_loop_niters and
123 slpeel_make_loop_iterate_ntimes.
124 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
125 (vect_gen_vector_loop_niters): Update declarations after above changes.
127 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
129 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
130 128-bit round to integer instructions.
131 (ceil<mode>2): Likewise.
132 (btrunc<mode>2): Likewise.
133 (round<mode>2): Likewise.
135 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
137 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
138 unaligned VSX load/store on P8/P9.
139 (expand_block_clear): Allow the use of unaligned VSX
142 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
144 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
146 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
147 swap associated with both a load and a store.
149 2018-01-02 Andrew Waterman <andrew@sifive.com>
151 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
152 * config/riscv/riscv.md (clear_cache): Use it.
154 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
156 * web.c: Remove out-of-date comment.
158 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
160 * expr.c (fixup_args_size_notes): Check that any existing
161 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
162 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
163 (emit_single_push_insn): ...here.
165 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
167 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
168 (const_vector_encoded_nelts): New function.
169 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
170 (const_vector_int_elt, const_vector_elt): Declare.
171 * emit-rtl.c (const_vector_int_elt_1): New function.
172 (const_vector_elt): Likewise.
173 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
176 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
178 * expr.c: Include rtx-vector-builder.h.
179 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
180 directly on the tree encoding.
181 (const_vector_from_tree): Likewise.
182 * optabs.c: Include rtx-vector-builder.h.
183 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
184 sequence of "u" values.
185 * vec-perm-indices.c: Include rtx-vector-builder.h.
186 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
187 directly on the vec_perm_indices encoding.
189 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
191 * doc/rtl.texi (const_vector): Describe new encoding scheme.
192 * Makefile.in (OBJS): Add rtx-vector-builder.o.
193 * rtx-vector-builder.h: New file.
194 * rtx-vector-builder.c: Likewise.
195 * rtl.h (rtx_def::u2): Add a const_vector field.
196 (CONST_VECTOR_NPATTERNS): New macro.
197 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
198 (CONST_VECTOR_DUPLICATE_P): Likewise.
199 (CONST_VECTOR_STEPPED_P): Likewise.
200 (CONST_VECTOR_ENCODED_ELT): Likewise.
201 (const_vec_duplicate_p): Check for a duplicated vector encoding.
202 (unwrap_const_vec_duplicate): Likewise.
203 (const_vec_series_p): Check for a non-duplicated vector encoding.
204 Say that the function only returns true for integer vectors.
205 * emit-rtl.c: Include rtx-vector-builder.h.
206 (gen_const_vec_duplicate_1): Delete.
207 (gen_const_vector): Call gen_const_vec_duplicate instead of
208 gen_const_vec_duplicate_1.
209 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
210 (gen_const_vec_duplicate): Use rtx_vector_builder.
211 (gen_const_vec_series): Likewise.
212 (gen_rtx_CONST_VECTOR): Likewise.
213 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
214 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
215 Build a new vector rather than modifying a CONST_VECTOR in-place.
216 (handle_special_swappables): Update call accordingly.
217 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
218 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
219 Build a new vector rather than modifying a CONST_VECTOR in-place.
220 (handle_special_swappables): Update call accordingly.
222 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
224 * simplify-rtx.c (simplify_const_binary_operation): Use
225 CONST_VECTOR_ELT instead of XVECEXP.
227 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
229 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
230 the selector elements to be different from the data elements
231 if the selector is a VECTOR_CST.
232 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
233 ssizetype for the selector.
235 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
237 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
238 before testing each element individually.
239 * tree-vect-generic.c (lower_vec_perm): Likewise.
241 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
243 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
244 * selftest-run-tests.c (selftest::run_tests): Call it.
245 * vector-builder.h (vector_builder::operator ==): New function.
246 (vector_builder::operator !=): Likewise.
247 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
248 (vec_perm_indices::all_from_input_p): New function.
249 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
250 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
251 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
252 instead of reading the VECTOR_CST directly. Detect whether both
253 vector inputs are the same before constructing the vec_perm_indices,
254 and update the number of inputs argument accordingly. Use the
255 utility functions added above. Only construct sel2 if we need to.
257 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
259 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
260 the broadcast of the low byte.
261 (expand_mult_highpart): Use an explicit encoding for the permutes.
262 * optabs-query.c (can_mult_highpart_p): Likewise.
263 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
264 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
265 (vectorizable_bswap): Likewise.
266 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
267 explicit encoding for the power-of-2 permutes.
268 (vect_permute_store_chain): Likewise.
269 (vect_grouped_load_supported): Likewise.
270 (vect_permute_load_chain): Likewise.
272 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
274 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
275 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
276 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
277 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
278 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
279 (vect_gen_perm_mask_any): Likewise.
281 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
283 * int-vector-builder.h: New file.
284 * vec-perm-indices.h: Include int-vector-builder.h.
285 (vec_perm_indices): Redefine as an int_vector_builder.
286 (auto_vec_perm_indices): Delete.
287 (vec_perm_builder): Redefine as a stand-alone class.
288 (vec_perm_indices::vec_perm_indices): New function.
289 (vec_perm_indices::clamp): Likewise.
290 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
291 (vec_perm_indices::new_vector): New function.
292 (vec_perm_indices::new_expanded_vector): Update for new
293 vec_perm_indices class.
294 (vec_perm_indices::rotate_inputs): New function.
295 (vec_perm_indices::all_in_range_p): Operate directly on the
296 encoded form, without computing elided elements.
297 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
298 encoding. Update for new vec_perm_indices class.
299 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
300 the given vec_perm_builder.
301 (expand_vec_perm_var): Update vec_perm_builder constructor.
302 (expand_mult_highpart): Use vec_perm_builder instead of
303 auto_vec_perm_indices.
304 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
305 vec_perm_indices instead of auto_vec_perm_indices. Use a single
306 or double series encoding as appropriate.
307 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
308 vec_perm_indices instead of auto_vec_perm_indices.
309 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
310 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
311 (vect_permute_store_chain): Likewise.
312 (vect_grouped_load_supported): Likewise.
313 (vect_permute_load_chain): Likewise.
314 (vect_shift_permute_load_chain): Likewise.
315 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
316 (vect_transform_slp_perm_load): Likewise.
317 (vect_schedule_slp_instance): Likewise.
318 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
319 (vectorizable_mask_load_store): Likewise.
320 (vectorizable_bswap): Likewise.
321 (vectorizable_store): Likewise.
322 (vectorizable_load): Likewise.
323 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
324 vec_perm_indices instead of auto_vec_perm_indices. Use
325 tree_to_vec_perm_builder to read the vector from a tree.
326 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
327 vec_perm_builder instead of a vec_perm_indices.
328 (have_whole_vector_shift): Use vec_perm_builder and
329 vec_perm_indices instead of auto_vec_perm_indices. Leave the
330 truncation to calc_vec_perm_mask_for_shift.
331 (vect_create_epilog_for_reduction): Likewise.
332 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
333 from auto_vec_perm_indices to vec_perm_indices.
334 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
335 instead of changing individual elements.
336 (aarch64_vectorize_vec_perm_const): Use new_vector to install
337 the vector in d.perm.
338 * config/arm/arm.c (expand_vec_perm_d::perm): Change
339 from auto_vec_perm_indices to vec_perm_indices.
340 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
341 instead of changing individual elements.
342 (arm_vectorize_vec_perm_const): Use new_vector to install
343 the vector in d.perm.
344 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
345 Update vec_perm_builder constructor.
346 (rs6000_expand_interleave): Likewise.
347 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
348 (rs6000_expand_interleave): Likewise.
350 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
352 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
353 to qimode could truncate the indices.
354 * optabs.c (expand_vec_perm_var): Likewise.
356 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
358 * Makefile.in (OBJS): Add vec-perm-indices.o.
359 * vec-perm-indices.h: New file.
360 * vec-perm-indices.c: Likewise.
361 * target.h (vec_perm_indices): Replace with a forward class
363 (auto_vec_perm_indices): Move to vec-perm-indices.h.
364 * optabs.h: Include vec-perm-indices.h.
365 (expand_vec_perm): Delete.
366 (selector_fits_mode_p, expand_vec_perm_var): Declare.
367 (expand_vec_perm_const): Declare.
368 * target.def (vec_perm_const_ok): Replace with...
369 (vec_perm_const): ...this new hook.
370 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
371 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
372 * doc/tm.texi: Regenerate.
373 * optabs.def (vec_perm_const): Delete.
374 * doc/md.texi (vec_perm_const): Likewise.
375 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
376 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
377 expand_vec_perm for constant permutation vectors. Assert that
378 the mode of variable permutation vectors is the integer equivalent
379 of the mode that is being permuted.
380 * optabs-query.h (selector_fits_mode_p): Declare.
381 * optabs-query.c: Include vec-perm-indices.h.
382 (selector_fits_mode_p): New function.
383 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
384 is defined, instead of checking whether the vec_perm_const_optab
385 exists. Use targetm.vectorize.vec_perm_const instead of
386 targetm.vectorize.vec_perm_const_ok. Check whether the indices
387 fit in the vector mode before using a variable permute.
388 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
389 vec_perm_indices instead of an rtx.
390 (expand_vec_perm): Replace with...
391 (expand_vec_perm_const): ...this new function. Take the selector
392 as a vec_perm_indices rather than an rtx. Also take the mode of
393 the selector. Update call to shift_amt_for_vec_perm_mask.
394 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
395 Use vec_perm_indices::new_expanded_vector to expand the original
396 selector into bytes. Check whether the indices fit in the vector
397 mode before using a variable permute.
398 (expand_vec_perm_var): Make global.
399 (expand_mult_highpart): Use expand_vec_perm_const.
400 * fold-const.c: Includes vec-perm-indices.h.
401 * tree-ssa-forwprop.c: Likewise.
402 * tree-vect-data-refs.c: Likewise.
403 * tree-vect-generic.c: Likewise.
404 * tree-vect-loop.c: Likewise.
405 * tree-vect-slp.c: Likewise.
406 * tree-vect-stmts.c: Likewise.
407 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
409 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
410 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
411 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
412 (aarch64_vectorize_vec_perm_const): ...this new function.
413 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
414 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
415 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
416 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
417 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
418 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
419 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
421 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
422 check for NEON modes.
423 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
424 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
425 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
426 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
428 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
429 the old VEC_PERM_CONST conditions.
430 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
431 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
432 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
433 (ia64_vectorize_vec_perm_const_ok): Merge into...
434 (ia64_vectorize_vec_perm_const): ...this new function.
435 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
436 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
437 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
438 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
439 * config/mips/mips.c (mips_expand_vec_perm_const)
440 (mips_vectorize_vec_perm_const_ok): Merge into...
441 (mips_vectorize_vec_perm_const): ...this new function.
442 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
443 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
444 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
445 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
446 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
447 (rs6000_expand_vec_perm_const): Delete.
448 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
450 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
451 (altivec_expand_vec_perm_const_le): Take each operand individually.
452 Operate on constant selectors rather than rtxes.
453 (altivec_expand_vec_perm_const): Likewise. Update call to
454 altivec_expand_vec_perm_const_le.
455 (rs6000_expand_vec_perm_const): Delete.
456 (rs6000_vectorize_vec_perm_const_ok): Delete.
457 (rs6000_vectorize_vec_perm_const): New function.
458 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
459 an element count and rtx array.
460 (rs6000_expand_extract_even): Update call accordingly.
461 (rs6000_expand_interleave): Likewise.
462 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
463 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
464 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
465 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
466 (rs6000_expand_vec_perm_const): Delete.
467 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
468 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
469 (altivec_expand_vec_perm_const_le): Take each operand individually.
470 Operate on constant selectors rather than rtxes.
471 (altivec_expand_vec_perm_const): Likewise. Update call to
472 altivec_expand_vec_perm_const_le.
473 (rs6000_expand_vec_perm_const): Delete.
474 (rs6000_vectorize_vec_perm_const_ok): Delete.
475 (rs6000_vectorize_vec_perm_const): New function. Remove stray
476 reference to the SPE evmerge intructions.
477 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
478 an element count and rtx array.
479 (rs6000_expand_extract_even): Update call accordingly.
480 (rs6000_expand_interleave): Likewise.
481 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
482 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
484 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
486 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
488 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
489 vector mode and that that mode matches the mode of the data
491 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
492 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
493 directly using expand_vec_perm_1 when forcing selectors into
495 (expand_vec_perm_var): New function, split out from expand_vec_perm.
497 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
499 * optabs-query.h (can_vec_perm_p): Delete.
500 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
501 * optabs-query.c (can_vec_perm_p): Split into...
502 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
503 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
504 particular selector is valid.
505 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
506 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
507 (vect_grouped_load_supported): Likewise.
508 (vect_shift_permute_load_chain): Likewise.
509 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
510 (vect_transform_slp_perm_load): Likewise.
511 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
512 (vectorizable_bswap): Likewise.
513 (vect_gen_perm_mask_checked): Likewise.
514 * fold-const.c (fold_ternary_loc): Likewise. Don't take
515 implementations of variable permutation vectors into account
516 when deciding which selector to use.
517 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
518 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
519 with a false third argument.
520 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
521 to test whether the constant selector is valid and can_vec_perm_var_p
522 to test whether a variable selector is valid.
524 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
526 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
527 * optabs-query.c (can_vec_perm_p): Likewise.
528 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
529 instead of vec_perm_indices.
530 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
531 (vect_gen_perm_mask_checked): Likewise,
532 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
533 (vect_gen_perm_mask_checked): Likewise,
535 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
537 * optabs-query.h (qimode_for_vec_perm): Declare.
538 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
539 (qimode_for_vec_perm): ...this new function.
540 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
542 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
544 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
545 does not have a conditional at the top.
547 2018-01-02 Richard Biener <rguenther@suse.de>
549 * ipa-inline.c (big_speedup_p): Fix expression.
551 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
554 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
557 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
561 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
562 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
563 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
564 cond_taken_branch_cost 3->4.
566 2018-01-01 Jakub Jelinek <jakub@redhat.com>
568 PR tree-optimization/83581
569 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
570 TODO_cleanup_cfg if any changes have been made.
573 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
574 convert_modes if target mode has the right side, but different mode
578 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
579 last argument when extracting from CONCAT. If either from_real or
580 from_imag is NULL, use expansion through memory. If result is not
581 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
582 the parts directly to inner mode, if even that fails, use expansion
586 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
587 check for bswap in mode rather than HImode and use that in expand_unop
590 2017-12-31 Jakub Jelinek <jakub@redhat.com>
593 * config/i386/i386.c (ix86_attribute_table): Remove excess
594 initializer for "shared" attribute.
596 2017-12-30 Tom de Vries <tom@codesourcery.com>
599 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
601 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
602 offload_funcs entries that no longer have a corresponding cgraph_node.
603 Mark the remaining ones as DECL_PRESERVE_P.
604 (output_lto): Call prune_offload_funcs.
606 2017-12-30 Jakub Jelinek <jakub@redhat.com>
608 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
609 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
610 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
611 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
612 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
613 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
614 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
615 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
616 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
617 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
618 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
619 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
620 vaesenc_<mode>, vpclmulqdq_<mode>,
621 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
623 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
625 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
626 _Float<N> and _Float<N>X variants for rounding built-in
628 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
629 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
630 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
631 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
632 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
633 * builtins.c (mathfn_built_in_2): Likewise.
634 * internal-fn.def (CEIL): Likewise.
636 (NEARBYINT): Likewise.
640 * convert.c (convert_to_integer_1): Likewise.
641 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
642 (integer_valued_real_call_p): Likewise.
643 * fold-const-call.c (fold_const_call_ss): Likewise.
644 * gencfn-macros.c (print_case_cfn): Change CFN and operator
645 printers to take a const char * suffix instead of a bool.
646 (print_define_operator_list): Likewise.
647 (fltall_suffixes): New list of suffixes, that include the
648 traditional suffixes as well as all of the _Float<N> and
650 (main): For _Float<N> and _Float<N>X functions, emit both
651 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
652 has the _Float<N> and _Float<N>X case names or operators. The
653 <name>_ALL variant has both the traditional and the
654 _Float<N>/_Float<N>X case names or operators.
655 * match.pd (COPYSIGN optimizations): Provide optimizations for
656 _Float<N> and _Float<N>X types where possible.
657 (MIN/MAX optimizations): Likewise.
658 (sqrt optimizations): Likewise.
659 (rounding optimizations): Likewise.
661 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
663 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
664 gen_int_mode rather than GEN_INT.
665 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
667 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
669 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
670 instead of CONSTANT_P.
671 (gen_vec_series): Likewise.
672 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
674 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
676 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
679 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
681 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
682 (aarch64_ld1x2<VDC:mode>): Likewise.
683 (aarch64_simd_ld1<mode>_x2): Likewise.
684 (aarch64_simd_ld1<mode>_x2): Likewise.
685 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
686 (vld1_s8_x2): Likewise.
687 (vld1_u16_x2): Likewise.
688 (vld1_s16_x2): Likewise.
689 (vld1_u32_x2): Likewise.
690 (vld1_s32_x2): Likewise.
691 (vld1_u64_x2): Likewise.
692 (vld1_s64_x2): Likewise.
693 (vld1_f16_x2): Likewise.
694 (vld1_f32_x2): Likewise.
695 (vld1_f64_x2): Likewise.
696 (vld1_p8_x2): Likewise.
697 (vld1_p16_x2): Likewise.
698 (vld1_p64_x2): Likewise.
699 (vld1q_u8_x2): Likewise.
700 (vld1q_s8_x2): Likewise.
701 (vld1q_u16_x2): Likewise.
702 (vld1q_s16_x2): Likewise.
703 (vld1q_u32_x2): Likewise.
704 (vld1q_s32_x2): Likewise.
705 (vld1q_u64_x2): Likewise.
706 (vld1q_s64_x2): Likewise.
707 (vld1q_f16_x2): Likewise.
708 (vld1q_f32_x2): Likewise.
709 (vld1q_f64_x2): Likewise.
710 (vld1q_p8_x2): Likewise.
711 (vld1q_p16_x2): Likewise.
712 (vld1q_p64_x2): Likewise.
714 2017-12-27 Martin Liska <mliska@suse.cz>
716 PR tree-optimization/83552
717 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
718 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
720 2017-12-27 Tom de Vries <tom@codesourcery.com>
722 * config/nvptx/nvptx.c (nvptx_option_override): Disable
723 -gstatement-frontiers.
725 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
727 PR rtl-optimization/83513
728 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
729 before priority comparison.
731 2017-12-25 Jakub Jelinek <jakub@redhat.com>
734 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
735 ix86_isa_flags2 to ix86_isa_flags.
736 * config/i386/i386-c.c (ix86_target_macros_internal): Test
737 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
738 isa_flags rather than isa_flags2.
739 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
740 and -mavx512bitalg from isa2_opts to isa_opts.
741 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
742 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
743 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
745 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
746 OPTION_MASK_ISA_AVX512F ored with another option similarly to
747 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
748 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
749 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
750 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
751 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
752 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
753 builtins from bdesc_args2 to bdesc_args section.
754 (__builtin_ia32_compressstoreuqi512_mask,
755 __builtin_ia32_compressstoreuhi512_mask,
756 __builtin_ia32_compressstoreuqi256_mask,
757 __builtin_ia32_expandloadqi512_mask,
758 __builtin_ia32_expandloadqi512_maskz,
759 __builtin_ia32_expandloadhi512_mask,
760 __builtin_ia32_expandloadhi512_maskz,
761 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
762 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
763 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
764 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
765 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
766 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
767 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
768 __builtin_ia32_vpshldv_v32hi_maskz,
769 __builtin_ia32_vpopcountb_v64qi_mask,
770 __builtin_ia32_vpopcountw_v32hi_mask,
771 __builtin_ia32_vpshufbitqmb512_mask,
772 __builtin_ia32_vpshufbitqmb256_mask): Add
773 " | OPTION_MASK_ISA_AVX512BW".
774 (__builtin_ia32_expandloadqi256_mask,
775 __builtin_ia32_expandloadqi256_maskz,
776 __builtin_ia32_vpopcountb_v32qi_mask): Add
777 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
778 (__builtin_ia32_expandloadhi256_mask,
779 __builtin_ia32_expandloadhi256_maskz,
780 __builtin_ia32_expandloadqi128_mask,
781 __builtin_ia32_expandloadqi128_maskz,
782 __builtin_ia32_expandloadhi128_mask,
783 __builtin_ia32_expandloadhi128_maskz,
784 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
785 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
786 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
787 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
788 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
789 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
790 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
791 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
792 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
793 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
794 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
795 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
796 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
797 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
798 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
799 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
800 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
801 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
802 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
803 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
804 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
805 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
806 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
807 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
808 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
809 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
810 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
811 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
812 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
813 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
814 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
815 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
816 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
817 __builtin_ia32_vpopcountw_v8hi_mask): Add
818 " | OPTION_MASK_ISA_AVX512VL".
819 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
820 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
821 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
822 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
823 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
824 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
825 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
826 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
827 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
828 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
829 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
830 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
831 _mm_mask_bitshuffle_epi64_mask): Likewise.
832 * common/config/i386/i386-common.c
833 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
834 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
835 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
836 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
837 OPTION_MASK_ISA_AVX512BITALG_UNSET.
838 (OPTION_MASK_ISA2_AVX512F_UNSET,
839 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
840 (ix86_handle_option): For -mno-general-regs-only, clear from
841 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
842 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
843 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
844 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
845 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
846 for moving from ix86_isa_flags2 to ix86_isa_flags.
848 2017-12-23 Jakub Jelinek <jakub@redhat.com>
851 * fold-const.c (struct contains_label_data): New type.
852 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
853 inside of a SWITCH_BODY seen during the walk.
854 (contains_label_p): Use walk_tree instead of
855 walk_tree_without_duplicates, prepare data for contains_label_1 and
858 2017-12-22 Jakub Jelinek <jakub@redhat.com>
861 * tree-iterator.c (alloc_stmt_list): Start with cleared
862 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
866 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
867 enable avx512f explicitly in #pragma GCC target.
868 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
869 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
870 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
871 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
872 __builtin_ia32_vpdpbusds_v8si_mask,
873 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
874 __builtin_ia32_vpdpbusds_v4si_mask,
875 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
876 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
877 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
878 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
879 __builtin_ia32_vpdpwssds_v8si_mask,
880 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
881 __builtin_ia32_vpdpwssds_v4si_mask,
882 __builtin_ia32_vpdpwssds_v4si_maskz): Use
883 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
884 just OPTION_MASK_ISA_AVX512VNNI.
886 2017-12-22 Martin Jambor <mjambor@suse.cz>
889 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
892 2017-12-22 Julia Koval <julia.koval@intel.com>
893 Sebastian Peryt <sebastian.peryt@intel.com>
895 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
896 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
897 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
898 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
899 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
900 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
901 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
902 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
903 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
904 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
905 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
906 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
907 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
908 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
909 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
910 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
911 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
912 _mm256_maskz_popcnt_epi64): New intrinsics.
913 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
914 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
915 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
916 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
917 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
918 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
919 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
920 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
921 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
922 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
923 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
924 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
925 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
926 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
927 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
928 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
929 __builtin_ia32_vpshufbitqmb256_mask,
930 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
931 * config/i386/i386-c.c (__AVX512BITALG__): New.
932 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
933 (ix86_valid_target_attribute_inner_p): Ditto.
934 (ix86_expand_args_builtin): Handle new types.
935 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
936 * config/i386/i386.opt: Add -mavx512bitalg.
937 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
938 avx512bitalgintrin.h.
939 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
940 (vpopcount<mode><mask_name>): Add more types.
941 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
942 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
944 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
946 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
947 Or in OPTION_MASK_ISA_AVX512F_SET.
948 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
949 OPTION_MASK_ISA_AVX512VNNI_UNSET.
950 (ix86_handle_option): Adjust for
951 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
952 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
954 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
955 isa_flag instead of isa_flag2.
956 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
957 isa_opts2 to isa_opts.
958 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
961 2017-12-22 Mike Stump <mikestump@comcast.net>
962 Eric Botcazou <ebotcazou@adacore.com>
964 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
966 2017-12-21 Jakub Jelinek <jakub@redhat.com>
969 * config/i386/i386.c (ix86_function_arg_boundary): Return
970 PARM_BOUNDARY for TYPE_EMPTY_P types.
973 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
974 if navail is >= dir.len.
976 2017-12-21 Steve Ellcey <sellcey@cavium.com>
978 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
981 2017-12-21 Jakub Jelinek <jakub@redhat.com>
983 PR rtl-optimization/80747
984 PR rtl-optimization/83512
985 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
986 succ edge from ENTRY, copy partition from e->dest to the newly
988 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
989 ENTRY, use BB_PARTITION of its successor block as current_partition.
990 Don't copy partition when splitting succ edge from ENTRY.
992 PR tree-optimization/83523
993 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
994 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
995 (convert_mult_to_fma): Likewise.
997 PR tree-optimization/83521
998 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
999 gimple_build_assign without code on result of
1000 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1001 a VIEW_CONVERT_EXPR.
1003 2017-12-21 Andrew Pinski <apinski@cavium.com>
1004 Steve Ellcey <sellcey@cavium.com>
1006 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1007 multi-arch for ilp32.
1009 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1012 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1013 constraints to operand 2.
1014 (*ashl<mode>3_mask_1): Ditto.
1015 (*<shift_insn><mode>3_mask): Ditto.
1016 (*<shift_insn><mode>3_mask_1): Ditto.
1017 (*<rotate_insn><mode>3_mask): Ditto.
1018 (*<rotate_insn><mode>3_mask_1): Ditto.
1020 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1022 * reorg.c (make_return_insns): Reemit each insn with its own location.
1024 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1026 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1027 generation for cases where splatting a value is not useful.
1028 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1029 across a vec_duplicate and a paradoxical subreg forming a vector
1030 mode to a vec_concat.
1032 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1034 * combine.c (simplify_set): Do not transform subregs to zero_extends
1035 if the destination is not a scalar int mode.
1037 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1040 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1041 shared zero if the input has overflowed.
1043 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1045 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1046 support for Cortex-A55 and Cortex-A75.
1048 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1050 * common/config/arm/arm-common.c (compare_opt_names): Add function
1051 comment. Use strcmp instead of manual loop.
1053 2017-12-21 Martin Liska <mliska@suse.cz>
1055 PR gcov-profile/83509
1056 * gcov-dump.c (dump_gcov_file): Do not read info about
1057 support_unexecuted_blocks for gcda files.
1059 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1061 PR rtl-optimization/82973
1062 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1063 (valid_for_const_vector_p): ... this.
1064 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1065 (valid_for_const_vector_p): ... this. Adjust function comment.
1066 (gen_vec_duplicate): Adjust caller.
1067 * optabs.c (expand_vector_broadcast): Likewise.
1068 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1069 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1071 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1072 instead of GET_CODE == CONST_FIXED.
1073 (simplify_subreg): Use CONST_FIXED_P macro instead of
1074 GET_CODE == CONST_FIXED.
1077 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1078 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1079 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1080 (ix86_option_override_internal): Adjust for
1081 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1082 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1083 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1084 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1085 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1086 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1087 def_builtin2 for CET builtins.
1088 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1089 up formatting in IX86_BUILTIN_RDPID code.
1090 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1091 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1092 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1093 ix86_isa_flags2 to ix86_isa_flags.
1094 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1096 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1097 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1098 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1100 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1101 Or in OPTION_MASK_ISA_AVX512F_SET.
1102 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1103 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1104 (ix86_handle_option): Adjust for
1105 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1106 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1108 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1109 Alan Hayward <alan.hayward@arm.com>
1110 David Sherwood <david.sherwood@arm.com>
1112 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1113 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1114 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1115 Track polynomial offsets.
1117 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1118 Alan Hayward <alan.hayward@arm.com>
1119 David Sherwood <david.sherwood@arm.com>
1121 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1122 Treat drb->init as a poly_int. Fail if its misalignment wrt
1123 vector_alignment isn't known.
1125 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1126 Alan Hayward <alan.hayward@arm.com>
1127 David Sherwood <david.sherwood@arm.com>
1129 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1131 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1132 (vect_loop_versioning): Take the loop versioning threshold as a
1134 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1135 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1136 versioning_threshold.
1137 (vect_analyze_loop_2): Compute the loop versioning threshold
1138 whenever loop versioning is needed, and store it in the new
1139 field rather than combining it with the cost model threshold.
1140 (vect_transform_loop): Update call to vect_loop_versioning.
1141 Try to combine the loop versioning and cost thresholds here.
1143 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1144 Alan Hayward <alan.hayward@arm.com>
1145 David Sherwood <david.sherwood@arm.com>
1147 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1148 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1149 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1150 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1151 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1152 (group_compare_offset): Update accordingly.
1153 (split_small_address_groups_p): Likewise.
1154 (record_use): Take addr_offset as a poly_uint64 rather than
1155 an unsigned HOST_WIDE_INT.
1156 (strip_offset): Return the offset as a poly_uint64 rather than
1157 an unsigned HOST_WIDE_INT.
1158 (record_group_use, split_address_groups): Track polynomial offsets.
1159 (add_iv_candidate_for_use): Likewise.
1160 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1161 than a HOST_WIDE_INT.
1162 (strip_offset_1): Return the offset as a poly_int64 rather than
1165 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1166 Alan Hayward <alan.hayward@arm.com>
1167 David Sherwood <david.sherwood@arm.com>
1169 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1170 rather than a HOST_WIDE_INT.
1171 * tree.c (get_binfo_at_offset): Likewise.
1173 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1174 Alan Hayward <alan.hayward@arm.com>
1175 David Sherwood <david.sherwood@arm.com>
1177 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1178 rather than a HOST_WIDE_INT.
1179 * tree-sra.c (build_ref_for_offset): Likewise.
1181 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1182 Alan Hayward <alan.hayward@arm.com>
1183 David Sherwood <david.sherwood@arm.com>
1185 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1187 * tree.c (mem_ref_offset): Likewise.
1188 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1189 * builtins.c (get_object_alignment_2): Likewise.
1190 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1191 * gimple-fold.c (get_base_constructor): Likewise.
1192 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1193 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1195 * ipa-polymorphic-call.c
1196 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1197 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1198 (get_ancestor_addr_info): Likewise.
1199 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1200 * match.pd: Likewise.
1201 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1202 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1203 * tree-eh.c (tree_could_trap_p): Likewise.
1204 * tree-object-size.c (addr_object_size): Likewise.
1205 * tree-ssa-address.c (copy_ref_info): Likewise.
1206 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1207 (indirect_refs_may_alias_p): Likewise.
1208 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1209 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1210 (non_rewritable_mem_ref_base): Likewise.
1211 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1212 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1213 * varasm.c (decode_addr_const): Likewise.
1215 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1216 Alan Hayward <alan.hayward@arm.com>
1217 David Sherwood <david.sherwood@arm.com>
1219 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1220 offsets for MEM_REFs.
1222 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1223 Alan Hayward <alan.hayward@arm.com>
1224 David Sherwood <david.sherwood@arm.com>
1226 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1227 as a poly_uint64 rather than an unsigned int.
1228 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1230 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1231 Alan Hayward <alan.hayward@arm.com>
1232 David Sherwood <david.sherwood@arm.com>
1234 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1236 (decode_addr_const): Update accordingly.
1238 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1239 Alan Hayward <alan.hayward@arm.com>
1240 David Sherwood <david.sherwood@arm.com>
1242 * tree.h (bit_field_size, bit_field_offset): New functions.
1243 * hsa-gen.c (gen_hsa_addr): Use them.
1244 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1245 (simplify_vector_constructor): Likewise.
1246 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1247 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1248 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1249 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1252 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1253 Alan Hayward <alan.hayward@arm.com>
1254 David Sherwood <david.sherwood@arm.com>
1256 * expr.h (emit_group_load, emit_group_load_into_temps)
1257 (emit_group_store): Take the size as a poly_int64 rather than an int.
1258 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1259 (emit_group_load_into_temp, emit_group_store): Likewise.
1261 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1262 Alan Hayward <alan.hayward@arm.com>
1263 David Sherwood <david.sherwood@arm.com>
1265 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1266 an unsigned int to a poly_uint64.
1267 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1268 sizes as poly_uint64s rather than unsigned ints.
1269 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1272 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1273 Alan Hayward <alan.hayward@arm.com>
1274 David Sherwood <david.sherwood@arm.com>
1276 * lra-constraints.c (emit_inc): Change inc_amount from an int
1279 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1280 Alan Hayward <alan.hayward@arm.com>
1281 David Sherwood <david.sherwood@arm.com>
1283 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1285 (add_stack_var, stack_var_cmp, partition_stack_vars)
1286 (dump_stack_var_partition): Update accordingly.
1287 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1288 than a HOST_WIDE_INT.
1289 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1290 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1291 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1292 rather than a HOST_WIDE_INT.
1294 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1295 Alan Hayward <alan.hayward@arm.com>
1296 David Sherwood <david.sherwood@arm.com>
1298 * target.def (return_pops_args): Treat both the input and output
1299 sizes as poly_int64s rather than HOST_WIDE_INTS.
1300 * targhooks.h (default_return_pops_args): Update accordingly.
1301 * targhooks.c (default_return_pops_args): Likewise.
1302 * doc/tm.texi: Regenerate.
1303 * emit-rtl.h (incoming_args): Change pops_args, size and
1304 outgoing_args_size from int to poly_int64_pod.
1305 * function.h (expr_status): Change x_pending_stack_adjust and
1306 x_stack_pointer_delta from int to poly_int64.
1307 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1308 (ARGS_SIZE_RTX): Update accordingly.
1309 * calls.c (highest_outgoing_arg_in_use): Change from int to
1311 (stack_usage_watermark, stored_args_watermark): New variables.
1312 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1313 (emit_call_1): Change the stack_size and rounded_stack_size
1314 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1316 (save_fixed_argument_area): Check stack_usage_watermark.
1317 (initialize_argument_information): Change old_pending_adj from
1318 a HOST_WIDE_INT * to a poly_int64_pod *.
1319 (compute_argument_block_size): Return the size as a poly_int64
1321 (finalize_must_preallocate): Track polynomial argument sizes.
1322 (compute_argument_addresses): Likewise.
1323 (internal_arg_pointer_based_exp): Track polynomial offsets.
1324 (mem_overlaps_already_clobbered_arg_p): Rename to...
1325 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1326 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1327 Check stored_args_used_watermark.
1328 (load_register_parameters): Update accordingly.
1329 (check_sibcall_argument_overlap_1): Likewise.
1330 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1331 args size as a poly_int64 rather than an int. Return a bool
1332 indicating whether the optimization was possible and return
1333 the new adjustment by reference.
1334 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1335 Update stored_args_watermark.
1336 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1337 (expand_call): Likewise. Maintain stack_usage_watermark and
1338 stored_args_watermark. Update calls to
1339 combine_pending_stack_adjustment_and_call.
1340 (emit_library_call_value_1): Handle polynomial argument sizes.
1341 Call stack_region_maybe_used_p and mark_stack_region_used.
1342 Maintain stack_usage_watermark.
1343 (store_one_arg): Likewise. Update call to
1344 mem_overlaps_already_clobbered_arg_p.
1345 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1347 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1348 * config/microblaze/microblaze.c (microblaze_function_prologue):
1350 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1351 TARGET_RETURN_POPS_ARGS interface.
1352 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1354 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1355 * config/i386/i386.c (ix86_return_pops_args): Update for new
1356 TARGET_RETURN_POPS_ARGS interface.
1357 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1358 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1359 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1360 TARGET_RETURN_POPS_ARGS interface.
1361 * config/vax/vax.c (vax_return_pops_args): Likewise.
1362 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1363 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1364 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1365 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1366 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1367 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1368 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1369 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1370 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1371 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1372 and x_stack_pointer_delta from int to poly_int64.
1373 * dojump.c (do_pending_stack_adjust): Update accordingly.
1374 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1375 stack_pointer_deltas.
1376 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1377 (pad_to_arg_alignment): Track polynomial offsets.
1378 (assign_parm_find_stack_rtl): Likewise.
1379 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1380 * toplev.c (output_stack_usage): Update reference to
1381 current_function_pushed_stack_size.
1383 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1384 Alan Hayward <alan.hayward@arm.com>
1385 David Sherwood <david.sherwood@arm.com>
1387 * function.c (in_arg_offset, var_offset, dynamic_offset)
1388 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1389 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1390 rather than a HOST_WIDE_INT.
1391 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1392 (instantiate_virtual_regs_in_insn): Likewise.
1394 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1395 Alan Hayward <alan.hayward@arm.com>
1396 David Sherwood <david.sherwood@arm.com>
1398 * rtl.h (get_args_size, add_args_size_note): New functions.
1399 (find_args_size_adjust): Return a poly_int64 rather than a
1401 (fixup_args_size_notes): Likewise. Make the same change to the
1402 end_args_size parameter.
1403 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1404 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1405 * calls.c (emit_call_1): Likewise.
1406 * explow.c (adjust_stack_1): Likewise.
1407 * cfgcleanup.c (old_insns_match_p): Update use of
1408 find_args_size_adjust.
1409 * combine.c (distribute_notes): Track polynomial arg sizes.
1410 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1411 end_true_args_size, beg_delay_args_size and end_delay_args_size
1412 from HOST_WIDE_INT to poly_int64.
1413 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1414 than a HOST_WIDE_INT.
1415 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1416 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1417 polynomial arg sizes.
1418 * emit-rtl.c (try_split): Use get_args_size.
1419 * recog.c (peep2_attempt): Likewise.
1420 * reload1.c (reload_as_needed): Likewise.
1421 * expr.c (find_args_size_adjust): Return the adjustment as a
1422 poly_int64 rather than a HOST_WIDE_INT.
1423 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1424 to a poly_int64 and change the return type in the same way.
1425 (emit_single_push_insn): Track polynomial arg sizes.
1427 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1428 Alan Hayward <alan.hayward@arm.com>
1429 David Sherwood <david.sherwood@arm.com>
1431 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1432 from HOST_WIDE_INT to poly_int64.
1433 * expr.c (push_block, emit_push_insn): Likewise.
1435 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1436 Alan Hayward <alan.hayward@arm.com>
1437 David Sherwood <david.sherwood@arm.com>
1439 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1441 (get_frame_size): Return the size as a poly_int64 rather than a
1443 (frame_offset_overflow): Take the offset as a poly_int64 rather
1444 than a HOST_WIDE_INT.
1445 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1446 (assign_stack_temp): Likewise for the size.
1447 * function.c (get_frame_size): Return a poly_int64 rather than
1449 (frame_offset_overflow): Take the offset as a poly_int64 rather
1450 than a HOST_WIDE_INT.
1451 (try_fit_stack_local): Take the start, length and size as poly_int64s
1452 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1453 rather than a HOST_WIDE_INT.
1454 (add_frame_space): Take the start and end as poly_int64s rather than
1456 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1457 (assign_stack_temp): Likewise for the size.
1458 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1460 (find_temp_slot_from_address): Handle polynomial offsets.
1461 (combine_temp_slots): Likewise.
1462 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1464 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1465 poly_int64 rather than a HOST_WIDE_INT.
1466 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1467 than a HOST_WIDE_INT.
1468 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1469 polynomial frame offsets.
1470 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1471 as a poly_int64 rather than an int.
1472 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1473 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1474 * config/v850/v850.c (compute_frame_size): Likewise.
1475 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1476 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1477 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1478 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1479 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1480 rather than a HOST_WIDE_INT.
1481 * explow.c (get_dynamic_stack_base): Likewise.
1482 * final.c (final_start_function): Use the constant lower bound
1483 of the frame size for -Wframe-larger-than.
1484 * ira.c (do_reload): Adjust for new get_frame_size return type.
1485 * lra.c (lra): Likewise.
1486 * reload1.c (reload): Likewise.
1487 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1488 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1489 * rtlanal.c (get_initial_register_offset): Return the offset as
1490 a poly_int64 rather than a HOST_WIDE_INT.
1492 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1493 Alan Hayward <alan.hayward@arm.com>
1494 David Sherwood <david.sherwood@arm.com>
1496 * reload1.c (elim_table): Change initial_offset, offset and
1497 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1498 (offsets_at): Change the target array's element type from
1499 HOST_WIDE_INT to poly_int64_pod.
1500 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1501 (elimination_costs_in_insn, update_eliminable_offsets)
1502 (verify_initial_elim_offsets, set_offsets_for_label)
1503 (init_eliminable_invariants): Update after above changes.
1505 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1506 Alan Hayward <alan.hayward@arm.com>
1507 David Sherwood <david.sherwood@arm.com>
1509 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1510 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1511 (decomposition): Change start and end from HOST_WIDE_INT
1513 (decompose, immune_p): Update accordingly.
1514 (find_inc_amount): Return a poly_int64 rather than an int.
1515 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1518 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1519 Alan Hayward <alan.hayward@arm.com>
1520 David Sherwood <david.sherwood@arm.com>
1522 * tree.h (get_inner_reference): Return the bitsize and bitpos
1523 as poly_int64_pods rather than HOST_WIDE_INT.
1524 * fold-const.h (ptr_difference_const): Return the pointer difference
1525 as a poly_int64_pod rather than a HOST_WIDE_INT.
1526 * expr.c (get_inner_reference): Return the bitsize and bitpos
1527 as poly_int64_pods rather than HOST_WIDE_INT.
1528 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1530 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1531 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1532 (optimize_bit_field_compare): Update call to get_inner_reference.
1533 (decode_field_reference): Likewise.
1534 (fold_unary_loc): Track polynomial offsets and sizes.
1535 (split_address_to_core_and_offset): Return the bitpos as a
1536 poly_int64_pod rather than a HOST_WIDE_INT.
1537 (ptr_difference_const): Likewise for the pointer difference.
1538 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1539 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1540 * dbxout.c (dbxout_expand_expr): Likewise.
1541 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1542 (loc_list_from_tree_1, fortran_common): Likewise.
1543 * gimple-laddress.c (pass_laddress::execute): Likewise.
1544 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1545 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1546 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1547 * tree-affine.c (tree_to_aff_combination): Likewise.
1548 (get_inner_reference_aff): Likewise.
1549 * tree-data-ref.c (split_constant_offset_1): Likewise.
1550 (dr_analyze_innermost): Likewise.
1551 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1552 * tree-sra.c (ipa_sra_check_caller): Likewise.
1553 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1554 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1555 (instrument_bool_enum_load, instrument_object_size): Likewise.
1556 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1557 to get_inner_reference.
1558 * hsa-gen.c (gen_hsa_addr): Likewise.
1559 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1560 * tsan.c (instrument_expr): Likewise.
1561 * match.pd: Update call to ptr_difference_const.
1563 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1564 Alan Hayward <alan.hayward@arm.com>
1565 David Sherwood <david.sherwood@arm.com>
1567 * fold-const.c (fold_comparison): Track sizes and offsets as
1568 poly_int64s rather than HOST_WIDE_INTs when folding address
1571 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1572 Alan Hayward <alan.hayward@arm.com>
1573 David Sherwood <david.sherwood@arm.com>
1575 * expr.h (get_bit_range): Return the bitstart and bitend as
1576 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1577 as a poly_int64 rather than a HOST_WIDE_INT.
1578 * expr.c (get_bit_range): Likewise.
1579 (expand_assignment): Update call accordingly.
1580 * fold-const.c (optimize_bit_field_compare): Likewise.
1582 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1583 Alan Hayward <alan.hayward@arm.com>
1584 David Sherwood <david.sherwood@arm.com>
1586 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1587 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1588 as an optional final parameter.
1589 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1590 (aarch64_legitimate_address_p): Likewise.
1591 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1592 instead of an rtx code.
1593 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1594 (aarch64_legitimate_address_hook_p): Likewise.
1595 (aarch64_print_ldpstp_address): Likewise.
1596 (aarch64_print_operand_address): Likewise.
1597 (aarch64_address_cost): Likewise.
1598 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1599 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1600 (aarch64_mem_pair_lanes_operand): Likewise.
1602 2017-12-20 Richard Biener <rguenther@suse.de>
1604 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1605 update_stmt_if_modified.
1607 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1609 PR tree-optimization/83491
1610 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1611 before walking uses. Improve coding style and comments.
1613 2017-12-20 Tom de Vries <tom@codesourcery.com>
1615 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1617 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1620 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1622 (pass_ipa_free_fn_summary): Add small_p private data member,
1623 initialize to false in the ctor.
1624 (pass_ipa_free_fn_summary::clone,
1625 pass_ipa_free_fn_summary::set_pass_param,
1626 pass_ipa_free_fn_summary::gate): New methods.
1627 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1628 | TODO_dump_symtab if small_p.
1629 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1630 entry and add another instance of the pass with false parm after
1632 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1633 ipa_free_fn_summary here.
1635 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1637 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1639 2017-12-20 Martin Sebor <msebor@redhat.com>
1642 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1643 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1644 (expand_builtin_strncmp): Same.
1646 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1649 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1650 allowed debug stmts before labels.
1651 (expand_gimple_basic_block): Likewise.
1652 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1653 * gimple-iterator.h (gsi_after_labels): Likewise.
1654 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1655 rename reused variable, and simplify using gsi_move_before.
1656 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1657 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1658 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1659 (gimple_verify_flow_info, gimple_block_label): Likewise.
1660 (make_blocks): Move debug markers after adjacent labels.
1661 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1662 allowed debug insns outside blocks.
1663 * df-scan.c (df_insn_delete): Likewise.
1664 * lra-constraints.c (update_ebb_live_info): Likewise.
1665 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1666 (vt_initialize, delete_vta_debug_insns): Likewise.
1667 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1669 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1670 Alan Hayward <alan.hayward@arm.com>
1671 David Sherwood <david.sherwood@arm.com>
1673 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1674 (round_up_to_byte_boundary): Likewise.
1675 * expr.h (get_bit_range): Add temporary shim.
1676 * gimple-ssa-store-merging.c (store_operand_info): Change the
1677 bitsize, bitpos, bitregion_start and bitregion_end fields from
1678 unsigned HOST_WIDE_INT to poly_uint64.
1679 (merged_store_group): Likewise load_align_base.
1680 (compatible_load_p, compatible_load_p): Update accordingly.
1681 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1682 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1683 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1684 bitregion_start and bitregion_end as poly_uint64s rather than
1685 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1686 (handled_load): Take the bitsize, bitpos,
1687 bitregion_start and bitregion_end as poly_uint64s rather than
1688 unsigned HOST_WIDE_INTs.
1689 (pass_store_merging::process_store): Update call to
1690 mem_valid_for_store_merging.
1692 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1693 Alan Hayward <alan.hayward@arm.com>
1694 David Sherwood <david.sherwood@arm.com>
1696 * builtins.c (get_object_alignment_2): Track polynomial offsets
1697 and sizes. Update the alignment handling.
1699 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1700 Alan Hayward <alan.hayward@arm.com>
1701 David Sherwood <david.sherwood@arm.com>
1703 * tree.h (get_inner_reference): Add a version that returns the
1704 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1705 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1706 the case in which bitpos is not associated with the first byte.
1708 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1709 Alan Hayward <alan.hayward@arm.com>
1710 David Sherwood <david.sherwood@arm.com>
1712 * tree-affine.h (get_inner_reference_aff): Return the size as a
1714 * tree-affine.c (get_inner_reference_aff): Likewise.
1715 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1716 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1718 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1719 Alan Hayward <alan.hayward@arm.com>
1720 David Sherwood <david.sherwood@arm.com>
1722 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1723 HOST_WIDE_INT rather than a poly_int64.
1725 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1726 Alan Hayward <alan.hayward@arm.com>
1727 David Sherwood <david.sherwood@arm.com>
1729 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1730 HOST_WIDE_INT to poly_int64_pod.
1731 (perform_symbolic_merge): Update accordingly.
1732 (bswap_replace): Likewise.
1734 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1735 Alan Hayward <alan.hayward@arm.com>
1736 David Sherwood <david.sherwood@arm.com>
1738 * tree-affine.h (aff_tree::offset): Change from widest_int
1740 (wide_int_ext_for_comb): Delete.
1741 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1742 constants as poly_widest_int rather than widest_int.
1743 (aff_combination_constant_multiple_p): Return the multiplier
1744 as a poly_widest_int.
1745 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1747 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1748 version static and add an overload for poly_widest_int.
1749 (aff_combination_const, aff_combination_add_cst)
1750 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1751 the constants as poly_widest_int rather than widest_int.
1752 (tree_to_aff_combination): Generalize INTEGER_CST case to
1754 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1755 (aff_combination_add_product, aff_combination_mult): Handle
1757 (aff_combination_constant_multiple_p): Return the multiplier
1758 as a poly_widest_int.
1759 * tree-predcom.c (determine_offset): Return the offset as a
1761 (split_data_refs_to_components, suitable_component_p): Update
1763 (valid_initializer_p): Update call to
1764 aff_combination_constant_multiple_p.
1765 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1766 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1767 as a poly_int64 rather than a HOST_WIDE_INT.
1768 (get_address_cost): Handle polynomial offsets.
1769 (iv_elimination_compare_lt): Likewise.
1770 (rewrite_use_nonlinear_expr): Likewise.
1772 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1773 Alan Hayward <alan.hayward@arm.com>
1774 David Sherwood <david.sherwood@arm.com>
1776 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1777 as a poly_int64_pod rather than a HOST_WIDE_INT.
1778 (get_addr_base_and_unit_offset): Likewise.
1779 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1780 (get_addr_base_and_unit_offset): Likewise.
1781 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1782 to poly_int64 in example.
1783 * fold-const.c (fold_binary_loc): Update call to
1784 get_addr_base_and_unit_offset.
1785 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1786 (maybe_canonicalize_mem_ref_addr): Likewise.
1787 (gimple_fold_stmt_to_constant_1): Likewise.
1788 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1790 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1791 * match.pd: Likewise.
1792 * omp-low.c (lower_omp_target): Likewise.
1793 * tree-sra.c (build_ref_for_offset): Likewise.
1794 (build_debug_ref_for_model): Likewise.
1795 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1796 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1797 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1798 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1799 (constant_pointer_difference): Likewise.
1800 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1801 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1802 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1803 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1804 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1805 (set_ssa_val_to): Likewise.
1806 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1807 (maybe_diag_stxncpy_trunc): Likewise.
1808 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1809 * tree.c (build_simple_mem_ref_loc): Likewise.
1810 (array_at_struct_end_p): Likewise.
1812 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1813 Alan Hayward <alan.hayward@arm.com>
1814 David Sherwood <david.sherwood@arm.com>
1816 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1817 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1818 (get_ref_base_and_extent_hwi): Declare.
1819 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1820 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1821 (get_ref_base_and_extent_hwi): New function.
1822 * cfgexpand.c (expand_debug_expr): Update call to
1823 get_ref_base_and_extent.
1824 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1825 * gimple-fold.c (get_base_constructor): Return the offset as a
1826 poly_int64_pod rather than a HOST_WIDE_INT.
1827 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1828 * ipa-polymorphic-call.c
1829 (ipa_polymorphic_call_context::set_by_invariant)
1830 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1831 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1832 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1833 rather than get_ref_base_and_extent.
1834 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1835 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1836 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1838 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1839 call to get_ref_base_and_extent.
1840 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1841 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1842 (stmt_kills_ref_p): Likewise.
1843 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1844 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1846 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1847 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1848 when calling native_encode_expr.
1849 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1850 call to get_ref_base_and_extent.
1851 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1852 get_ref_base_and_extent.
1853 * var-tracking.c (track_expr_p): Likewise.
1855 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1856 Alan Hayward <alan.hayward@arm.com>
1857 David Sherwood <david.sherwood@arm.com>
1859 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1860 HOST_WIDE_INT to poly_int64_pod.
1861 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1862 polynomail parameter offsets.
1864 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1865 Alan Hayward <alan.hayward@arm.com>
1866 David Sherwood <david.sherwood@arm.com>
1868 * gengtype.c (main): Handle poly_int64_pod.
1869 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1870 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1871 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1872 from HOST_WIDE_INT to poly_int64_pod.
1873 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1874 (copy_cfa): New function.
1875 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1876 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1877 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1878 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1879 (initial_return_save): Treat offsets as poly_ints.
1880 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1881 offset is nonconstant.
1882 (reg_save): Take the offset as a poly_int64. Fall back to
1883 DW_CFA_expression for nonconstant offsets.
1884 (queue_reg_save): Take the offset as a poly_int64.
1885 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1887 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1888 Alan Hayward <alan.hayward@arm.com>
1889 David Sherwood <david.sherwood@arm.com>
1891 * rtl.h (operand_subword, operand_subword_force): Take the offset
1892 as a poly_uint64 an unsigned int.
1893 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1895 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1896 Alan Hayward <alan.hayward@arm.com>
1897 David Sherwood <david.sherwood@arm.com>
1899 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1900 'p' format code. Use INT_LIST rather than SUBREG as the example of
1901 a code with an XINT and an XEXP. Remove the implication that
1902 accessing an rtx field using XINT is expected to work.
1903 * rtl.def (SUBREG): Change format from "ei" to "ep".
1904 * rtl.h (rtunion::rt_subreg): New field.
1905 (XCSUBREG): New macro.
1906 (SUBREG_BYTE): Use it.
1907 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1908 Update constructor accordingly.
1909 (subreg_shape::operator ==): Update accordingly.
1910 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1911 than an unsigned int.
1912 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1913 a poly_uint64 rather than an unsigned int.
1914 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1915 than an unsigned int.
1916 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1917 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1918 an unsigned int. Take the sizes as poly_uint64s.
1919 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1920 an unsigned int. Take the shift as a poly_uint64 rather than
1922 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1923 as a poly_uint64 rather than an unsigned int.
1924 (simplify_subreg_regno): Likewise.
1925 (byte_lowpart_offset): Return the memory offset as a poly_int64
1927 (subreg_memory_offset): Likewise. Take the subreg offset as a
1928 poly_uint64 rather than an unsigned int.
1929 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1930 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1931 poly_uint64 rather than an unsigned int.
1932 * rtl.c (rtx_format): Describe 'p' in comment.
1933 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1934 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1935 offset as a poly_uint64 rather than an unsigned int.
1936 (byte_lowpart_offset): Return the memory offset as a poly_int64
1938 (subreg_memory_offset): Likewise. Take the subreg offset as a
1939 poly_uint64 rather than an unsigned int.
1940 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1941 mode sizes as poly_uint64s rather than unsigned ints. Return a
1942 poly_uint64 rather than an unsigned int.
1943 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1944 (copy_insn_1): Handle 'p'.
1945 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1946 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1947 an unsigned int. Return the shift in the same way.
1948 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1950 (subreg_size_offset_from_lsb): Take the sizes and shift as
1951 poly_uint64s rather than unsigned ints. Return the offset as
1953 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1954 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1956 * rtlhash.c (add_rtx): Handle 'p'.
1957 * genemit.c (gen_exp): Likewise.
1958 * gengenrtl.c (type_from_format, gendef): Likewise.
1959 * gensupport.c (subst_pattern_match, get_alternatives_number)
1960 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1961 (subst_dup): Likewise.
1962 * gengtype.c (adjust_field_rtx_def): Likewise.
1963 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1964 (match_pattern_2): Likewise.
1965 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1966 (rtx_test::subreg_field): New function.
1967 (operator ==, safe_to_hoist_p, transition_parameter_type)
1968 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1969 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1970 * genpeep.c (match_rtx): Likewise.
1971 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1972 (rtx_writer::print_rtx_operand): Handle 'p'.
1973 (print_value): Handle SUBREG.
1974 * read-rtl.c (apply_int_iterator): Likewise.
1975 (rtx_reader::read_rtx_operand): Handle 'p'.
1976 * alias.c (rtx_equal_for_memref_p): Likewise.
1977 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1978 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1980 * calls.c (expand_call): Likewise.
1981 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1982 (make_extraction, gen_lowpart_for_combine): Likewise.
1983 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1985 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1986 rather than an unsigned int. Treat subreg offsets as poly_ints.
1987 (exp_equiv_p): Handle 'p'.
1988 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1989 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1990 * dse.c (find_shift_sequence): Likewise.
1991 * dwarf2out.c (rtl_for_decl_location): Likewise.
1992 * expmed.c (extract_low_bits): Likewise.
1993 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1994 (expand_expr_real_2): Likewise.
1995 * final.c (alter_subreg): Likewise.
1996 (leaf_renumber_regs_insn): Handle 'p'.
1997 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1998 Treat subreg offsets as poly_ints.
1999 * fwprop.c (forward_propagate_and_simplify): Likewise.
2000 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2001 * ira.c (get_subreg_tracking_sizes): Likewise.
2002 * ira-conflicts.c (go_through_subreg): Likewise.
2003 * ira-lives.c (process_single_reg_class_operands): Likewise.
2004 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2005 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2006 as a poly_uint64 rather than an unsigned int.
2007 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2008 subreg offsets as poly_ints.
2009 * lra-constraints.c (operands_match_p): Handle 'p'.
2010 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2011 * lra-spills.c (assign_mem_slot): Likewise.
2012 * postreload.c (move2add_valid_value_p): Likewise.
2013 * recog.c (general_operand, indirect_operand): Likewise.
2014 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2015 (copyprop_hardreg_forward_1): Likewise.
2016 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2017 (record_subregs_of_mode): Likewise.
2018 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2019 * reload.c (operands_match_p): Handle 'p'.
2020 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2021 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2022 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2023 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2024 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2025 (simplify_const_poly_int_tests<N>::run): Likewise.
2026 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2027 a poly_uint64 rather than an unsigned int.
2028 * valtrack.c (debug_lowpart_subreg): Likewise.
2029 * var-tracking.c (var_lowpart): Likewise.
2030 (loc_cmp): Handle 'p'.
2032 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2033 Alan Hayward <alan.hayward@arm.com>
2034 David Sherwood <david.sherwood@arm.com>
2036 * ira.c (get_subreg_tracking_sizes): New function.
2037 (init_live_subregs): Take an integer size rather than a register.
2038 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2039 to init_live_subregs.
2041 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2042 Alan Hayward <alan.hayward@arm.com>
2043 David Sherwood <david.sherwood@arm.com>
2045 * expr.c (store_constructor_field): Change bitsize from a
2046 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2047 HOST_WIDE_INT to a poly_int64.
2048 (store_constructor): Change size from a HOST_WIDE_INT to
2050 (store_field): Likewise bitsize and bitpos.
2052 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2053 Alan Hayward <alan.hayward@arm.com>
2054 David Sherwood <david.sherwood@arm.com>
2056 * expmed.h (store_bit_field): Change bitregion_start and
2057 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2058 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2059 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2060 (store_fixed_bit_field, store_split_bit_field): Likewise.
2061 * expr.c (store_constructor_field, store_field): Likewise.
2062 (optimize_bitfield_assignment_op): Likewise. Make the same change
2063 to bitsize and bitpos.
2064 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2065 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2066 same change in the constructor arguments.
2067 (get_best_mode): Change bitregion_start and bitregion_end from
2068 unsigned HOST_WIDE_INT to poly_uint64.
2069 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2070 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2072 (bit_field_mode_iterator::next_mode): Update for new types
2073 of m_bitregion_start and m_bitregion_end.
2074 (get_best_mode): Change bitregion_start and bitregion_end from
2075 unsigned HOST_WIDE_INT to poly_uint64.
2077 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2078 Alan Hayward <alan.hayward@arm.com>
2079 David Sherwood <david.sherwood@arm.com>
2081 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2082 accepts poly_uint64 offsets.
2083 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2084 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2085 * expmed.c (lowpart_bit_field_p): Likewise.
2086 (extract_bit_field_as_subreg): New function, split out from...
2087 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2088 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2089 extractions, check that BITSIZE matches the size of the extracted
2090 value and that BITNUM is an exact multiple of that size.
2091 If all else fails, try forcing the value into memory if
2092 BITNUM is variable, and adjusting the address so that the
2093 offset is constant. Split the part that can only handle constant
2094 bitsize and bitnum out into...
2095 (extract_integral_bit_field): ...this new function.
2096 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2097 rather than unsigned HOST_WIDE_INTs.
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 * expmed.h (store_bit_field): Take bitsize and bitnum as
2104 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2105 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2106 that returns the byte size.
2107 (store_bit_field_1): Take bitsize and bitnum as
2108 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2109 to simple_mem_bitfield_p. Split the part that can only handle
2110 constant bitsize and bitnum out into...
2111 (store_integral_bit_field): ...this new function.
2112 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2113 than unsigned HOST_WIDE_INTs.
2114 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2116 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2117 Alan Hayward <alan.hayward@arm.com>
2118 David Sherwood <david.sherwood@arm.com>
2120 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2121 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2123 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2124 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2125 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2126 offset as a poly_int64 rather than an int.
2127 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2128 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2129 * lra-constraints.c (equiv_address_substitution): Track offsets
2131 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2132 (curr_insn_transform): Handle the new form of sp_offset.
2133 * lra-eliminations.c (lra_elim_table): Change previous_offset
2134 and offset from HOST_WIDE_INT to poly_int64.
2135 (print_elim_table, update_reg_eliminate): Update accordingly.
2136 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2137 (get_elimination): Update accordingly.
2138 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2139 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2140 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2141 poly_int64 offsets generally.
2142 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2143 (mark_not_eliminable, init_elimination): Update accordingly.
2144 (remove_reg_equal_offset_note): Return a bool and pass the new
2145 offset back by pointer as a poly_int64.
2146 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2147 rather than a HOST_WIDE_INT.
2148 (do_remat): Track offsets poly_int64s.
2149 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2151 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2152 Alan Hayward <alan.hayward@arm.com>
2153 David Sherwood <david.sherwood@arm.com>
2155 * rtl.h (mem_attrs): Add a default constructor. Change size and
2156 offset from HOST_WIDE_INT to poly_int64.
2157 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2158 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2159 (widen_memory_access): Take the sizes and offsets as poly_int64s
2160 rather than HOST_WIDE_INTs.
2161 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2162 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2164 (adjust_offset_for_component_ref): Change the offset from a
2165 HOST_WIDE_INT to a poly_int64.
2166 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2167 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2168 * dce.c (find_call_stack_args): Likewise.
2169 * dse.c (record_store): Likewise.
2170 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2171 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2172 * read-rtl-function.c (test_loading_mem): Likewise.
2173 * rtlanal.c (may_trap_p_1): Likewise.
2174 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2175 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2176 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2177 (mem_attrs::mem_attrs): New function.
2178 (set_mem_attributes_minus_bitpos): Change bitpos from a
2179 HOST_WIDE_INT to poly_int64.
2180 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2181 (clear_mem_offset, clear_mem_size, change_address)
2182 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2183 initialize mem_attrs.
2184 (set_mem_offset, set_mem_size, adjust_address_1)
2185 (adjust_automodify_address_1, offset_address, widen_memory_access):
2186 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2188 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2189 Alan Hayward <alan.hayward@arm.com>
2190 David Sherwood <david.sherwood@arm.com>
2192 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2193 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2194 rather than 0 to represent an unknown size. Assert that the size
2195 is known when the mode isn't BLKmode.
2196 (may_trap_p_1): Use -1 for unknown sizes.
2197 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2199 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2200 Alan Hayward <alan.hayward@arm.com>
2201 David Sherwood <david.sherwood@arm.com>
2203 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2204 to poly_int64. Update commentary for positions_needed.large.
2205 (read_info_type): Change offset and width from HOST_WIDE_INT
2207 (set_usage_bits): Likewise.
2208 (canon_address): Return the offset as a poly_int64 rather than
2209 a HOST_WIDE_INT. Use strip_offset_and_add.
2210 (set_all_positions_unneeded, any_positions_needed_p): Use
2211 positions_needed.large to track stores with non-constant widths.
2212 (all_positions_needed_p): Likewise. Take the offset and width
2213 as poly_int64s rather than ints. Assert that rhs is nonnull.
2214 (record_store): Cope with non-constant offsets and widths.
2215 Nullify the rhs of an earlier store if we can't tell which bytes
2217 (find_shift_sequence): Take the access_size and shift as poly_int64s
2219 (get_stored_val): Take the read_offset and read_width as poly_int64s
2220 rather than HOST_WIDE_INTs.
2221 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2222 non-constant offsets and widths.
2224 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2225 Alan Hayward <alan.hayward@arm.com>
2226 David Sherwood <david.sherwood@arm.com>
2228 * inchash.h (inchash::hash::add_poly_int): New function.
2229 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2230 Use poly_int64 rather than HOST_WIDE_INT.
2231 (ao_ref::max_size_known_p): New function.
2232 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2233 rather than HOST_WIDE_INT.
2234 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2235 to temporaries until its interface is adjusted to match.
2236 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2237 (aliasing_component_refs_p, decl_refs_may_alias_p)
2238 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2239 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2240 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2242 * alias.c (ao_ref_from_mem): Likewise.
2243 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2244 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2245 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2246 (maybe_trim_complex_store, maybe_trim_constructor_store)
2247 (live_bytes_read, dse_classify_store): Likewise.
2248 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2249 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2250 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2251 (vn_reference_lookup_3): Likewise.
2252 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2254 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2255 Alan Hayward <alan.hayward@arm.com>
2256 David Sherwood <david.sherwood@arm.com>
2258 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2259 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2260 instead of ranges_overlap_p.
2262 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2263 Alan Hayward <alan.hayward@arm.com>
2264 David Sherwood <david.sherwood@arm.com>
2266 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2267 sizes as poly_int64s rather than HOST_WIDE_INTs.
2269 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2270 Alan Hayward <alan.hayward@arm.com>
2271 David Sherwood <david.sherwood@arm.com>
2273 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2274 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2275 * gimple-fold.c (fold_ctor_reference): Likewise.
2277 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2278 Alan Hayward <alan.hayward@arm.com>
2279 David Sherwood <david.sherwood@arm.com>
2281 * target.def (dwarf_poly_indeterminate_value): New hook.
2282 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2283 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2284 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2285 * doc/tm.texi: Regenerate.
2286 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2287 offset as a poly_int64.
2288 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2289 offset as a poly_int64.
2290 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2291 Take the offset as a poly_int64.
2292 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2293 (frame_pointer_fb_offset): Change to a poly_int64.
2294 (int_loc_descriptor): Take the offset as a poly_int64. Use
2295 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2296 (based_loc_descr): Take the offset as a poly_int64.
2297 Use strip_offset_and_add to handle (plus X (const)).
2298 Use new_reg_loc_descr instead of an open-coded version of the
2299 previous implementation.
2300 (mem_loc_descriptor): Handle CONST_POLY_INT.
2301 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2302 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2304 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2305 Alan Hayward <alan.hayward@arm.com>
2306 David Sherwood <david.sherwood@arm.com>
2308 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2309 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2310 * inchash.h (inchash::hash::add_poly_hwi): New function.
2311 * gengtype.c (main): Register poly_int64.
2312 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2313 offset as a poly_int.
2314 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2315 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2316 offset as a poly_int64.
2317 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2318 * print-rtl.c (print_poly_int): New function.
2319 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2321 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2323 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2324 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2325 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2326 Enforce track_offset_p for parts of a PARALLEL.
2327 (vt_add_function_parameter): Use const_offset for the final
2328 offset to track. Use get_tracked_reg_offset for the parts
2331 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2332 Alan Hayward <alan.hayward@arm.com>
2333 David Sherwood <david.sherwood@arm.com>
2335 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2336 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2337 * doc/tm.texi: Regenerate.
2338 * hooks.h (hook_bool_uint_uint_true): Delete.
2339 (hook_bool_puint64_puint64_true): Declare.
2340 * hooks.c (hook_bool_uint_uint_true): Delete.
2341 (hook_bool_puint64_puint64_true): New function.
2342 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2343 instead of unsigned ints.
2344 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2345 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2347 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2348 Alan Hayward <alan.hayward@arm.com>
2349 David Sherwood <david.sherwood@arm.com>
2351 * optabs.h (expand_operand): Add an int_value field.
2352 (create_expand_operand): Add an int_value parameter and use it
2353 to initialize the new expand_operand field.
2354 (create_integer_operand): Replace with a declaration of a function
2355 that accepts poly_int64s. Move the implementation to...
2356 * optabs.c (create_integer_operand): ...here.
2357 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2358 the mode preserves the value of int_value, instead of calling
2359 const_int_operand on the rtx. Use gen_int_mode to generate
2362 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2363 Alan Hayward <alan.hayward@arm.com>
2364 David Sherwood <david.sherwood@arm.com>
2366 * dumpfile.h (dump_dec): Declare.
2367 * dumpfile.c (dump_dec): New function.
2368 * pretty-print.h (pp_wide_integer): Turn into a function and
2369 declare a poly_int version.
2370 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2372 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2373 Alan Hayward <alan.hayward@arm.com>
2374 David Sherwood <david.sherwood@arm.com>
2376 * doc/generic.texi (POLY_INT_CST): Document.
2377 * tree.def (POLY_INT_CST): New tree code.
2378 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2379 * tree-core.h (tree_poly_int_cst): New struct.
2380 (tree_node): Add a poly_int_cst field.
2381 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2382 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2383 instead of a wide_int_ref.
2384 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2386 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2387 instead of an unsigned HOST_WIDE_INT.
2388 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2389 (ptrdiff_tree_p): Declare.
2390 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2391 extern inline implementations if the target doesn't use POLY_INT_CST.
2392 (poly_int_tree_p): New function.
2393 (wi::unextended_tree): New class.
2394 (wi::int_traits <unextended_tree>): New override.
2395 (wi::extended_tree): Add a default constructor.
2396 (wi::extended_tree::get_tree): New function.
2397 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2398 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2399 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2400 (wi::tree_to_poly_wide_ref): New typedefs.
2401 (wi::ints_for): Provide overloads for extended_tree and
2403 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2404 (wi::to_wide): New functions.
2405 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2406 * tree.c (poly_int_cst_hasher): New struct.
2407 (poly_int_cst_hash_table): New variable.
2408 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2409 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2411 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2412 (init_ttree): Initialize poly_int_cst_hash_table.
2413 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2414 a poly_int64 instead of a HOST_WIDE_INT.
2415 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2416 instead of an unsigned HOST_WIDE_INT.
2417 (wide_int_to_tree): Rename to...
2418 (wide_int_to_tree_1): ...this.
2419 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2420 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2421 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2422 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2423 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2424 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2426 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2427 (streamer_read_tree_body): Likewise.
2428 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2429 (streamer_write_tree_body): Likewise.
2430 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2431 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2432 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2433 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2434 * gimple-expr.h (is_gimple_constant): Likewise.
2435 * gimplify.c (maybe_with_size_expr): Likewise.
2436 * print-tree.c (print_node): Likewise.
2437 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2438 * tree-pretty-print.c (dump_generic_node): Likewise.
2439 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2440 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2441 * tree-vrp.c (compare_values_warnv): Likewise.
2442 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2443 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2444 (force_expr_to_var_cost): Likewise.
2445 * tree-ssa-loop.c (for_each_index): Likewise.
2446 * fold-const.h (build_invariant_address, size_int_kind): Take a
2447 poly_int64 instead of a HOST_WIDE_INT.
2448 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2449 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2451 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2452 (int_const_binop_2): New function, split out from...
2453 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2454 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2455 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2456 * gimple-ssa-strength-reduction.c (slsr_process_add)
2457 (slsr_process_mul): Check for INTEGER_CSTs before using them
2459 * stor-layout.c (bits_from_bytes): New function.
2460 (bit_from_pos): Use it.
2461 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2462 by BITS_PER_UNIT to get the TYPE_SIZE.
2463 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2464 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
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 * doc/rtl.texi (const_poly_int): Document. Also document the
2471 rtl sharing behavior.
2472 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2473 * rtl.h (const_poly_int_def): New struct.
2474 (rtx_def::u): Add a cpi field.
2475 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2476 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2477 (wi::rtx_to_poly_wide_ref): New typedef
2478 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2479 (poly_int_rtx_p): New functions.
2480 (trunc_int_for_mode): Declare a poly_int64 version.
2481 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2482 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2484 (strip_offset): Declare.
2485 (strip_offset_and_add): New function.
2486 * rtl.def (CONST_POLY_INT): New rtx code.
2487 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2488 (shared_const_p): Use poly_int_rtx_p.
2489 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2491 (gen_int_shift_amount): Likewise.
2492 * emit-rtl.c (const_poly_int_hasher): New class.
2493 (const_poly_int_htab): New variable.
2494 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2495 (const_poly_int_hasher::hash): New function.
2496 (const_poly_int_hasher::equal): Likewise.
2497 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2498 (immed_wide_int_const): Rename to...
2499 (immed_wide_int_const_1): ...this and make static.
2500 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2501 instead of a wide_int_ref.
2502 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2503 (gen_lowpart_common): Handle CONST_POLY_INT.
2504 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2505 * cselib.c (cselib_hash_rtx): Likewise.
2506 * dwarf2out.c (const_ok_for_output_1): Likewise.
2507 * expr.c (convert_modes): Likewise.
2508 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2509 * rtlhash.c (add_rtx): Likewise.
2510 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2511 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2512 Handle existing CONST_POLY_INT rtxes.
2513 * expmed.h (expand_shift): Take a poly_int64 instead of a
2515 * expmed.c (expand_shift): Likewise.
2516 * rtlanal.c (strip_offset): New function.
2517 (commutative_operand_precedence): Give CONST_POLY_INT the same
2518 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2520 * rtl-tests.c (const_poly_int_tests): New struct.
2521 (rtl_tests_c_tests): Use it.
2522 * simplify-rtx.c (simplify_const_unary_operation): Handle
2524 (simplify_const_binary_operation): Likewise.
2525 (simplify_binary_operation_1): Fold additions of symbolic constants
2526 and CONST_POLY_INTs.
2527 (simplify_subreg): Handle extensions and truncations of
2529 (simplify_const_poly_int_tests): New struct.
2530 (simplify_rtx_c_tests): Use it.
2531 * wide-int.h (storage_ref): Add default constructor.
2532 (wide_int_ref_storage): Likewise.
2533 (trailing_wide_ints): Use GTY((user)).
2534 (trailing_wide_ints::operator[]): Add a const version.
2535 (trailing_wide_ints::get_precision): New function.
2536 (trailing_wide_ints::extra_size): Likewise.
2538 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2539 Alan Hayward <alan.hayward@arm.com>
2540 David Sherwood <david.sherwood@arm.com>
2542 * emit-rtl.h (gen_int_shift_amount): Declare.
2543 * emit-rtl.c (gen_int_shift_amount): New function.
2544 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2546 * calls.c (shift_return_value): Likewise.
2547 * cse.c (fold_rtx): Likewise.
2548 * dse.c (find_shift_sequence): Likewise.
2549 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2550 (expand_shift, expand_smod_pow2): Likewise.
2551 * lower-subreg.c (shift_cost): Likewise.
2552 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2553 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2554 (expand_vec_perm_var): Likewise.
2555 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2556 (simplify_binary_operation_1): Likewise.
2557 * combine.c (try_combine, find_split_point, force_int_to_mode)
2558 (simplify_shift_const_1, simplify_shift_const): Likewise.
2559 (change_zero_ext): Likewise. Use simplify_gen_binary.
2561 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2563 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2565 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2567 * doc/invoke.texi (ARM Options): Document accepted extension options
2568 for -march=armv8.3-a.
2570 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2573 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2574 or --with-float=softfp, set the default CPU to arm10e.
2576 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2578 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2579 * config/visium/predicates.md (const_shift_operand): Likewise.
2580 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2581 (visium_legitimize_reload_address): Likewise.
2583 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2586 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2588 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2591 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2593 2017-12-20 Julia Koval <julia.koval@intel.com>
2595 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2596 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2597 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2598 * config.gcc: Include vpclmulqdqintrin.h.
2599 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2600 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2602 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2603 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2604 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2605 * config/i386/i386.c (isa2_opts): Add -mcx16.
2606 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2607 (ix86_option_override_internal): Move mcx16 to flags2.
2608 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2609 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2610 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2611 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2612 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2613 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2614 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2615 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2616 * doc/invoke.texi: Add -mvpclmulqdq.
2618 2017-12-20 Tom de Vries <tom@codesourcery.com>
2621 * config/i386/i386.c (ix86_static_chain): Move
2622 DECL_STATIC_CHAIN test ...
2623 * calls.c (rtx_for_static_chain): ... here. New function.
2624 * calls.h (rtx_for_static_chain): Declare.
2625 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2626 instead of targetm.calls.static_chain.
2627 * df-scan.c (df_get_entry_block_def_set): Same.
2629 2017-12-19 Tom de Vries <tom@codesourcery.com>
2631 PR tree-optimization/83493
2632 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2635 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2637 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2638 inform with hardcoded english plural handling.
2640 2017-12-18 Jeff Law <law@redhat.com>
2642 PR tree-optimization/83477
2643 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2644 a non-virtual PHI, always push a new range.
2646 2017-12-19 Martin Sebor <msebor@redhat.com>
2649 * builtins.c (compute_objsize): Handle non-constant offsets.
2651 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2653 PR tree-optimization/83444
2654 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2655 character load case, if get_stridx on MEM_REF's operand doesn't
2656 look usable, retry with get_addr_stridx.
2658 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2661 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2662 (variable_tracking_main_1): Keep markers even when VTA fails.
2665 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2666 even if there are markers before it.
2667 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2669 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2671 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2672 typical order conditions.
2673 * sel-sched.c (extract_new_fences_from): Likewise.
2674 * config/visium/constraints.md (J, K, L): Likewise.
2675 * config/visium/predicates.md (const_shift_operand): Likewise.
2676 * config/visium/visium.c (visium_legitimize_address,
2677 visium_legitimize_reload_address): Likewise.
2678 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2679 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2680 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2681 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2682 SET_DUMP_DETAIL): Likewise.
2683 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2684 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2685 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2686 avr_set_core_architecture, avr_set_current_function,
2687 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2688 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2689 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2690 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2691 avr_map_decompose, avr_fold_builtin): Likewise.
2692 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2693 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2694 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2695 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2696 * config/m32c/m32c.c (m32c_conditional_register_usage,
2697 m32c_address_cost): Likewise.
2698 * config/m32c/predicates.md (shiftcount_operand,
2699 longshiftcount_operand): Likewise.
2700 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2701 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2702 can_use_cdx_ldstw): Likewise.
2703 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2704 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2706 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2707 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2708 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2709 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2710 * config/vax/vax.c (adjacent_operands_p): Likewise.
2711 * config/ft32/constraints.md (L, b, KA): Likewise.
2712 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2714 * cfgexpand.c (expand_stack_alignment): Likewise.
2715 * gcse.c (insert_expr_in_table): Likewise.
2716 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2717 * cgraphunit.c (cgraph_node::expand): Likewise.
2718 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2719 * emit-rtl.c (add_insn): Likewise.
2720 * input.c (dump_location_info): Likewise.
2721 * passes.c (NEXT_PASS): Likewise.
2722 * read-rtl-function.c (parse_note_insn_name,
2723 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2725 * sched-rgn.c (sched_rgn_init): Likewise.
2726 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2727 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2728 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2730 * explow.c (eliminate_constant_term): Likewise.
2731 * final.c (leaf_renumber_regs_insn): Likewise.
2732 * cfgrtl.c (print_rtl_with_bb): Likewise.
2733 * genhooks.c (emit_init_macros): Likewise.
2734 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2735 * tree-data-ref.c (conflict_fn): Likewise.
2736 * selftest.c (assert_streq): Likewise.
2737 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2738 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2739 fold_binary_loc, multiple_of_p): Likewise.
2740 * reload.c (push_reload, find_equiv_reg): Likewise.
2741 * et-forest.c (et_nca, et_below): Likewise.
2742 * dbxout.c (dbxout_symbol_location): Likewise.
2743 * reorg.c (relax_delay_slots): Likewise.
2744 * dojump.c (do_compare_rtx_and_jump): Likewise.
2745 * gengtype-parse.c (type): Likewise.
2746 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2747 simplify_const_relational_operation): Likewise.
2748 * reload1.c (do_output_reload): Likewise.
2749 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2750 * gengtype.c (type_for_name): Likewise.
2751 * gimple-ssa-sprintf.c (format_directive): Likewise.
2753 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2756 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2757 accessing it. Adjust comment.
2759 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2762 * predict.c (zero_one_minusone): New function.
2763 (apply_return_prediction): Avoid return prediction for functions
2764 returning only -1, 0 and 1 values, unless they only return -1 and 0
2767 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2769 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2770 fall-through warning.
2771 (arc_compute_frame_size): Remove unused variables.
2772 (arc_print_operand): Fix fprintif format.
2773 (arc_can_follow_jump): Clean fall-through warning.
2775 2017-12-19 Marek Polacek <polacek@redhat.com>
2778 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2781 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2783 * config/arc/arc.c (overriderregs): New variable.
2784 (arc_override_options): Track fixed/call saved/call options.
2785 (arc_conditional_register_usage): Check against overrideregs
2786 variable whenever we change register properties.
2788 2017-12-19 Nathan Sidwell <nathan@acm.org>
2790 * opts.c (finish_options): Don't prefix dump_base_name if it
2791 already contains directories.
2793 2017-12-19 Martin Liska <mliska@suse.cz>
2795 PR rtl-optimization/82675
2796 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2797 more element in sbitmap.
2799 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2801 * gimplify.c (gimplify_expr): Use error_operand_p.
2803 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2806 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2809 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2811 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2812 the jump is to a label.
2814 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2816 PR tree-optimization/83444
2817 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2822 * ipa-inline.c (flatten_remove_node_hook): New function.
2823 (ipa_inline): Keep only nodes with flatten attribute at the end of
2824 the array in the order from ipa_reverse_postorder, only walk that
2825 portion of array for flattening, if there is more than one such
2826 node, temporarily register a removal hook and ignore removed nodes.
2828 PR tree-optimization/80631
2829 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2830 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2831 IFN_REDUC_MAX or IFN_REDUC_MIN.
2833 2017-12-18 Jeff Law <law@redhat.com>
2835 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2836 record symbolic equivalences from backedges in the CFG.
2839 2017-11-19 Jeff Law <law@redhat.com>
2841 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2842 of degenerates resulting from ignoring an edge.
2844 2017-12-18 Martin Sebor <msebor@redhat.com>
2847 PR tree-optimization/78450
2848 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2849 (handle_builtin_strlen): Call it.
2851 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2853 PR rtl-optimization/83424
2854 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2856 2017-12-18 Marek Polacek <polacek@redhat.com>
2859 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2860 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2863 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2865 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2866 (umaddsidi4, umaddsidi_split): Likewise.
2868 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2870 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2873 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2876 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2878 2017-12-18 Richard Biener <rguenther@suse.de>
2880 PR tree-optimization/81877
2881 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2882 (outermost_indep_loop): Adjust.
2883 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2884 (can_sm_ref_p): Adjust.
2886 2017-12-18 Richard Biener <rguenther@suse.de>
2889 * tree.c (array_at_struct_end_p): Return true if the underlying
2890 object has space for at least one element in excess of what
2891 the array domain specifies.
2893 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2895 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2896 example to avoid overfull hbox.
2897 * doc/invoke.texi (Option Summary): Add missing @gol.
2898 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2901 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2902 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2904 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2905 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2906 (AArch64 Options): Document -mverbose-cost-dump.
2907 (ARM Options): Likewise, plus -mflip-thumb.
2909 2017-12-17 Martin Sebor <msebor@redhat.com>
2912 * gimple-ssa-warn-restrict.c
2913 (builtin_memref::offset_out_of_bounds): Correct the handling of
2916 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2918 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2920 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2922 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2924 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2926 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2929 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2931 * doc/invoke.texi: Fix some typos.
2933 2017-12-16 Martin Sebor <msebor@redhat.com>
2935 PR tree-optimization/78918
2936 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2937 * builtins.c (check_sizes): Rename...
2938 (check_access): ...to this. Rename function arguments for clarity.
2939 (check_memop_sizes): Adjust names.
2940 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2941 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2942 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2943 (check_strncat_sizes, expand_builtin_strncat): Same.
2944 (expand_builtin_strncpy, expand_builtin_memset): Same.
2945 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2946 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2947 (maybe_emit_sprintf_chk_warning): Same.
2948 (expand_builtin_strcpy): Adjust.
2949 (expand_builtin_stpcpy): Same.
2950 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2951 in pointer-checking forms of memcpy, memmove, and mempcpy.
2952 (gcall_to_tree_minimal, max_object_size): Define new functions.
2953 * builtins.h (max_object_size): Declare.
2954 * calls.c (alloc_max_size): Call max_object_size instead of
2955 hardcoding ssizetype limit.
2956 (get_size_range): Handle new argument.
2957 * calls.h (get_size_range): Add a new argument.
2958 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2959 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2960 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2962 (gimple_fold_builtin_memory_chk): Same.
2963 (gimple_fold_builtin_stxcpy_chk): New function.
2964 * gimple-ssa-warn-restrict.c: New source.
2965 * gimple-ssa-warn-restrict.h: New header.
2966 * gimple.c (gimple_build_call_from_tree): Propagate location.
2967 * passes.def (pass_warn_restrict): Add new pass.
2968 * tree-pass.h (make_pass_warn_restrict): Declare.
2969 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2971 (handle_builtin_strcat): Same.
2972 (strlen_optimize_stmt): Rename...
2973 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2974 stpncpy, strncpy, and their checking forms.
2976 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2978 PR rtl-optimization/82849
2979 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2980 and get_max_loop_iterations_int.
2982 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2983 Alan Hayward <alan.hayward@arm.com>
2984 David Sherwood <david.sherwood@arm.com>
2986 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2987 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2988 size as a poly_uint64.
2989 (mode_for_vector, mode_for_int_vector): Take the number of vector
2990 elements as a poly_uint64.
2991 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2992 size as a poly_uint64.
2993 (mode_for_vector, mode_for_int_vector): Take the number of vector
2994 elements as a poly_uint64.
2996 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2997 Alan Hayward <alan.hayward@arm.com>
2998 David Sherwood <david.sherwood@arm.com>
3000 * machmode.h (MACRO_MODE): New macro.
3001 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3002 * targhooks.c (default_libcall_value, default_secondary_reload)
3003 (default_memory_move_cost, default_register_move_cost)
3004 (default_class_max_nregs): Likewise.
3006 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3007 Alan Hayward <alan.hayward@arm.com>
3008 David Sherwood <david.sherwood@arm.com>
3010 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3011 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3013 * genautomata.c (main): Likewise.
3014 * genconditions.c (write_header): Likewise.
3015 * genemit.c (main): Likewise.
3016 * genextract.c (print_header): Likewise.
3017 * genopinit.c (main): Likewise.
3018 * genoutput.c (output_prologue): Likewise.
3019 * genpeep.c (main): Likewise.
3020 * genpreds.c (write_insn_preds_c): Likewise.
3021 * genrecog.c (writer_header): Likewise.
3022 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3023 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3024 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3025 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3026 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3027 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3028 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3029 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3030 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3031 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3032 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3033 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3034 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3035 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3036 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3037 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3038 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3039 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3040 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3041 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3042 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3043 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3044 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3045 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3046 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3047 * config/darwin.c (IN_TARGET_CODE): Likewise.
3048 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3049 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3050 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3051 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3052 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3053 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3054 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3055 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3056 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3057 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3058 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3059 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3060 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3061 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3062 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3063 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3064 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3065 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3066 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3067 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3068 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3069 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3070 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3071 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3072 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3073 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3074 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3075 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3076 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3077 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3078 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3079 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3080 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3081 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3082 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3083 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3084 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3085 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3086 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3087 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3088 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3089 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3090 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3091 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3092 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3093 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3094 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3095 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3096 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3097 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3098 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3099 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3100 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3101 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3102 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3103 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3104 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3105 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3106 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3107 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3108 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3109 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3110 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3111 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3112 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3113 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3114 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3115 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3116 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3117 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3118 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3119 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3120 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3121 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3122 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3123 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3124 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3125 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3126 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3127 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3128 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3129 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3130 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3131 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3132 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3133 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3134 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3135 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3136 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3137 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3138 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3139 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3140 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3141 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3142 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3143 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3144 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3145 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3146 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3147 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3148 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3149 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3150 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3151 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3152 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3153 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3155 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3156 Alan Hayward <alan.hayward@arm.com>
3157 David Sherwood <david.sherwood@arm.com>
3159 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3162 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3163 Alan Hayward <alan.hayward@arm.com>
3164 David Sherwood <david.sherwood@arm.com>
3166 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3167 * doc/md.texi (vec_series@var{m}): Document.
3168 * tree.def (VEC_SERIES_EXPR): New tree code.
3169 * tree.h (build_vec_series): Declare.
3170 * tree.c (build_vec_series): New function.
3171 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3172 * tree-pretty-print.c (dump_generic_node): Likewise.
3173 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3174 * tree-inline.c (estimate_operator_cost): Likewise.
3175 * expr.c (expand_expr_real_2): Likewise.
3176 * optabs-tree.c (optab_for_tree_code): Likewise.
3177 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3178 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3179 * expmed.c (make_tree): Handle VEC_SERIES.
3180 * optabs.def (vec_series_optab): New optab.
3181 * optabs.h (expand_vec_series_expr): Declare.
3182 * optabs.c (expand_vec_series_expr): New function.
3183 * tree-vect-generic.c (expand_vector_operations_1): Check that
3184 the operands also have vector type.
3186 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3187 Alan Hayward <alan.hawyard@arm.com>
3188 David Sherwood <david.sherwood@arm.com>
3190 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3191 (VEC_COND_EXPR): Add missing @tindex.
3192 * doc/md.texi (vec_duplicate@var{m}): Document.
3193 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3194 * tree.c (build_vector_from_val): Add stubbed-out handling of
3195 variable-length vectors, using VEC_DUPLICATE_EXPR.
3196 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3197 * cfgexpand.c (expand_debug_expr): Likewise.
3198 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3199 * tree-inline.c (estimate_operator_cost): Likewise.
3200 * tree-pretty-print.c (dump_generic_node): Likewise.
3201 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3202 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3203 (test_vec_duplicate_folding): New function.
3204 (fold_const_c_tests): Call it.
3205 * optabs.def (vec_duplicate_optab): New optab.
3206 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3207 * optabs.h (expand_vector_broadcast): Declare.
3208 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3209 vec_duplicate_optab.
3210 * expr.c (store_constructor): Try using vec_duplicate_optab for
3212 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3214 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3217 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3218 div/mod latencies a bit.
3220 2017-12-15 Jeff Law <law@redhat.com>
3222 PR tree-optimization/36550
3223 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3224 (mark_threaded_blocks): Rewrite code to avoid block copying when
3225 optimizing for size. Don't pessimize blocks which will be
3226 copied, but all the statements will be dead.
3228 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3230 PR tree-optimization/81165
3231 * tree-ssa-threadupdate.c (uses_in_bb): New.
3232 (estimate_threading_killed_stmts): New.
3233 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3234 * tree-ssa-threadedge.c
3235 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3238 2017-12-15 Jeff Law <law@redhat.com>
3240 PR tree-optimization/83410
3241 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3242 threads when parallelizing loops.
3244 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3246 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3248 * config/alpha/alpha.c (vms_attribute_table): Swap
3249 affects_type_identity and handler fields, adjust comments.
3250 * config/mips/mips.c (mips_attribute_table): Likewise.
3251 * config/visium/visium.c (visium_attribute_table): Likewise.
3252 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3253 * config/microblaze/microblaze.c (microblaze_attribute_table):
3255 * config/spu/spu.c (spu_attribute_table): Likewise.
3256 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3257 * config/arc/arc.c (arc_attribute_table): Likewise.
3258 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3259 * config/v850/v850.c (v850_handle_interrupt_attribute,
3260 v850_handle_data_area_attribute): Formatting fixes.
3261 (v850_attribute_table): Swap affects_type_identity and handler
3262 fields, adjust comments.
3263 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3264 * config/arm/arm.c (arm_attribute_table): Likewise.
3265 * config/avr/avr.c (avr_attribute_table): Likewise.
3266 * config/s390/s390.c (s390_attribute_table): Likewise.
3267 * config/sh/sh.c (sh_attribute_table): Likewise.
3268 * config/i386/i386.c (ix86_handle_cconv_attribute,
3269 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3270 (ix86_attribute_table): Swap affects_type_identity and handler
3271 fields, adjust comments.
3272 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3273 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3274 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3275 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3276 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3277 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3278 * config/rx/rx.c (rx_attribute_table): Likewise.
3279 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3280 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3281 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3282 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3283 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3284 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3285 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3286 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3287 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3288 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3289 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3291 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3293 (register_scoped_attributes, decl_attributes): Formatting fixes.
3295 PR tree-optimization/83269
3296 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3297 subtraction in arg0's type if type is signed and arg0 is unsigned.
3301 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3302 convert? on @0 instead of convert. Check type of @1, not @0.
3303 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3304 convert? on @0 instead of convert. Check type of @1, not @0.
3305 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3306 only optimize if either both @1 and @2 types are narrower
3307 precision, or both are wider or equal precision, and in the former
3308 case only if both have undefined overflow.
3310 2017-12-15 Richard Biener <rguenther@suse.de>
3313 * internal-fn.def (IFN_NOP): Add.
3314 * internal-fn.c (expand_NOP): Do nothing.
3315 * lto-streamer-in.c (input_function): Instead of removing
3316 sanitizer calls replace them with IFN_NOP calls.
3318 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3319 Alan Hayward <alan.hayward@arm.com>
3320 David Sherwood <david.sherwood@arm.com>
3322 * dse.c (store_info, read_info_type): Replace begin and end with
3324 (print_range): New function.
3325 (set_all_positions_unneeded, any_positions_needed_p)
3326 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3328 (record_store): Likewise. Optimize the case in which all positions
3330 (get_stored_val): Replace read_begin and read_end with read_offset
3332 (replace_read): Update call accordingly.
3334 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3336 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3337 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3338 (loop_cand::loop_cand): Initialize above members.
3339 (loop_cand::supported_operations): Delete.
3340 (loop_cand::can_interchange_p): Inline above function.
3341 (loop_cand::classify_simple_reduction): Record number of constant
3342 initialized simple reductions.
3343 (should_interchange_loops): New parameters. Check stmt cost of loops
3345 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3346 Update call to should_interchange_loops.
3347 (should_interchange_loop_nest): Update call to
3348 should_interchange_loops.
3350 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3353 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3354 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3355 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3357 2017-12-15 Julia Koval <julia.koval@intel.com>
3359 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3360 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3362 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3363 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3364 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3366 2017-12-15 Julia Koval <julia.koval@intel.com>
3368 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3369 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3370 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3371 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3372 _mm_aesenc_epi128): New intrinsics.
3374 2017-12-15 Julia Koval <julia.koval@intel.com>
3376 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3377 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3379 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3380 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3381 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3383 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3385 * gimple-ssa-strength-reduction.c (analyze_increments):
3386 Distinguish replacement costs for constant strides from those for
3389 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3391 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3393 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3395 * doc/invoke.texi: Document -Wcast-function-type.
3396 * recog.h (stored_funcptr): Change signature.
3397 * tree-dump.c (dump_node): Avoid warning.
3398 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3400 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3403 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3406 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3408 PR tree-optimization/83312
3409 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3410 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3411 "val" param, and to cope with arbitrary basic blocks.
3412 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3413 handle NULL_TREE for "val", dropping "bb" param.
3414 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3415 drop "bb" param. Handle NULL_TREE for "val".
3416 (find_case_label_for_value): Make "switch_stmt" param const.
3417 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3419 (vrp_prop::check_all_array_refs): Reimplement as...
3420 (check_array_bounds_dom_walker::before_dom_children): ...this new
3421 vfunc. Replace linear search through BB block list, excluding
3422 those with non-executable in-edges via dominator walk.
3424 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3426 * config/arm/arm.opt (mverbose-cost-dump): New option.
3427 * config/arm/arm.c (arm_rtx_costs): Use it.
3429 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3432 * reload1.c (emit_input_reload_insns): Skip debug markers.
3434 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3436 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3440 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3441 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3442 * config/sh/sh.c: Skip debug insns besides notes.
3443 * config/sh/sh.md: Likewise.
3444 * config/sh/sh_treg_combine.cc: Likewise.
3445 * config/sh/sync.md: Likewise.
3447 2017-12-14 Tom de Vries <tom@codesourcery.com>
3449 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3450 Add item for weak_undefined.
3452 2017-12-14 Richard Biener <rguenther@suse.de>
3454 PR tree-optimization/67842
3455 * sese.h (bb_in_region): Remove #if 0'ed code.
3457 2017-12-14 Richard Biener <rguenther@suse.de>
3459 PR tree-optimization/83326
3460 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3461 may_be_zero parameter and handle it by not marking the first
3462 peeled copy as not exiting the loop.
3463 (try_peel_loop): Likewise.
3464 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3465 to handle the case of constant or zero iterations and perform
3466 loop header copying on-the-fly.
3468 2017-12-14 Richard Biener <rguenther@suse.de>
3470 PR tree-optimization/83418
3471 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3472 Instead of asserting we don't get unfolded comparisons deal with them.
3474 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3477 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3478 basic blocks. Assert debug bind insns don't appear outside of bbs,
3479 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3482 PR tree-optimization/83198
3483 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3484 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3485 value if arg is a REAL_CST with incompatible type.
3487 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3488 Bin Cheng <bin.cheng@arm.com>
3491 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3493 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3496 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3498 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3499 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3501 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3502 Alan Hayward <alan.hayward@arm.com>
3503 David Sherwood <david.sherwood@arm.com>
3505 * poly-int.h: New file.
3506 * poly-int-types.h: Likewise.
3507 * coretypes.h: Include them.
3508 (POLY_INT_CONVERSION): Define.
3509 * target.def (estimated_poly_value): New hook.
3510 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3511 * doc/tm.texi: Regenerate.
3512 * doc/poly-int.texi: New file.
3513 * doc/gccint.texi: Include it.
3514 * doc/rtl.texi: Describe restrictions on subreg modes.
3515 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3516 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3517 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3518 * targhooks.h (default_estimated_poly_value): Declare.
3519 * targhooks.c (default_estimated_poly_value): New function.
3520 * target.h (estimated_poly_value): Likewise.
3521 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3522 (wi::unary_traits): Delete.
3523 (wi::binary_traits::signed_shift_result_type): Define for
3524 offset_int << HOST_WIDE_INT, etc.
3525 (generic_wide_int::operator <<=): Define for all types and use
3526 wi::lshift instead of <<.
3527 (wi::hwi_with_prec): Add a default constructor.
3528 (wi::ints_for): New class.
3529 (operator <<): Define for all wide-int types.
3530 (operator /): New function.
3531 (operator %): Likewise.
3532 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3533 (ASSERT_MAYBE_NE_AT): New macros.
3535 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3536 Dominik Vogt <vogt@linux.vnet.ibm.com>
3539 * emit-rtl.c (init_emit): Remove ??? comment.
3540 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3541 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3542 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3543 alignment of 3 virtual registers to BITS_PER_WORD.
3545 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3547 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3549 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3550 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3552 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3555 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3557 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3559 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3560 with the exception of debug begin stmt markers appear before
3564 * final.c (rest_of_handle_final): Call variable_tracking_main only
3565 if !flag_var_tracking.
3567 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3568 Jakub Jelinek <jakub@redhat.com>
3572 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3573 labels debug stmts that can only appear after labels.
3575 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3577 PR rtl-optimization/82398
3578 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3579 EXPR_USEFULNESS in priority comparison.
3581 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3583 PR rtl-optimization/83393
3584 * combine.c (move_deaths): If reg_stat points to a too new insn in
3585 last_death, do not use it: find the proper insn instead.
3587 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3590 * config/rs6000/rs6000-string.c (expand_block_compare,
3591 expand_strn_compare): Fix set_mem_size() calls.
3593 2017-12-12 Jeff Law <law@redhat.com>
3595 PR tree-optimization/83298
3596 PR tree-optimization/83362
3597 PR tree-optimization/83383
3598 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3599 push_value_range a public interface. Add new argument to
3600 record_ranges_from_stmt.
3601 * gimple-ssa-evrp-analyze.c
3602 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3603 Update comments. Handle recording temporary equivalences.
3604 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3605 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3606 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3607 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3608 gimple-ssa-evrp-analyze.h.
3609 (record_temporary_equivalences_from_phis): Add new argument. When
3610 the PHI arg is an SSA_NAME, set the result's range to the range
3612 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3613 from statements too.
3614 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3615 Pass it down to children as needed.
3616 (thread_outgoing_edges): Likewise.
3617 (thread_across_edge): Likewise. Push/pop range state as needed.
3618 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3620 2017-12-12 Julia Koval <julia.koval@intel.com>
3622 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3623 (PTA_CANNONLAKE): Remove PTA_CLWB.
3625 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3628 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3629 (vcondv2div2df): Likewise.
3630 (vconduv2dfv2di): Likewise.
3632 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3634 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3635 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3636 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3637 ATTR_MATHFN_FPROUNDING_ERRNO.
3639 2017-12-12 Richard Biener <rguenther@suse.de>
3641 PR tree-optimization/83385
3642 * graphite-scop-detection.c (get_order, order): Remove.
3643 (bb_to_rpo): New global.
3645 (build_scops): Sort pbbs in RPO order.
3647 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3649 * combine.c (simplify_set): Do not transform subregs to zero_extends
3650 if the destination mode is a vector mode.
3652 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3654 PR tree-optimization/83359
3655 * tree-cfg.h (fold_loop_internal_call): Declare.
3656 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3657 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3658 (find_loop_dist_alias): New function.
3659 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3660 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3663 PR tree-optimization/80631
3664 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3665 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3666 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3667 hardcoding zero as the value if COND_EXPR is never true. For
3668 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3669 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3670 hardcoding MAX_EXPR as the reduction operation.
3671 (is_nonwrapping_integer_induction): Allow negative step.
3672 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3673 vect_create_epilog_for_reduction, if no value is suitable, don't
3674 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3676 2017-12-12 Richard Biener <rguenther@suse.de>
3678 PR tree-optimization/81889
3679 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3680 range info from the non-wrapping IV instead of just the range
3683 2017-12-12 Julia Koval <julia.koval@intel.com>
3685 * config.gcc: Add vaesintrin.h.
3686 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3688 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3689 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3691 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3692 * config/i386/immintrin.h: Include vaesintrin.h.
3693 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3694 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3695 _mm_aesdec_epi128): New intrinsics.
3697 2017-12-12 Julia Koval <julia.koval@intel.com>
3699 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3700 OPTION_MASK_ISA_VAES_UNSET): New.
3701 (ix86_handle_option): Handle -mvaes.
3702 * config/i386/cpuid.h: Define bit_VAES.
3703 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3704 * config/i386/i386-c.c (__VAES__): New.
3705 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3706 (ix86_valid_target_attribute_inner_p): Ditto.
3707 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3708 * config/i386/i386.opt: Add -mvaes.
3709 * doc/invoke.texi: Ditto.
3711 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3713 * debug.h (gcc_debug_hooks): Add inline_entry.
3714 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3715 * debug.c (do_nothing_debug_hooks): Likewise.
3716 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3717 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3718 (dwarf2_lineno_debug_hooks): Likewise.
3720 * common.opt (gstatement-frontiers): New, setting
3721 debug_nonbind_markers_p.
3722 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3723 * toplev.c (process_options): Autodetect value for debug statement
3725 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3726 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3728 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3729 markers. Integrate source bind into debug stmt expand loop.
3730 (pass_expand::execute): Check debug marker limit. Avoid deep
3731 TER and expand debug locations for debug bind insns only.
3732 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3734 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3735 * final.c (reemit_insn_block_notes): Take current block from
3736 nonbind markers. Declare note where it's first set.
3737 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3738 begin stmt markers if enabled.
3739 (notice_source_line): Handle nonbind markers. Fail if their
3740 location is unknown or that of builtins.
3741 (rest_of_handle_final): Convert begin stmt markers to notes if
3742 var-tracking didn't run.
3743 (rest_of_clean_state): Skip begin stmt markers.
3744 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3746 * function.c (allocate_struct_function): Set begin_stmt_markers.
3747 * function.h (struct function): Add debug_marker_count counter
3748 and debug_nonbind_markers flag.
3749 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3750 * gimple-low.c (lower_function_body): Adjust
3751 debug_nonbind_markers.
3752 (lower_stmt): Drop or skip gimple debug stmts.
3753 (lower_try_catch): Skip debug stmts.
3754 * gimple.c (gimple_build_debug_begin_stmt): New.
3755 (gimple_copy): Increment debug_marker_count if copying one.
3756 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3757 * gimplify.c (rexpr_location): New.
3758 (rexpr_has_location): New.
3759 (warn_switch_unreachable_r): Handle gimple debug stmts.
3760 (shortcut_cond_r): Call expr_location.
3762 (find_goto_label): New.
3763 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3765 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3767 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3768 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3769 (LANG_HOOKS_INITIALIZER): ... this.
3770 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3771 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3773 (update_ebb_live_info): Skip debug insn markers.
3774 * lra.c (debug_insn_static_data): Rename to...
3775 (debug_bind_static_data): ... this.
3776 (debug_marker_static_data): New.
3777 (lra_set_insn_recog_data): Select one of the above depending
3779 (lra_update_isn_regno_info): Don't assume debug insns have
3781 (push_insns): Skip debug insns.
3782 * lto-streamer-in.c (input_function): Drop debug stmts
3783 depending on active options. Adjust debug_nonbind_markers.
3784 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3785 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3786 begin stmt marker notes.
3787 (print_insn): Likewise.
3788 * recog.c (extract_insn): Recognize rtl for debug markers.
3789 * rtl.def (DEBUG_MARKER): New.
3790 * tree-inline.c: Include params.h.
3791 (remap_gimple_stmt): Handle nonbind markers.
3792 (maybe_move_debug_stmts_to_successors): Likewise.
3793 (copy_debug_stmt): Likewise.
3794 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3795 markers regardless of no side effects.
3796 (tsi_link_before): Don't update container's side effects when adding
3797 a begin stmt marker.
3798 (tsi_link_after): Likewise.
3799 (expr_first): Skip begin stmt markers.
3800 (expr_last): Likewise.
3801 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3802 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3803 Disregard nonbind markers.
3804 * tree.c (make_node_stat): Don't set side effects for begin stmt
3806 (build1_stat): Likewise.
3807 * tree.def (DEBUG_BEGIN_STMT): New.
3808 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3809 * var-tracking.c (delete_debug_insns): Renamed to...
3810 (delete_vta_debug_insns): ... this.
3811 (reemit_marker_as_note): New.
3812 (vt_initialize): Reemit markers.
3813 (delete_vta_debug_insns): Likewise.
3814 (vt_debug_insns_local): Reemit or delete markers.
3815 (variable_tracking_main_1): Likewise.
3816 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3817 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3818 (gimple_debug_nonbind_marker_p): New.
3819 (gimple_build_debug_bind): Adjust.
3820 (gimple_build_debug_begin_stmt): New.
3821 * doc/invoke.texi (max-debug-marker-count): New param.
3822 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3823 (debug_parameter_ref, debug_marker): New.
3824 (NOTE_INSN_BEGIN_STMT): New.
3825 (DEBUG_INSN): Describe begin stmt markers.
3827 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3828 without debug insns, we wouldn't, but clean up debug insns
3829 after a control flow insn nevertheless.
3831 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3832 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3833 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3834 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3835 (rtl_tidy_fallthru_edge): Likewise.
3836 (rtl_verify_fallthru): Likewise.
3837 (rtl_verify_bb_layout): Likewise.
3838 (skip_insns_after_block): Likewise.
3839 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3840 * dwarf2out.c: Include print-rtl.h.
3841 (dwarf2out_next_real_insn): New.
3842 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3843 Dump debug binds in asm comments.
3844 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3845 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3846 callers to use gsi_start_nondebug_bb instead.
3847 (gsi_after_labels): Skip gimple debug stmts.
3848 (gsi_start_nondebug): New.
3849 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3850 (proper_loop_form_for_interchange): Adjust.
3851 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3852 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3853 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3854 (collect_fallthrough_labels): Likewise.
3855 (should_warn_for_implicit_fallthrough): Likewise.
3856 (warn_implicit_fallthrough_r): Likewise.
3857 (expand_FALLTHROUGH_r): Likewise.
3858 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3859 (graphite_copy_stmts_from_block): Skip nonbind markers.
3860 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3861 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3862 * jump.c (clean_barriers): Skip debug insns.
3863 * omp-expand.c (expand_parallel_call): Skip debug insns.
3864 (expand_task_call): Likewise.
3865 (remove_exit_barrier): Likewise.
3866 (expand_omp_taskreg): Likewise.
3867 (expand_omp_for_init_counts): Likewise.
3868 (expand_omp_for_generic): Likewise.
3869 (expand_omp_for_static_nochunk): Likewise.
3870 (expand_omp_for_static_chunk): Likewise.
3871 (expand_omp_simd): Likewise.
3872 (expand_omp_taskloop_for_outer): Likewise.
3873 (expand_omp_taskloop_for_inner): Likewise.
3874 (expand_oacc_for): Likewise.
3875 (expand_omp_sections): Likewise.
3876 (expand_omp_single): Likewise.
3877 (expand_omp_synch): Likewise.
3878 (expand_omp_atomic_load): Likewise.
3879 (expand_omp_atomic_store): Likewise.
3880 (expand_omp_atomic_fetch_op): Likewise.
3881 (expand_omp_atomic_pipeline): Likewise.
3882 (expand_omp_atomic_mutex): Likewise.
3883 (expand_omp_target): Likewise.
3884 (grid_expand_omp_for_loop): Likewise.
3885 (grid_expand_target_grid_body): Likewise.
3886 (build_omp_regions_1): Likewise.
3887 * omp-low.c (check_combined_parallel): Skip debug stmts.
3888 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3889 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3890 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3892 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3893 (make_edges): Likewise.
3894 (cleanup_dead_labels): Likewise.
3895 (gimple_can_merge_blocks_p): Likewise.
3896 (stmt_starts_bb_p): Likewise.
3897 (gimple_block_label): Likewise.
3898 (gimple_redirect_edge_and_branch): Likewise.
3899 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3901 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3903 * tree-pretty-print (print_declaration): Omit initializer in slim
3905 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3907 (eliminate_unnecessary_stmts): Stabilize block removal order.
3908 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3909 * var-tracking.c (get_first_insn): New.
3910 (vt_emit_notes): Call it.
3911 (vt_initialize): Walk any insns before the first BB.
3912 (delete_debug_insns): Likewise.
3914 * gimple.h (enum gimple_debug_subcode): Add
3915 GIMPLE_DEBUG_BEGIN_STMT.
3916 (gimple_debug_begin_stmt_p): New.
3917 (gimple_debug_nonbind_marker_p): New.
3918 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3919 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3920 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3921 * insn-notes.def (BEGIN_STMT): New.
3922 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3923 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3924 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3925 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3926 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3927 (INSN_DEBUG_MARKER_KIND): New.
3928 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3929 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3930 (INSN_VAR_LOCATION_PTR): New.
3931 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3933 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3934 deps only if debug bind insns are enabled.
3935 (pass_expand::execute): Avoid deep TER and expand
3936 debug locations for debug bind insns only.
3937 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3938 debug stmts special handling down to debug bind stmts.
3939 * combine.c (try_combine): Narrow debug insns special handling
3940 down to debug bind insns.
3941 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3942 Narrow debug insns preexisting special handling down to debug
3944 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3945 handling down to debug bind insns.
3946 * function.c (instantiate_virtual_regs): Skip debug markers,
3947 adjust handling of debug binds.
3948 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3949 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3950 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3951 insns to debug bind insns.
3952 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3953 special handling of debug stmts to debug bind stmts.
3954 * ipa-split.c (split_function): Likewise.
3955 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3956 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3958 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3959 * regrename.c (build_def_use): Likewise.
3960 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3961 (pass_cprop_hardreg): Narrow special casing of debug insns to
3963 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3964 * reload1.c (reload): Likewise.
3965 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3966 casing of debug stmts to debug bind stmts.
3967 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3968 * ssa-iterators.h (num_imm_uses): Likewise.
3969 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3970 debug stmts to debug bind stmts.
3971 * tree-inline.c (tree_function_versioning): Narrow special casing
3972 of debug stmts to debug bind stmts.
3973 * tree-loop-distribution.c (generate_loops_for_partition):
3974 Narrow special casing of debug stmts to debug bind stmts.
3975 * tree-sra.c (analyze_access_subtree): Narrow special casing
3976 of debug stmts to debug bind stmts.
3977 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3978 stmts to debug bind stmts.
3979 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3980 casing of debug stmts to debug bind stmts.
3981 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3982 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3983 casing of debug stmts to debug bind stmts.
3984 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3986 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3987 debug stmts to debug bind stmts.
3988 (gimple_replace_ssa_lhs): Likewise.
3989 (insert_debug_temp_for_var_def): Likewise.
3990 (insert_debug_temps_for_defs): Likewise.
3991 (reset_debug_uses): Likewise.
3992 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3993 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3994 (adjust_debug_stmts): Likewise.
3995 (adjust_phi_and_debug_stmts): Likewise.
3996 (vect_do_peeling): Likewise.
3997 * tree-vect-loop.c (vect_transform_loop): Likewise.
3998 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3999 * var-tracking.c (adjust_mems): Narrow special casing of debug
4000 insns to debug bind insns.
4001 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4002 (compute_bb_dataflow, vt_find_locations): Likewise.
4003 (vt_expand_loc, emit_notes_for_changes): Likewise.
4004 (vt_init_cfa_base): Likewise.
4005 (vt_emit_notes): Likewise.
4006 (vt_initialize): Likewise.
4007 (vt_finalize): Likewise.
4009 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4010 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4011 (next_nonnote_nondebug_insn_bb): New.
4012 (prev_nonnote_nondebug_insn_bb): New.
4013 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4014 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4015 (prev_nonnote_nondebug_insn_bb): Declare.
4016 (next_nonnote_nondebug_insn_bb): Declare.
4017 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4018 * cfgrtl.c (get_last_bb_insn): Likewise.
4019 * lra.c (push_insns): Likewise.
4021 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4024 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4026 * selftest.c (selftest::test_runner::test_runner): New ctor.
4027 (selftest::test_runner::~test_runner): New dtor.
4028 * selftest.h (class selftest::test_runner): New class.
4030 2017-12-11 Carl Love <cel@us.ibm.com>
4032 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4033 vec_extract_fp32_from_shortl]): Add #defines.
4034 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4035 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4036 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4037 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4038 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4039 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4040 * doc/extend.texi: Add documentation for the added builtins.
4042 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4044 PR rtl-optimization/80693
4045 PR rtl-optimization/81019
4046 PR rtl-optimization/81020
4047 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4048 are not mentioned in i3. Place the REG_UNUSED note on i2,
4049 possibly modified to REG_DEAD, if it did not originate in i3.
4051 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4053 * recog.c (store_data_bypass_p_1): New function.
4054 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4055 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4057 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4059 PR rtl-optimization/83361
4060 * ifcvt.c (if_convert): Call fixup_partitions.
4062 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4064 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4065 early folding of splat_u{8,16,32}.
4067 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4069 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4070 output_operand_lossage first argument with capital letter.
4071 (aarch64_override_options): Don't start error and sorry first argument
4072 with capital letter.
4074 2017-12-11 Andi Kleen <ak@linux.intel.com>
4076 PR gcov-profile/83355
4077 * auto-profile.c (string_table::get_index_by_decl): Don't
4078 recurse when abstract origin points to itself.
4080 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4082 PR tree-optimization/83320
4083 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4084 (prune_datarefs_not_in_loop): Ditto.
4086 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4088 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4090 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4092 PR tree-optimization/83337
4093 * gimple-loop-interchange.cc (compute_access_stride): Handle
4094 bitfield DRs properly.
4096 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4098 PR tree-optimization/83338
4099 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4100 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4101 vector of unsigned integers to vector of signed integers.
4103 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4105 PR rtl-optimization/83317
4106 * lra-constraints.c (process_address_1): Add insn code check.
4108 2017-12-08 Michael Matz <matz@suse.de>
4110 Fix PR tree-optimization/83323
4111 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4112 head-controlled loops and loop BBs.
4113 * common.opt (funroll-and-jam): Remove, instead ...
4114 (floop-unroll-and-jam): ... reuse this option.
4115 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4116 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4117 (-floop-unroll-and-jam): ... this option.
4119 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4121 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4122 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4123 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4125 2017-12-08 Richard Biener <rguenther@suse.de>
4128 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4129 handle accesses outside of zero-sized vars.
4131 2017-12-08 Martin Jambor <mjambor@suse.cz>
4133 PR tree-optimization/83141
4134 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4135 test for MEM_REFs implicitely changing types with padding. Remove
4137 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4139 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4141 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4143 * config/avr/avr.c (avr_attribute_table): Likewise.
4144 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4145 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4146 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4147 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4148 of struct attribute_spec.
4149 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4151 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4154 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4156 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4159 * profile-count.c (profile_count::from_gcov_type): Move from
4160 profile-count.h; handle overflow.
4161 * profile-count.h (profile_count::from_gcov_type): Move offline.
4163 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4165 PR rtl-optimization/83304
4166 * combine.c (move_deaths): If we do not know where a register died,
4169 2017-12-08 Richard Biener <rguenther@suse.de>
4171 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4172 Provide -fopt-info-loop feedback when we interchange in a nest.
4174 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4176 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4177 for armv6 ARM CPU IDs.
4179 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4181 * common/config/arm/arm-common.c: Include <algorithm>.
4182 (INCLUDE_VECTOR): Define.
4183 (compare_opt_names): New function.
4184 (arm_rewrite_selected_arch): Only strip out extensions that can be
4185 expressed through -mfpu. Sort the remaining extensions
4188 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4190 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4191 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4192 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4193 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4195 2017-06-08 Tristan Gingold <gindold@adacore.com>
4198 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4199 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4201 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4203 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4204 of struct attribute_spec.
4206 2017-12-08 Julia Koval <julia.koval@intel.com>
4208 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4209 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4210 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4211 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4212 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4213 _mm_maskz_dpwssds_epi32): Ditto.
4215 2017-12-08 Richard Biener <rguenther@suse.de>
4217 PR tree-optimization/81303
4218 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4219 conditions try to create a comparison vector type matching
4220 the data vector type.
4221 (vectorizable_condition): Adjust.
4222 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4223 Leave invariant conditions alone in case we can vectorize those.
4225 2017-12-08 Julia Koval <julia.koval@intel.com>
4227 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4228 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4229 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4230 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4231 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4233 2017-12-08 Julia Koval <julia.koval@intel.com>
4235 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4236 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4237 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4238 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4239 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4240 _mm_maskz_dpbusds_epi32): New intrinsics.
4242 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4244 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4246 (ld<bh>io_signed): Likewise.
4247 (st<bhw>io): Likewise.
4248 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4249 SMALL_INT12 constant integer operand.
4251 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4253 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4255 (nios2_large_constant_allowed): ...this. Adjust uses.
4256 (nios2_plus_symbolic_constant_p): Rename to...
4257 (nios2_plus_large_constant_p): ...this. Adjust uses.
4258 (nios2_legitimate_address_p): Correct CONST_INT handling.
4259 (nios2_symbolic_memory_operand_p): Rename to...
4260 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4261 (nios2_large_constant_p): Check for large constant integers too.
4262 (nios2_split_large_constant): Handle constant integers.
4263 (nios2_split_symbolic_memory_operand): Rename to...
4264 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4265 (nios2_legitimize_constant_address): Handle constant integers.
4266 (r0rel_constant_p): Handle small constant integers.
4267 (nios2_print_operand_address): Handle r0-relative integer addresses.
4268 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4269 * config/nios2/nios2.md: Adjust for renamed functions.
4271 2017-12-07 Andrew Waterman <andrew@sifive.com>
4273 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4274 (TARGET_HAVE_SRODATA_SECTION): New define.
4275 (riscv_select_section): New function.
4277 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4278 Alexander Monakov <amonakov@ispras.ru>
4279 Jakub Jelinek <jakub@redhat.com>
4282 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4284 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4286 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4287 a bool success value. Don't call output_operand_lossage here.
4288 (aarch64_print_ldpstp_address): Return a bool success value.
4289 (aarch64_print_operand_address): Call output_addr_const if
4290 aarch64_print_address_internal fails.
4291 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4292 'y'; call output_operand_lossage instead. Call output_operand_lossage
4293 if aarch64_print_ldpstp_address fails.
4295 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4297 * tree-vector-builder.h
4298 (tree_vector_builder::binary_encoded_nelts): Declare.
4299 * tree-vector-builder.c
4300 (tree_vector_builder::binary_encoded_nelts): New function.
4301 * fold-const.c (negate_expr_p): Likewise.
4302 (operand_equal_p, fold_checksum_tree): Likewise.
4303 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4304 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4305 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4306 (uniform_vector_p): Likewise.
4307 * varasm.c (const_hash_1, compare_constant): Likewise.
4308 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4309 (valid_lattice_transition): Operate directly on the VECTOR_CST
4311 * ipa-icf.c: Include tree-vector-builder.h.
4312 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4313 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4315 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4317 * tree.c (build_vector): Delete.
4318 * tree.h (build_vector): Make static and move into the self-testing
4321 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4323 * vector-builder.h (vector_builder::derived): New const overload.
4324 (vector_builder::elt): New function.
4325 * tree-vector-builder.h (tree_vector_builder::type): New function.
4326 (tree_vector_builder::apply_step): Declare.
4327 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4329 * gimple-fold.h (tree_vector_builder): Declare.
4330 (gimple_build_vector): Take a tree_vector_builder instead of a
4331 type and vector of elements.
4332 * gimple-fold.c (gimple_build_vector): Likewise.
4333 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4335 (get_initial_defs_for_reduction): Likewise.
4336 (vectorizable_induction): Likewise.
4338 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4340 * tree-vector-builder.h
4341 (tree_vector_builder::new_binary_operation): Declare.
4342 * tree-vector-builder.c
4343 (tree_vector_builder::new_binary_operation): New function.
4344 * fold-const.c (fold_relational_const): Use it.
4345 (const_binop): Likewise. Check that both input vectors have
4346 the same number of elements, thus excluding things like WIDEN_SUM.
4347 Check whether it is possible to operate directly on the encodings
4350 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4352 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4353 new_unary_operation, operating only on the encoded elements.
4354 (const_unop): Likewise.
4355 (exact_inverse): Likewise.
4356 (distributes_over_addition_p): New function.
4357 (const_binop): Use tree_vector_builder and new_unary_operation
4358 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4359 on the encoded elements unless the encoding is strided and the
4360 operation does not distribute over addition.
4361 (fold_convert_const): Use tree_vector_builder and
4362 new_unary_operation. Operate only on the encoded elements
4363 for truncating integer conversions, or for non-stepped encodings.
4365 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4367 * config/sparc/sparc.c: Include tree-vector-builder.h.
4368 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4369 * expmed.c: Include tree-vector-builder.h.
4370 (make_tree): Use tree_vector_builder instead of build_vector.
4371 * fold-const.c: Include tree-vector-builder.h.
4372 (const_binop): Use tree_vector_builder instead of build_vector.
4373 (const_unop): Likewise.
4374 (native_interpret_vector): Likewise.
4375 (fold_vec_perm): Likewise.
4376 (fold_ternary_loc): Likewise.
4377 * gimple-fold.c: Include tree-vector-builder.h.
4378 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4380 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4381 (simplify_vector_constructor): Use tree_vector_builder instead
4383 * tree-vect-generic.c: Include tree-vector-builder.h.
4384 (add_rshift): Use tree_vector_builder instead of build_vector.
4385 (expand_vector_divmod): Likewise.
4386 (optimize_vector_constructor): Likewise.
4387 * tree-vect-loop.c: Include tree-vector-builder.h.
4388 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4389 of build_vector. Explicitly use a stepped encoding for
4391 * tree-vect-slp.c: Include tree-vector-builder.h.
4392 (vect_get_constant_vectors): Use tree_vector_builder instead
4394 (vect_transform_slp_perm_load): Likewise.
4395 (vect_schedule_slp_instance): Likewise.
4396 * tree-vect-stmts.c: Include tree-vector-builder.h.
4397 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4398 (vect_gen_perm_mask_any): Likewise.
4399 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4400 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4402 (build_vector_from_val): Likewise. Explicitly use a duplicate
4405 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4407 * doc/generic.texi (VECTOR_CST): Describe new representation of
4409 * vector-builder.h: New file.
4410 * tree-vector-builder.h: Likewise.
4411 * tree-vector-builder.c: Likewise.
4412 * Makefile.in (OBJS): Add tree-vector-builder.o.
4413 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4414 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4415 (tree_vector): Change the number of elements to
4416 vector_cst_encoded_nelts.
4417 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4418 (VECTOR_CST_ELTS): Delete.
4419 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4420 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4421 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4422 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4423 (VECTOR_CST_ENCODED_ELT): Likewise.
4424 (vector_cst_encoded_nelts): New function.
4425 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4426 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4427 (vector_cst_int_elt, vector_cst_elt): Declare.
4428 * tree.c: Include tree-vector-builder.h.
4429 (tree_code_size): Abort if passed VECTOR_CST.
4430 (tree_size): Update for new VECTOR_CST layout.
4431 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4432 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4433 (build_vector): Use tree_vector_builder.
4434 (vector_cst_int_elt, vector_cst_elt): New functions.
4435 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4436 encoded elements and then create the vector in the canonical form.
4437 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4438 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4439 (tree_c_tests): Call test_vector_cst_patterns.
4440 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4442 (hash_tree): Likewise.
4443 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4444 (streamer_write_tree_header): Likewise.
4445 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4446 (streamer_alloc_tree): Likewise. Update call to make_vector.
4447 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4449 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4451 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4452 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4453 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4455 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4456 Richard Biener <rguenther@suse.de>
4458 PR tree-optimization/81303
4459 * Makefile.in (gimple-loop-interchange.o): New object file.
4460 * common.opt (floop-interchange): Reuse the option from graphite.
4461 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4462 -floop-interchange and mention it for -O3.
4463 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4464 * gimple-loop-interchange.cc: New file.
4465 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4466 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4467 * passes.def (pass_linterchange): New pass.
4468 * timevar.def (TV_LINTERCHANGE): New time var.
4469 * tree-pass.h (make_pass_linterchange): New declaration.
4470 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4471 interchange. Record IV before/after increment in new parameters.
4472 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4473 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4475 (check_reduction_path): ...New function here.
4476 * tree-vectorizer.h (check_reduction_path): New declaration.
4478 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4481 PR rtl-optimization/80818
4482 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4483 always early clobbered.
4484 * lra-lives.c (process_bb_lives): Check input hard regs for early
4485 clobbered non-operand hard reg.
4487 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4490 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4491 types_compatible_p, just that TYPE_MODE is the same.
4493 2017-12-07 Martin Sebor <msebor@redhat.com>
4496 * attribs.c (empty_attribute_table): Initialize new member of
4497 struct attribute_spec.
4498 (decl_attributes): Add argument. Handle mutually exclusive
4499 combinations of attributes.
4500 (selftests::test_attribute_exclusions): New function.
4501 (selftests::attribute_c_tests): Ditto.
4502 * attribs.h (decl_attributes): Add default argument.
4503 * selftest.h (attribute_c_tests): Declare.
4504 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4505 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4507 * doc/extend.texi (Common Function Attributes): Update const and pure.
4508 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4509 of struct attribute_spec.
4510 * config/arc/arc.c (arc_attribute_table): Same.
4511 * config/arm/arm.c (arm_attribute_table): Same.
4512 * config/avr/avr.c ( avr_attribute_table): Same.
4513 * config/bfin/bfin.c (bfin_attribute_table): Same.
4514 * config/cr16/cr16.c (cr16_attribute_table): Same.
4515 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4516 * config/h8300/h8300.c (h8300_attribute_table): Same.
4517 * config/i386/i386.c (ix86_attribute_table): Same.
4518 * config/ia64/ia64.c (ia64_attribute_table): Same.
4519 * config/m32c/m32c.c (m32c_attribute_table): Same.
4520 * config/m32r/m32r.c (m32r_attribute_table): Same.
4521 * config/m68k/m68k.c (m68k_attribute_table): Same.
4522 * config/mcore/mcore.c (mcore_attribute_table): Same.
4523 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4524 * config/mips/mips.c (mips_attribute_table): Same.
4525 * config/msp430/msp430.c (msp430_attribute_table): Same.
4526 * config/nds32/nds32.c (nds32_attribute_table): Same.
4527 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4528 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4529 * config/rl78/rl78.c (rl78__attribute_table): Same.
4530 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4531 * onfig/rx/rx.c (rx_attribute_table): Same.
4532 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4533 * config/sh/sh.c (sh_attribute_table): Same.
4534 * config/sparc/sparc.c (sparc_attribute_table): Same.
4535 * config/spu/spu.c (spu_attribute_table): Same.
4536 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4537 * config/v850/v850.c (v850_attribute_table): Same.
4538 * config/visium/visium.c (visium_attribute_table): Same.
4540 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4543 * config/arm/arm.c (INCLUDE_STRING): Define.
4544 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4545 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4548 2017-12-07 Michael Matz <matz@suse.de>
4550 Add unroll and jam pass
4552 * gimple-loop-jam.c: New file.
4553 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4554 * common.opt (funroll-and-jam): New option.
4555 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4556 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4557 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4558 * passes.def: Add pass_loop_jam.
4559 * timevar.def (TV_LOOP_JAM): Add.
4560 * tree-pass.h (make_pass_loop_jam): Declare.
4561 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4562 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4563 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4564 to flow_loop_tree_node_add.
4565 (duplicate_subloops, copy_loops_to): Append to sibling list.
4566 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4567 * doc/invoke.texi (-funroll-and-jam): Document new option.
4568 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4570 2017-12-07 Richard Biener <rguenther@suse.de>
4572 PR tree-optimization/83296
4573 PR tree-optimization/67769
4574 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4575 flow sensitive info in an unrelated BB.
4576 (value_replacement): Use reset_flow_sensitive_info.
4577 (minmax_replacement): Reset flow sensitive info on the def
4578 we move. Do not reset flow sensitive info in the whole BB
4579 we move the stmt to.
4580 (abs_replacement): Likewise.
4582 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4585 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4586 rs6000_cpu to the given -mcpu=, or to the default processor.
4588 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4590 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4591 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4592 instead of rs6000_cpu_attr.
4593 (rs6000_adjust_cost): Ditto.
4594 (is_microcoded_insn): Ditto.
4595 (rs6000_adjust_priority): Ditto.
4596 (rs6000_issue_rate): Ditto.
4597 (rs6000_use_sched_lookahead): Ditto.
4598 (rs6000_use_sched_lookahead_guard): Ditto.
4599 (rs6000_sched_reorder): Ditto.
4600 (force_new_group): Ditto.
4601 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4602 (group_ending_nop): Ditto.
4604 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4606 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4607 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4608 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4609 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4611 (rs6000_reassociation_width): Ditto.
4612 (rs6000_emit_epilogue): Ditto.
4613 (rs6000_adjust_cost): Ditto.
4614 (is_microcoded_insn): Ditto.
4615 (is_cracked_insn): Ditto.
4616 (rs6000_adjust_priority): Ditto.
4617 (rs6000_sched_reorder): Ditto.
4618 (rs6000_sched_reorder2): Ditto.
4619 (insn_must_be_first_in_group): Ditto.
4620 (insn_must_be_last_in_group): Ditto.
4621 (rs6000_register_move_cost): Ditto.
4622 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4625 2017-12-07 Julia Koval <julia.koval@intel.com>
4627 * config.gcc: Add vaesintrin.h.
4628 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4630 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4631 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4633 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4634 * config/i386/immintrin.h: Include vaesintrin.h.
4635 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4636 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4637 _mm_aesdec_epi128): New intrinsics.
4639 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4641 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4642 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4643 c-family/c-spellcheck.cc.
4644 (best_macro_match::best_macro_match): Likewise.
4646 (struct edit_distance_traits<cpp_hashnode *>): Move to
4647 c-family/c-spellcheck.h.
4648 (class best_macro_match): Likewise.
4650 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4652 PR tree-optimization/83293
4653 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4654 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4655 might insert into empty bb.
4658 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4659 simplify for plus with :c added, and pointer_plus without that.
4660 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4661 with undefined overflow and the conversion is not widening,
4662 perform negation in utype and only convert to type afterwards.
4663 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4664 simplify for plus with :c added, and pointer_plus without that.
4665 If type is integral with undefined overflow and the conversion is
4666 not widening, perform minus in utype and only convert to type
4667 afterwards. Move the last pointer_diff_expr simplify into the
4670 2017-12-06 Martin Sebor <msebor@redhat.com>
4672 PR tree-optimization/82646
4673 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4674 strncpy, not maxlen.
4676 2017-12-06 Martin Sebor <msebor@redhat.com>
4678 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4681 PR tree-optimization/83075
4682 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4683 strncat/strncpy don't change length of source string.
4685 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4688 2017-11-29 Martin Aberg <maberg@gaisler.com>
4690 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4691 to prevent b2bst errata sequence.
4692 (sqrtdf2_fix): Likewise.
4694 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4696 PR tree-optimization/81945
4697 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4698 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4699 to dest_cfun has orig_loop_num set, either remap it to the new
4700 loop number if the loop got moved too, or clear it.
4702 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4704 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4706 (thunderx2t99_nothing): New insn reservation.
4707 (thunderx2t99_mrs): New insn reservation.
4708 (thunderx2t99_multiple): New insn reservation.
4709 (thunderx2t99_alu_basi): Add bfx to reservation.
4710 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4712 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4715 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4717 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4719 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4720 tree-ssa-pre.c::remove_dead_inserted_code.
4721 * tree-ssa-dce.h: New file.
4722 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4723 (remove_dead_inserted_code): Move and rename to function
4724 tree-ssa-dce.c::simple_dce_from_worklist.
4725 (pass_pre::execute): Update use.
4727 2017-12-05 Richard Biener <rguenther@suse.de>
4729 PR tree-optimization/83277
4730 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4731 to code-gen liveout vars.
4733 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4735 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4736 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4737 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4740 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4742 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4744 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4745 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4748 2017-12-05 Richard Biener <rguenther@suse.de>
4750 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4751 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4752 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4753 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4755 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4757 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4759 2017-12-05 Martin Liska <mliska@suse.cz>
4760 Jakub Jelinek <jakub@redhat.com>
4762 * doc/invoke.texi: Document the options.
4763 * flag-types.h (enum sanitize_code): Add
4764 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4765 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4766 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4767 * opts.c: Define new sanitizer options.
4768 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4769 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4771 2017-12-05 Julia Koval <julia.koval@intel.com>
4773 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4774 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4775 (ix86_handle_option): Handle -mavx512vnni.
4776 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4777 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4778 * config/i386/i386-c (__AVX512VNNI__): New.
4779 * config/i386/i386.c (ix86_target_string): Handle new option.
4780 (ix86_valid_target_attribute_inner_p): Handle new option.
4781 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4782 * config/i386/i386.opt (mavx512vnni): New option.
4784 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4787 * athlon.md: Disable for generic.
4788 * haswell.md: Enable for generic.
4789 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4790 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4792 (ix86_adjust_cost): Move generic to haswell path.
4794 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4796 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4797 instead of MEM_P in a couple more places. Fix formatting issues.
4799 2017-12-04 Jim Wilson <jimw@sifive.com>
4801 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4802 instead of GP_REG_LAST-1.
4803 (riscv_adjust_libcall_cfi_prologue): Likewise.
4804 (riscv_adjust_libcall_cri_epilogue): Likewise.
4805 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4808 2017-12-04 Luis Machado <luis.machado@linaro.org>
4810 * ipa-pure-const.c (check_decl): Add missing newline.
4811 (state_from_flags): Likewise.
4813 2017-12-04 Jeff Law <law@redhat.com>
4815 PR tree-optimizatin/78496
4816 * gimple-ssa-evrp-analyze.h
4817 (evrp_range_analyzer::get_vr_values): Simplify.
4818 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4819 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4820 and gimple-ssa-evrp-analyze.h.
4821 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4822 (simplify_stmt_for_jump_threading): Copy a blob of code from
4823 tree-vrp.c to use ranges to simplify statements.
4824 (dom_opt_dom_walker::before_dom_children): Call
4825 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4826 (dom_opt_dom_walker::after_dom_children): Similarly for
4827 evrp_range_analyzer::leave.
4828 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4831 * gimple-ssa-evrp-analyze.c
4832 (evrp_range_analyzer::extract_range_from_stmt): Always use
4833 vr_values::update_value_range so preexisting range info is
4834 medged with new range info, even if the new range is VR_VARYING.
4836 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4838 * combine.c: Adjust comment.
4839 (use_crosses_set_p): Delete.
4840 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4841 (try_combine): Ditto.
4843 2017-12-04 Richard Biener <rguenther@suse.de>
4845 PR tree-optimization/83255
4846 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4847 Re-add zero-iteration check.
4849 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4851 PR rtl-optimization/83245
4852 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4853 hard registers as earlyclobber, also if not in an asm.
4855 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4859 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4862 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4863 rs6000_cpu based on cpu_index, not tune_index.
4865 2017-12-04 Richard Biener <rguenther@suse.de>
4867 PR tree-optimization/83238
4868 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4869 code match comment, rejecting invalid SESE regions.
4871 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4873 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4874 require base operand is a REG_POINTER prior to reload on targets
4875 with non-equivalent space registers.
4877 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4879 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4880 (update_specialized_profile): Fix updating of counts.
4881 (perhaps_add_new_callers): Likewise.
4883 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4886 * x86-tune.def: Remove obsolette FIXMEs.
4887 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4888 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4889 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4891 (X86_TUNE_PAD_RETURNS): Disable for generic.
4893 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4895 PR tree-optimization/83170
4896 PR tree-optimization/83241
4897 * gimple-ssa-store-merging.c
4898 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4899 gimple_vuse (ins_stmt) in case it has changed.
4900 (imm_store_chain_info::output_merged_store): Likewise.
4902 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4906 * tree-cfg.c (pass_warn_function_return::execute): Handle
4907 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4908 with BUILTINS_LOCATION.
4912 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4913 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4914 instead of DECL_MODE.
4916 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4917 last argument to rtx pointer.
4918 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4919 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4920 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4921 depending on the chosen ISAs.
4922 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4923 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4924 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4926 * config/i386/sse.md (mov<mode>_internal): Likewise.
4927 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4929 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4931 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4933 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4935 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4936 parameters from prototype.
4937 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4938 parameters. Don't print a comment.
4939 (emit_fusion_gpr_load): Adjust.
4940 (emit_fusion_load_store): Adjust.
4941 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4942 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4943 comment on the second line.
4945 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4948 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4949 rs6000_cpu based on cpu_index, not tune_index.
4951 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4953 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4954 which_alternative instead of which_alternative + 1.
4955 (output_asm_insn): Print an extra tab if the template is short.
4957 2017-12-01 Jim Wilson <jimw@sifive.com>
4959 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4961 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4962 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4964 * doc/tm.texi: Regenerate.
4966 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4969 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4970 whether we can allocate pseudos before trying to fix an address.
4971 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4972 memory address is indexed or indirect.
4973 (floatuns_<mode>si2_hw2): Likewise.
4975 2017-12-01 Jason Merrill <jason@redhat.com>
4977 * Makefile.in (TAGS): Add c-family/*.cc.
4979 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4981 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4982 (call_value_insn): Likewise.
4983 (sibcall_insn): Likewise.
4984 (sibcall_value_insn): Likewise.
4985 (movsi_aarch64): Likewise.
4986 (movdi_aarch64): Likewise.
4987 (add_losym_): Likewise.
4988 (ldr_got_small_): Likewise.
4989 (ldr_got_small_sidi): Likewise.
4990 (ldr_got_small_28k_): Likewise.
4991 (ldr_got_small_28k_sidi): Likewise.
4992 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4993 Move output_addr_const to symbolic case. Add error check.
4995 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4997 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4998 * config/s390/s390.md: Disable MVC merging peephole if it would
4999 disable operand forwarding.
5000 (new peephole2): Split MVCs if it would turn them into up to 2
5003 2017-12-01 Richard Biener <rguenther@suse.de>
5005 PR tree-optimization/83232
5006 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5007 detection of same access. Instead of breaking the group here
5008 do not consider the duplicate. Add comment explaining real fix.
5010 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5012 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5014 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5016 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5018 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5020 * function.h (struct function): Remove cilk_frame_decl,
5021 is_cilk_function and calls_cilk_spawn fields.
5022 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5024 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5025 cilk_elemental field.
5026 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5027 * target.def: Adjust comment.
5028 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5029 Don't test cilk_elemental.
5031 PR tree-optimization/83233
5032 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5033 bswap_stat name for the struct.
5036 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5037 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5038 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5039 SWITCH_BREAK_LABEL_P set on the label.
5040 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5041 added for default case if it was missing and not all cases covered.
5042 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5043 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5045 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5046 to avoid -Wimplicit-fallthrough warning.
5047 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5048 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5050 PR tree-optimization/83221
5051 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5053 (init_reassoc): Formatting fix.
5056 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5057 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5061 * tree-cfg.c: Include asan.h.
5062 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5064 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5066 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5068 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5070 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5071 vpcmpeqd instruction.
5073 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5075 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5076 registers type for 128bit mode.
5078 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5080 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5082 * gengtype-state.c (read_state_pair): Likewise.
5083 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5084 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5085 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5086 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5087 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5088 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5089 * builtins.c (fold_builtin_3): Likewise.
5090 * graphite-scop-detection.c
5091 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5092 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5095 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5098 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5099 parameter. Return true if ignore_decl_rtl_set_p is true and other
5100 conditions are satisfied.
5101 * asan.h (asan_protect_global): Add new parameter.
5102 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5103 to asan_protect_global calls.
5105 2017-11-30 Jim Wilson <jimw@sifive.com>
5107 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5108 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5109 -mstrict-align, add info on default value. Delete redundant lines for
5110 -mabi. Add missing -mexplicit-relocs docs.
5112 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5114 * config/arc/arc.md (trap): New pattern.
5116 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5118 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5119 instruction to end into a delay slot.
5120 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5121 can be placed into a delay slot against reg_note.
5123 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5125 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5126 labels number of usages.
5128 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5130 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5132 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5134 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5137 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5138 multiplication by power of 2 constant into two shifts + comparison.
5140 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5143 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5145 2017-11-30 Richard Biener <rguenther@suse.de>
5147 PR tree-optimization/83202
5148 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5149 allow_peel argument and guard peeling.
5150 (canonicalize_loop_induction_variables): Likewise.
5151 (canonicalize_induction_variables): Pass false.
5152 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5153 peeling from cunrolli.
5155 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5157 * combine.c (try_combine): Print a message to dump file whenever
5158 I0, I1, or I2 cannot be combined into I3.
5160 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5162 PR rtl-optimization/83156
5163 PR rtl-optimization/82621
5164 * combine.c (try_combine): Don't split an I2 if one of the dests is
5165 set again before I3. Allow unused dests.
5167 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5169 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5171 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5173 PR rtl-optimization/80818
5174 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5175 recursively. Use insn code for clobber.
5176 (lra_set_insn_recog_data): Pass the new arg to
5177 collect_non_operand_hard_regs.
5178 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5180 (lra_update_insn_regno_info): Pass insn to
5181 add_regs_to_insn_regno_info.
5183 2017-11-29 Jim Wilson <jimw@sifive.com>
5184 Andrew Waterman <andrew@sifive.com>
5186 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5187 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5188 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5189 use SINGLE_SHIFT_COST.
5190 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5191 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5193 2017-11-29 Julia Koval <julia.koval@intel.com>
5195 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5196 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5197 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5198 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5199 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5200 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5201 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5202 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5203 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5204 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5205 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5206 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5207 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5208 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5209 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5210 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5211 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5212 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5213 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5214 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5215 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5216 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5217 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5218 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5219 __builtin_ia32_vpshldv_v2di_mask,
5220 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5221 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5222 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5224 2017-11-29 Julia Koval <julia.koval@intel.com>
5226 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5227 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5228 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5229 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5230 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5231 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5232 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5233 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5234 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5235 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5236 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5237 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5238 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5239 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5240 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5241 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5242 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5243 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5244 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5245 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5246 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5248 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5250 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5251 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5254 2017-11-29 Martin Aberg <maberg@gaisler.com>
5256 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5257 to prevent b2bst errata sequence.
5258 (sqrtdf2_fix): Likewise.
5260 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5262 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5263 (div_sqrt_insn_p): New function.
5264 (sparc_do_work_around_errata): Insert NOP instructions to
5265 prevent sequences that could trigger the TN-0013 errata for
5266 certain LEON3 processors.
5267 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5268 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5269 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5270 (in_branch_delay): Prevent div and sqrt in delay slot if
5272 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5274 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5276 * config/sparc/sparc.c (atomic_insn_p): New function.
5277 (sparc_do_work_around_errata): Insert NOP instructions to
5278 prevent sequences that could trigger the TN-0010 errata for
5280 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5281 instruction referable in atomic_insns_p.
5283 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5285 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5286 (atomic_compare_and_swap_leon3_1): Likewise.
5289 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5291 * config/sparc/sparc.c (fpop_insn_p): New function.
5292 (sparc_do_work_around_errata): Insert NOP instructions to
5293 prevent sequences that could trigger the TN-0012 errata for
5295 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5296 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5297 (in_branch_annul_delay): Prevent floating-point instructions
5298 in delay slot of annulled integer branch.
5300 2017-11-29 Richard Biener <rguenther@suse.de>
5302 PR tree-optimization/83202
5303 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5305 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5306 nodes vectorized to the same stmts multiple times.
5307 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5308 (vect_analyze_slp_instance): Adjust.
5309 (scalar_stmts_to_slp_tree_map_t): New typedef.
5310 (vect_schedule_slp_instance): Add a map recording the SLP node
5311 representing the vectorized stmts for a set of scalar stmts.
5312 Avoid code-generating redundancies.
5313 (vect_schedule_slp): Allocate map and pass it down.
5315 2017-11-29 Nathan Sidwell <nathan@acm.org>
5318 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5321 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5323 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5324 cases for printing LDP/STP memory addresses.
5325 (aarch64_print_address_internal): Renamed from
5326 aarch64_print_operand_address, added parameter, add Pmode check.
5327 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5328 (aarch64_print_operand_address): Indirect to
5329 aarch64_print_address_internal.
5330 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5331 'y' operand output specifier.
5333 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5336 * tree.c (build_simple_mem_ref_loc): Handle
5337 get_addr_base_and_unit_offset returning a MEM_REF.
5340 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5343 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5346 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5348 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5350 (layout::move_to_column): Likewise.
5352 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5355 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5356 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5358 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5359 Martin Jambor <mjambor@suse.cz>
5362 * tree.h (expr_type_first_operand_type_p): Declare
5363 * tree.c (expr_type_first_operand_type_p): New function.
5364 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5365 (ipa_value_from_jfunc): Adjust declaration.
5366 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5367 Use it as result type for arithmetics, unless it is NULL in which case
5368 be more conservative.
5369 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5370 ipa_get_jf_pass_through_result.
5371 (propagate_vals_across_pass_through): Likewise.
5372 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5373 is to propagate_vals_across_pass_through.
5374 (propagate_constants_across_call): Pass PARM_TYPE to
5375 propagate_scalar_across_jump_function.
5376 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5377 ipa_value_from_jfunc.
5378 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5379 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5380 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5381 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5382 target_type, pass it to ipa_value_from_jfunc.
5383 (update_indirect_edges_after_inlining): Pass parameter type to
5384 try_make_edge_direct_simple_call.
5386 2017-11-28 Jeff Law <law@redhat.com>
5388 * gimple-ssa-evrp-analyze.c
5389 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5390 refine ranges if scev_initialized_p returns true.
5391 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5393 2017-11-28 Julia Koval <julia.koval@intel.com>
5395 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5396 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5397 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5398 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5399 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5400 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5401 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5402 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5403 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5404 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5405 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5406 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5407 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5408 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5409 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5410 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5411 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5412 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5413 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5414 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5415 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5416 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5418 2017-11-28 Julia Koval <julia.koval@intel.com>
5420 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5421 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5422 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5423 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5424 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5425 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5426 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5427 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5428 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5429 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5430 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5431 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5432 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5433 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5434 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5435 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5436 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5437 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5438 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5439 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5440 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5441 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5442 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5443 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5444 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5445 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5446 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5447 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5448 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5449 __builtin_ia32_vpshld_v2di_mask): New builtins.
5450 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5451 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5453 2017-11-28 Richard Biener <rguenther@suse.de>
5455 PR tree-optimization/80776
5456 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5458 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5460 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5461 If the incoming edge is an effective fallthru because the other
5462 edge only reaches a __builtin_unreachable () then record ranges
5463 derived from the controlling condition in SSA info.
5464 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5465 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5467 2017-11-28 Olivier Hainque <hainque@adacore.com>
5469 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5470 on mingw build hosts.
5472 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5475 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5477 2017-11-28 Richard Biener <rguenther@suse.de>
5480 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5481 copies generated from memcpy use a character array as reference
5484 2017-11-28 Julia Koval <julia.koval@intel.com>
5485 Sebastian Peryt <sebastian.peryt@intel.com>
5487 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5488 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5489 cilk-common.o, cilk.h, cilk-common.c): Remove.
5491 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5492 * builtins.c (is_builtin_name): Remove cilkplus condition.
5493 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5494 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5495 cilk-builtins.def, cilkplus.def): Remove.
5496 * cif-code.def (CILK_SPAWN): Remove.
5497 * cilk-builtins.def: Delete.
5498 * cilk-common.c: Ditto.
5500 * cilkplus.def: Ditto.
5501 * config/darwin.h (fcilkplus): Delete.
5502 * cppbuiltin.c: Ditto.
5503 * doc/extend.texi: Remove cilkplus doc.
5504 * doc/generic.texi: Ditto.
5505 * doc/invoke.texi: Ditto.
5506 * doc/passes.texi: Ditto.
5507 * gcc.c (fcilkplus): Remove.
5508 * gengtype.c (cilk.h): Remove.
5509 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5511 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5513 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5515 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5516 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5517 cilkplus conditions.
5518 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5519 inline_read_section): Ditto.
5520 * ipa-inline-analysis.c (cilk.h): Remove.
5521 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5522 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5523 append_linker_options): Remove condition for fcilkplus.
5524 * lto/lto-lang.c (cilk.h): Remove.
5525 (lto_init): Remove condition for fcilkplus.
5526 * omp-expand.c (expand_cilk_for_call): Delete.
5527 (expand_omp_taskreg, expand_omp_for_static_chunk,
5528 expand_omp_for): Remove cilkplus
5530 (expand_cilk_for): Delete.
5531 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5532 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5533 execute_lower_omp, diagnose_sb_0): Ditto.
5534 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5535 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5536 * tree-nested.c: Ditto.
5537 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5538 (dump_generic_node): Ditto.
5539 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5540 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5542 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5544 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5546 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5547 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5548 for vector mode and !TARGET_SIMD.
5550 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5552 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5554 * tree.h (SWITCH_LABELS): Remove.
5555 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5556 assert SWITCH_BODY is non-NULL.
5557 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5559 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5561 PR tree-optimization/80788
5562 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5563 has TREE_OVERFLOW set, call drop_tree_overflow.
5565 2017-11-28 Richard Biener <rguenther@suse.de>
5567 PR tree-optimization/83158
5568 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5570 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5573 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5574 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5576 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5578 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5579 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5581 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5584 * varasm.c (compare_constant): Take the mode of the constants into
5585 account when comparing floating point constants.
5587 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5589 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5590 from explicit instantiation of debug_helper.
5591 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5593 2017-11-27 Richard Biener <rguenther@suse.de>
5595 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5598 2017-11-27 Richard Biener <rguenther@suse.de>
5600 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5601 (type_hash_canon): Also clear int_cst_hash_table entry for
5603 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5605 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5607 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5608 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5609 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5611 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5613 * hash-map.h (gt_cleare_cache): Avoid UB.
5615 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5617 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5618 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5619 (decide_unroll_runtime_iterations): Likewise.
5620 (decide_unroll_stupid): Likewise.
5622 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5625 * config/i386/i386.md: Add a loop with incssp.
5627 2017-11-27 Martin Jambor <mjambor@suse.cz>
5629 PR tree-optimization/81248
5630 * tree-sra.c (splice_param_accesses): Remove size check.
5631 (decide_one_param_reduction): Fix size check.
5632 * gimple-pretty-print.c (dump_profile): Silence warning.
5633 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5635 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5638 * dbxout.c (lastlineno): New variable.
5639 (dbx_debug_hooks): Use dbxout_switch_text_section as
5640 switch_text_section debug hook.
5641 (dbxout_function_end): Switch to current_function_section
5642 rather than function_section. If crtl->has_bb_partition,
5643 output just one N_FUN, depending on in_cold_section_p.
5644 (dbxout_source_line): Remember last lineno in lastlineno.
5645 (dbxout_switch_text_section): New function.
5646 (dbxout_function_decl): Adjust dbxout_block caller.
5647 (dbx_block_with_cold_children): New function.
5648 (dbxout_block): Return true if any LBRAC/RBRAC have been
5649 emitted. Use dbx_block_with_cold_children at depth == 0
5650 in second partition. Add PARENT_BLOCKNUM argument, pass
5651 it optionally adjusted to children. Output LBRAC/RBRAC
5652 around recursive call only if the block is in the current
5653 partition, if not and anything was output, emit empty
5655 * final.c (final_scan_insn): Compute cold_function_name
5656 before calling switch_text_section debug hook. Call
5657 that hook even if dwarf2out_do_frame if not emitting
5661 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5662 TREE_READONLY decls.
5664 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5666 PR rtl-optimization/82488
5667 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5669 2017-11-26 Julia Koval <julia.koval@intel.com>
5671 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5673 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5676 2017-11-26 Julia Koval <julia.koval@intel.com>
5678 * config/i386/driver-i386.c (host_detect_local_cpu):
5679 Detect skylake-avx512.
5681 2017-11-26 Julia Koval <julia.koval@intel.com>
5683 * config.gcc: Add -march=cannonlake.
5684 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5685 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5686 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5687 (PTA_CANNONLAKE): New.
5688 (processor_target_table): Add cannonlake.
5689 (ix86_option_override_internal): Ditto.
5690 (fold_builtin_cpu): Ditto.
5691 (get_builtin_code_for_version): Handle cannonlake.
5692 (M_INTEL_COREI7_CANNONLAKE): New.
5693 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5694 * doc/invoke.texi: Add -march=cannonlake.
5696 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5698 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5699 (try_init_one_plugin): Alternative implementation for MinGW.
5700 * Makefile.in (plugin_implib): New.
5701 (gengtype-lex.c): Fix broken AIX workaround.
5702 * configure: Regenerate.
5703 * doc/plugins.texi: Document support for MinGW.
5705 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5707 PR rtl-optimization/81553
5708 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5709 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5710 is a shift where C1 has different mode than the whole shift, use C1's
5711 mode for MULT rather than the shift's mode.
5714 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5715 builtins not enabled in the currently selected ISA.
5717 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5719 PR tree-optimization/71026
5720 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5721 (insert_reciprocals): Change to insert reciprocals before a division
5722 by a square and to insert the square of a reciprocal.
5723 (execute_cse_reciprocals_1): Change to consider division by a square.
5724 (register_division_in): Add importance parameter.
5726 2017-11-24 Richard Biener <rguenther@suse.de>
5728 PR tree-optimization/82402
5729 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5730 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5732 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5734 * match.pd (0-ptr): New transformation.
5736 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5739 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5740 when skipping checking.
5742 2017-11-24 Richard Biener <rguenther@suse.de>
5744 PR tree-optimization/83128
5745 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5746 (vn_reference_lookup_3): Likewise.
5748 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5751 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5752 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5755 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5756 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5757 check for the latter with gimple_call_builtin_p. Do not handle
5758 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5760 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5762 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5765 2017-11-23 Julia Koval <julia.koval@intel.com>
5767 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5768 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5769 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5770 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5771 _mm512_maskz_expandloadu_epi16): New intrinsics.
5772 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5773 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5774 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5775 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5776 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5777 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5778 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5779 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5780 _mm256_maskz_expandloadu_epi8): New intrinsics.
5781 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5782 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5783 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5784 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5785 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5786 * config/i386/sse.md (VI248_VLBW): New iterator.
5787 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5789 2017-11-23 Julia Koval <julia.koval@intel.com>
5791 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5792 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5793 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5794 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5795 _mm512_mask_compressstoreu_epi16): New.
5796 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5797 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5798 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5799 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5800 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5801 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5802 _mm256_mask_compressstoreu_epi8): New.
5803 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5804 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5805 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5806 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5807 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5808 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5809 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5810 __builtin_ia32_compresshi128_mask,
5811 __builtin_ia32_compressstoreuqi512_mask,
5812 __builtin_ia32_compressstoreuhi512_mask,
5813 __builtin_ia32_compressstoreuqi256_mask,
5814 __builtin_ia32_compressstoreuqi128_mask,
5815 __builtin_ia32_compressstoreuhi256_mask,
5816 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5817 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5819 (ix86_expand_special_args_builtin): Handle new types.
5820 (s4fma_expand): Handle new builtin array.
5821 * config/i386/immintrin.h: Include new headers.
5822 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5823 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5825 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5827 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5828 fixes. Declare temp and g variables at the top in order to avoid
5829 {} in most of the cases.
5831 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5833 * match.pd (ptr-0): New transformation.
5835 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5837 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5838 (struct addr_mode_cost_table): New.
5839 (struct tune_params): Add field addr_mode_costs.
5840 * config/arm/arm.c (generic_addr_mode_costs): New.
5841 (arm_slowmul_tune): Initialise addr_mode_costs field.
5842 (arm_fastmul_tune): Likewise.
5843 (arm_strongarm_tune): Likewise.
5844 (arm_xscale_tune): Likewise.
5845 (arm_9e_tune): Likewise.
5846 (arm_marvell_pj4_tune): Likewise.
5847 (arm_v6t2_tune): Likewise.
5848 (arm_cortex_tune): Likewise.
5849 (arm_cortex_a8_tune): Likewise.
5850 (arm_cortex_a7_tune): Likewise.
5851 (arm_cortex_a15_tune): Likewise.
5852 (arm_cortex_a35_tune): Likewise.
5853 (arm_cortex_a53_tune): Likewise.
5854 (arm_cortex_a57_tune): Likewise.
5855 (arm_exynosm1_tune): Likewise.
5856 (arm_xgene1_tune): Likewise.
5857 (arm_cortex_a5_tune): Likewise.
5858 (arm_cortex_a9_tune): Likewise.
5859 (arm_cortex_a12_tune): Likewise.
5860 (arm_cortex_a73_tune): Likewise.
5861 (arm_v7m_tune): Likewise.
5862 (arm_cortex_m7_tune): Likewise.
5863 (arm_v6m_tune): Likewise.
5864 (arm_fa726te_tune): Likewise.
5865 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5867 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5869 * config/arm/arm.c (arm_mem_costs): New function.
5870 (arm_rtx_costs_internal): Use arm_mem_costs.
5872 2017-11-23 Mark Wielaard <mark@klomp.org>
5874 * dwarf2out.c (init_sections_and_labels): Use generation to create
5875 unique ranges_section_label and ranges_base_label. Return generation.
5876 (output_rnglists): Add generation argument. Use generation to create
5877 unique ranges labels.
5878 (dwarf2out_finish): Get generation from init_sections_and_labels
5879 and pass generation to output_rnglists.
5881 2017-11-23 Mike Stump <mikestump@comcast.net>
5882 Eric Botcazou <ebotcazou@adacore.com>
5884 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5885 * cfgloop.h (struct loop): Add unroll field.
5886 * function.h (struct function): Add has_unroll bitfield.
5887 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5888 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5889 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5890 (pass_rtl_unroll_loops::gate): Likewise.
5891 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5892 for which loop->unroll==1.
5893 (decide_unroll_constant_iterations): Use note for consistency and
5894 take loop->unroll into account. Return early if loop->unroll is set.
5895 Fix thinko in existing test.
5896 (decide_unroll_runtime_iterations): Use note for consistency and
5897 take loop->unroll into account.
5898 (decide_unroll_stupid): Likewise.
5899 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5900 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5901 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5903 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5904 (print_loop): Print loop->unroll if set.
5905 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5906 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5907 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5909 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5910 loop->unroll is set and smaller than the trip count. Otherwise bypass
5911 entirely the heuristics if loop->unroll is set. Remove dead note.
5912 Fix off-by-one bug in other note.
5913 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5914 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5916 (tree_unroll_loops_completely): Make static.
5917 (pass_complete_unroll::execute): Use correct type for variable.
5918 (pass_complete_unrolli::execute): Fix formatting.
5919 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5921 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5923 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5924 enable when TARGET_PREFER_AVX128 is set.
5926 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5928 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5930 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5931 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5933 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5935 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5936 frequencies. Fix estimation of aggregate parameters.
5938 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5940 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5943 2017-11-23 Tom de Vries <tom@codesourcery.com>
5945 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5947 2017-11-23 Tom de Vries <tom@codesourcery.com>
5949 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5952 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5955 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5956 sibcall_value_pcrel_fdpic): Use local variable instead of
5958 (calli_tbr_rel): Add missing operand 2.
5959 (call_valuei_tbr_rel): Add missing operand 3.
5961 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5964 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5965 bitpos/bitsize covering the whole destination, use store_expr only if
5966 the complex mode is the same. Otherwise, use expand_normal and if
5967 it returns CONCAT, subreg each part separately instead of trying to
5968 subreg the whole result.
5970 2017-11-23 Richard Biener <rguenther@suse.de>
5972 PR tree-optimization/23094
5973 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5974 come here from walking over backedges in the first iteration.
5975 (vn_reference_lookup_3): Skip clobbers that store the same value.
5977 2017-11-23 Richard Biener <rguenther@suse.de>
5979 PR tree-optimization/81403
5980 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5981 a block we need a leader relative to.
5982 (phi_translate_1): For nary processing require a leader from
5983 get_representative_for given we run expression simplification
5984 using match-and-simplify. Remove previous fix.
5986 2017-11-22 Jeff Law <law@redhat.com>
5988 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5989 Use new method allocate_value_range rather than accessing the
5990 vrp_value_range_pool data member directly.
5991 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5992 to use extract_range_from_stmt method to avoid need for
5993 extract_range_from_assignment method.
5994 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5995 method rather than setting values_propgated data member directly.
5996 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5997 and values propagated data members and extract_range_from_assignment
5998 method. Reorder private data members to conform to standards.
5999 Add new methods set_lattice_propagation_complete and
6000 allocate_value_range.
6002 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6004 PR rtl-optimization/83030
6005 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6006 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6007 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6008 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6009 CROSSING_JUMP_P flag is already set before setting it.
6010 * cfgrtl.c (fixup_partition_crossing): Likewise.
6011 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6014 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6016 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6017 Handle the case where both arguments are using gen_const_vec_series.
6019 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6022 * pretty-print.c (pp_format): Move quoting implementation to
6023 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6024 to pass address of "quote" local.
6025 (pp_begin_quote): New function.
6026 (pp_end_quote): New function.
6027 * pretty-print.h (printer_fn): Convert penultimate param from bool
6029 (pp_begin_quote): New decl.
6030 (pp_end_quote): New decl.
6031 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6032 param from bool to bool *.
6033 * tree-diagnostic.h (default_tree_printer): Likewise.
6035 2017-11-22 Jeff Law <law@redhat.com>
6037 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6038 Initialize vr_values.
6039 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6040 vr_values via vr_values class instance rather than delegators.
6041 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6042 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6043 (evrp_range_analyzer::push_value_range): Likewise.
6044 (evrp_range_analyzer::pop_value_range): Likewise.
6045 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6046 most delegators. Those remaining are exposed as public interfaces.
6047 Make vr_values a pointer and private.
6048 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6050 (evrp_range_analyzer::get_vr_value): New method.
6051 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6052 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6053 (class evrp_dom_walker): Attach evrp_folder class, initialize
6054 it in the ctor. Remove temporary delegators.
6055 (evrp_dom_walker::before_dom_children): Call methods in attached
6056 evrp_range_analyzer class via class instance pointer. Use
6057 free value_range_constant_singleton to remove need for
6058 op_with_constant_singleton_value delegator method. Do not
6059 create a vrp_prop class instance for every call! Narrow
6060 scope of a couple variables.
6061 (evrp_dom_walker::cleanup): Call methods in attached
6062 evrp_range_analyzer class via class instance pointer.
6063 * vr-values.h (class vr_values): Privatize many methods and
6066 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6068 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6069 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6070 * expr.c (expand_expr_real_2): Likewise.
6071 * fold-const.c (const_unop): Likewise.
6072 * optabs-tree.c (optab_for_tree_code): Likewise.
6073 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6074 * tree-inline.c (estimate_operator_cost): Likewise.
6075 * tree-pretty-print.c (dump_generic_node): Likewise.
6076 (op_code_prio): Likewise.
6077 (op_symbol_code): Likewise.
6078 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6079 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6080 * internal-fn.c (direct_internal_fn_optab): New function.
6081 (direct_internal_fn_array, direct_internal_fn_supported_p
6082 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6083 * fold-const-call.c (fold_const_reduction): New function.
6084 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6086 * tree-vect-loop.c: Include internal-fn.h.
6087 (reduction_code_for_scalar_code): Rename to...
6088 (reduction_fn_for_scalar_code): ...this and return an internal
6090 (vect_model_reduction_cost): Take an internal_fn rather than
6092 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6094 (vectorizable_reduction): Use internal functions rather than tree
6095 codes for the reduction operation. Update calls to the functions
6097 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6098 Use calls to internal functions rather than REDUC tree codes.
6099 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6101 2017-11-22 Olivier Hainque <hainque@adacore.com>
6103 * config/vxworks.c (vxworks_override_options): Pick default
6104 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6105 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6106 DWARF_GNAT_ENCODINGS_DEFAULT.
6107 * config/vxworksae.h: Likewise.
6109 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6111 PR tree-optimization/83104
6112 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6115 2017-11-22 Marek Polacek <polacek@redhat.com>
6116 H.J. Lu <hongjiu.lu@intel.com>
6117 Jason Merrill <jason@redhat.com>
6122 * calls.c (initialize_argument_information): Call
6123 warn_parameter_passing_abi target hook.
6124 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6125 argument onto stack.
6126 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6127 * common.opt: Update -fabi-version description.
6128 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6129 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6131 (ix86_is_empty_record): New function.
6132 (ix86_warn_parameter_passing_abi): New function.
6133 (TARGET_EMPTY_RECORD_P): Redefine.
6134 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6135 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6136 * doc/tm.texi: Regenerated.
6137 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6138 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6139 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6140 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6141 instead of int_size_in_bytes.
6142 * expr.c (copy_blkmode_to_reg): Likewise.
6143 * function.c (aggregate_value_p): Return 0 for empty types.
6144 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6146 (locate_and_pad_parm): Call arg size_in_bytes instead
6148 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6149 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6150 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6152 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6153 (std_gimplify_va_arg_expr): Skip empty records. Call
6154 arg_size_in_bytes instead size_in_bytes.
6155 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6156 * tree-core.h (tree_type_common): Add empty_flag.
6157 (tree_decl_common): Update comments.
6158 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6160 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6161 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6163 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6164 * tree.c (default_is_empty_type): New function.
6165 (default_is_empty_record): New function.
6166 (arg_int_size_in_bytes): New function.
6167 (arg_size_in_bytes): New function.
6168 (get_ultimate_context): New function.
6169 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6170 TRANSLATION_UNIT_WARN_EMPTY_P.
6171 (default_is_empty_record, arg_int_size_in_bytes,
6172 arg_size_in_bytes, get_ultimate_context): Declare.
6174 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6176 * config/arm/arm.c (cmse_clear_registers): New function.
6177 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6178 code by call to cmse_clear_registers.
6179 (cmse_nonsecure_entry_clear_before_return): Likewise.
6181 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6183 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6184 (vdot_s32, vdotq_s32): New.
6185 (vdot_lane_u32, vdotq_lane_u32): New.
6186 (vdot_lane_s32, vdotq_lane_s32): New.
6189 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6192 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6193 for unsigned values with fewer HWIs than the precision.
6194 (test_overflow): New function.
6195 (wide_int_cc_tests): Call it.
6197 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6199 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6200 has the same mode class as Pmode.
6202 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6204 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6205 padding_bits_to_clear_ptr.
6206 (cmse_nonsecure_entry_clear_before_return): Likewise.
6208 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6210 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6211 auto_sbitap instead of integer bitfield to control register needing
6214 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6216 PR tree-optimization/83044
6217 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6218 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6219 that need the upper bound. Subtract offset from
6220 get_addr_base_and_unit_offset only if positive and subtract it
6221 before division by eltsize rather than after it.
6224 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6225 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6226 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6230 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6231 expand_binop with *mul_widen_optab, make sure at least one of the
6232 operands doesn't have VOIDmode.
6235 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6237 PR rtl-optimization/82044
6238 PR tree-optimization/82042
6239 * dse.c (record_store): Check for overflow.
6240 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6241 add_wild_read instead of clear_rhs_from_active_local_stores on
6242 overflow and log it into dump_file.
6244 2017-11-22 Richard Biener <rguenther@suse.de>
6246 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6247 fake edges to exit when looking for a place to insert.
6248 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6250 (insert_into_preds_of_block): Commit edge insertion immediately,
6251 assert that doesn't require new BBs.
6252 (fini_pre): Release expressions.
6253 (pass_pre::execute): Shuffle things around a bit, if the fn
6254 is too large do not compute AVAIL either as this is really the
6257 2017-11-22 Richard Biener <rguenther@suse.de>
6259 PR tree-optimization/83089
6260 * tree-if-conv.c (pass_if_conversion::execute): If anything
6261 changed reset SCEV and free the number of iteration estimates.
6263 2017-11-21 Martin Sebor <msebor@redhat.com>
6265 PR tree-optimization/82945
6266 * calls.h (warn_nonstring_bound): Remove unused function.
6268 2017-11-21 Martin Sebor <msebor@redhat.com>
6270 PR tree-optimization/82945
6271 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6272 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6273 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6275 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6276 * calls.h (get_attr_nonstring_decl): Declare new function.
6277 * doc/extend.texi (attribute nonstring): Update.
6278 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6279 get_attr_nonstring_decl and handle it.
6280 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6281 detection of nul-termination.
6282 (strlen_to_stridx): Change to a pointer.
6283 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6284 (pass_strlen::execute): Same.
6286 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6288 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6289 for the new option -mprefer-vector-width=[none|128|256|512].
6290 * config/i386/i386.c (ix86_target_string): remove old style options
6291 -mprefer-avx256 and make -mprefer-avx128 as alias.
6292 (ix86_option_override_internal): Apply defaults for the
6293 -mprefer-vector-width=[128|256] option.
6294 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6295 Implement macros to work with -mprefer-vector-width=.
6296 * config/i386/i386.opt: Implemented option
6297 -mprefer-vector-width=[none|128|256|512].
6298 * doc/invoke.texi: Documentation for
6299 -mprefer-vector-width=[none|128|256|512].
6301 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6303 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6305 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6307 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6309 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6311 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6313 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6314 * expr.c (expand_expr_real_2): Likewise.
6315 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6316 fold_binary_loc): Likewise.
6317 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6318 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6319 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6320 MINUS_EXPR transformations.
6321 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6322 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6323 * tree-inline.c (estimate_operator_cost): Likewise.
6324 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6325 op_symbol_code): Likewise.
6326 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6327 * vr-values.c (extract_range_from_binary_expr): Likewise.
6328 * varasm.c (initializer_constant_valid_p_1): Likewise.
6329 * tree.def: New tree code POINTER_DIFF_EXPR.
6331 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6333 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6334 integer suffix to movbe mnemonic.
6335 (*bswaphi2_movbe): Ditto.
6336 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6338 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6341 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6342 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6343 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6345 2017-11-21 Martin Liska <mliska@suse.cz>
6347 * tree-inline.c (expand_call_inline): Remove not needed
6350 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6351 Jakub Jelinek <jakub@redhat.com>
6354 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6355 Remove static keyword from f variable.
6357 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6359 PR tree-optimization/83086
6360 * gimple-ssa-store-merging.c
6361 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6362 rather than n.base_addr.
6364 2017-11-21 Martin Liska <mliska@suse.cz>
6366 PR rtl-optimization/82044
6367 PR tree-optimization/82042
6368 * dse.c (check_mem_read_rtx): Check for overflow.
6370 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6372 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6375 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6377 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6378 and its corresponding call arg location note.
6380 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6381 Andrew Burgess <andrew.burgess@embecosm.com>
6383 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6385 (arc_return_slot_offset): Likewise.
6386 (arc_eh_return_address_location): New declaration.
6387 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6388 (MUST_SAVE_REGISTER): Add exception handler case.
6389 (MUST_SAVE_RETURN_ADDR): Likewise.
6390 (arc_frame_pointer_required): Likewise.
6391 (arc_frame_pointer_needed): New function.
6392 (arc_compute_frame_size): Changed.
6393 (arc_expand_prologue): Likewise.
6394 (arc_expand_epilogue): Likewise.
6395 (arc_initial_elimination_offset): Likewise.
6396 (arc_return_slot_offset): Delete.
6397 (arc_eh_return_address_location): New function.
6398 (arc_builtin_setjmp_frame_value): Likewise.
6399 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6400 (EH_RETURN_STACKADJ_RTX): Define.
6401 (EH_RETURN_HANDLER_RTX): Likewise.
6402 * config/arc/arc.md (eh_return): Delete.
6404 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6406 * print-tree.h (debug_vec_tree): Remove prototype.
6407 * gdbinit.in (pvt): Remove macro.
6409 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6411 PR tree-optimization/83047
6412 * gimple-ssa-store-merging.c
6413 (imm_store_chain_info::output_merged_store): If the loads with the
6414 same vuse are in different basic blocks, for load_gsi pick a load
6415 location that is dominated by the other loads.
6418 * config/i386/i386.c (ix86_memmodel_check): Start
6419 -Winvalid-memory-model diagnostics with lowercase letter.
6422 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6423 set in_cold_section_p to the partition containing loc_list->first.
6424 When seeing loc_list->last_before_switch node, update secname and
6425 perform range_across_switch second partition handling only after that.
6428 * run-rtl-passes.c: Include debug.h.
6429 (run_rtl_passes): Call debug_hooks->assembly_start.
6430 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6434 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6435 OPTAB_DIRECT in calls to expand_simple_binop.
6437 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6440 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6442 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6445 * spellcheck.h (best_match::blithely_get_best_candidate): New
6448 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6450 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6451 with lower case letter.
6453 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6455 * config/i386/i386.md (bswaphi2): New expander.
6456 (*bswaphi2_movbe): New insn pattern.
6457 (bswaphi -> rorhi pepehole2): New peephole pattern.
6459 2017-11-20 Jeff Law <law@redhat.com>
6461 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6462 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6463 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6464 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6465 gimple-ssa-evrp-analyze.h.
6467 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6468 set BB_VISITED here.
6469 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6471 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6473 * tree-predcom.c: Add general comment on Store-Store chains.
6474 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6476 (get_chain_last_ref_at): Rename into...
6477 (get_chain_last_write_at): ...this.
6478 (get_chain_last_write_before_load): New function.
6479 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6480 CT_STORE_STORE when write reference is added.
6481 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6482 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6483 (initialize_root_vars_store_elim_1): Ditto.
6484 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6485 definition is created.
6486 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6487 chain by replacing it with dominant stored value.
6489 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6491 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6493 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6495 * vr-values.c (extract_range_from_binary_expr): Use a full range
6498 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6500 * config/arm/arm.md (R4_REGNUM): Define constant.
6501 (nonsecure_call_internal): Remove r4 clobber.
6502 (nonsecure_call_value_internal): Likewise.
6503 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6504 clobber and resequence match_operands.
6505 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6506 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6507 (nonsecure_call_value_reg_thumb2): Likewise.
6509 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6511 PR tree-optimization/78821
6512 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6513 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6514 to the address of the base rather than the base itself.
6515 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6516 (find_bswap_or_nop_finalize): New function.
6517 (find_bswap_or_nop): Use it.
6518 (bswap_replace): Return a tree rather than bool, change first
6519 argument from gimple * to gimple_stmt_iterator, allow inserting
6520 into an empty sequence, allow ins_stmt to be NULL - then emit
6521 all stmts into gsi. Fix up MEM_REF address gimplification.
6522 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6523 (struct store_immediate_info): Add N and INS_STMT non-static
6525 (store_immediate_info::store_immediate_info): Initialize them
6526 from newly added ctor args.
6527 (merged_store_group::apply_stores): Formatting fixes. Sort by
6529 (stmts_may_clobber_ref_p): For stores call also
6530 refs_anti_dependent_p.
6531 (gather_bswap_load_refs): New function.
6532 (imm_store_chain_info::try_coalesce_bswap): New method.
6533 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6534 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6535 (imm_store_chain_info::output_merged_store): Fail if number of
6536 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6537 and NOP_EXPR rhs_code.
6538 (pass_store_merging::process_store): Compute n and ins_stmt, if
6539 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6540 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6542 (pass_store_merging::execute): Calculate dominators.
6544 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6545 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6546 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6547 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6548 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6549 class pass_optimize_bswap, bswap_replace,
6550 pass_optimize_bswap::execute): Moved to ...
6551 * gimple-ssa-store-merging.c: ... this file.
6552 Include optabs-tree.h.
6553 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6554 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6555 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6556 anonymous namespace, remove static keywords.
6557 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6558 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6560 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6563 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6565 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6567 * vec.h (debug_helper): New function.
6568 (DEFINE_DEBUG_VEC): New macro.
6569 * hash-set.h (debug_helper): New function.
6570 (DEFINE_DEBUG_HASH_SET): New macro.
6571 * cfg.c (debug_slim (edge)): New function.
6572 Call DEFINE_DEBUG_VEC for edges.
6573 Call DEFINE_DEBUG_HASH_SET for edges.
6574 * cfghooks.c (debug_slim (basic_block)): New function.
6575 Call DEFINE_DEBUG_VEC for basic blocks.
6576 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6577 * print-tree.c (debug_slim): New function to handle trees.
6578 Call DEFINE_DEBUG_VEC for trees.
6579 Call DEFINE_DEBUG_HASH_SET for trees.
6580 (debug (vec<tree, va_gc>) &): Remove.
6581 (debug (<vec<tree, va_gc>) *): Remove.
6582 * print-rtl.c (debug_slim): New function to handle const_rtx.
6583 Call DEFINE_DEBUG_VEC for rtx_def.
6584 Call DEFINE_DEBUG_VEC for rtx_insn.
6585 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6586 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6587 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6588 (debug (vec<rtx_insn *> *ptr): Remove.
6589 (debug_insn_vector): Remove.
6590 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6592 2017-11-20 Tom de Vries <tom@codesourcery.com>
6594 PR rtl-optimization/82020
6595 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6596 IF_THEN_ELSE condition.
6598 2017-11-19 Jeff Law <law@redhat.com>
6600 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6601 of degenerates resulting from ignoring an edge.
6603 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6606 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6608 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6611 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6612 for uninitialied counts.
6614 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6617 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6619 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6622 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6624 2017-11-19 Tom de Vries <tom@codesourcery.com>
6626 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6629 2017-11-19 Tom de Vries <tom@codesourcery.com>
6631 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6634 2017-11-19 Tom de Vries <tom@codesourcery.com>
6636 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6639 2017-11-19 Tom de Vries <tom@codesourcery.com>
6641 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6643 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6645 2017-11-19 Tom de Vries <tom@codesourcery.com>
6647 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6649 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6650 semicolon after MCORE_EXPORT_NAME call.
6652 2017-11-19 Tom de Vries <tom@codesourcery.com>
6655 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6656 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6658 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6660 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6661 (cgraph_edge::clone): Cleanup updating of profile.
6662 * ipa-cp.c (update_profiling_info): Likewise.
6663 * ipa-inline-transform.c (inline_transform): Likewise.
6664 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6665 * ipa-split.c (execute_split_functions): Do not split when function
6667 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6668 * profile-count.c (profile_count::dump): Add global0.
6669 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6671 (profile_count::to_sreal_scale): Likewise.
6672 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6673 (profile_count::combine_with_ipa_count): New function.
6674 * profile-count.h (profile_guessed_global0adjusted): New.
6675 (profile_count::adjusted_zero): New.
6676 (profile_count::global0adjusted): New.
6677 (profile_count::combine_with_ipa_count): New.
6678 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6679 correct profile of return block of split functions.
6680 (copy_cfg_body): Remove unused profile_count.
6681 (copy_body): Likewise.
6682 (expand_call_inline): Update.
6683 (tree_function_versioning): Update.
6685 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6687 * hash-set.h (hash_set::empty): New.
6688 * tree-ssa-threadbackward.h: Delete.
6689 * tree-ssa-threadbackward.c (class thread_jumps): New.
6690 Move max_threaded_paths into class.
6691 (fsm_find_thread_path): Remove arguments that are now in class.
6692 (profitable_jump_thread_path): Rename to...
6693 (thread_jumps::profitable_jump_thread_path): ...this.
6694 (convert_and_register_jump_thread_path): Rename to...
6695 (thread_jumps::convert_and_register_current_path): ...this.
6696 (check_subpath_and_update_thread_path): Rename to...
6697 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6698 (register_jump_thread_path_if_profitable): Rename to...
6699 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6700 (handle_phi): Rename to...
6701 (thread_jumps::handle_phi): ...this.
6702 (handle_assignment): Rename to...
6703 (thread_jumps::handle_assignment): ...this.
6704 (fsm_find_control_statement_thread_paths): Rename to...
6705 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6706 (find_jump_threads_backwards): Rename to...
6707 (thread_jumps::find_jump_threads_backwards): ...this.
6708 Initialize path local data.
6709 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6710 from within thread_jumps class.
6711 (pass_early_thread_jumps::execute): Same.
6713 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6715 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6717 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6719 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6720 to canonical location.
6722 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6725 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6727 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6729 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6731 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6732 rcrt1.o%s/grcrt1.o%s for -static-pie.
6734 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6736 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6737 ix86_shift_rotate_cost): Break out from ...
6738 (ix86_rtx_costs): ... here.
6739 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6742 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6744 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6745 when entry block was promoted unlikely.
6746 (estimate_bb_frequencies): Increase frequency scale.
6747 * profile-count.h (profile_count): Export precision info.
6749 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6751 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6752 disturbin profile of entry block.
6754 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6756 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6759 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6761 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6762 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6764 2017-11-17 Jeff Law <law@redhat.com>
6766 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6767 from evrp_dom_walker class. Various methods moved into new class.
6768 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6769 (evrp_range_analyzer::enter): New method.
6770 (evrp_range_analyzer::leave): New method.
6771 (evrp_dom_walker): Remove delegators no longer needed by this class.
6772 Replace vr_values data member with evrp_range_analyzer
6774 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6775 method extracted from evrp_dom_walker::before_dom_children.
6776 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6777 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6779 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6780 Add private copy constructor and move assignment operators.
6781 Privatize methods and class data where trivially possible.
6782 (evrp_dom_walker::cleanup): New function, extracted from
6783 execute_early_vrp. Simplify access to class data.
6785 * vr-values.h (get_output_for_vrp): Prototype.
6786 * vr-values.c (get_output_for_vrp): New function extracted from
6787 vrp_visit_assignment_or_call and extract_range_from_stmt.
6788 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6790 2017-11-17 Luis Machado <luis.machado@linaro.org>
6792 * config/aarch64/aarch64.c
6793 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6794 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6795 tune_params::AUTOPREFETCHER_WEAK.
6797 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6800 * config/arm/arm.c (arm_valid_target_attribute_rec):
6801 Parse "arch=" and "+<ext>".
6802 (arm_valid_target_attribute_tree): Re-init global options.
6803 (arm_option_override): Make non-static.
6804 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6805 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6806 (__ARM_FEATURE_CRC32): Support undef.
6807 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6808 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6810 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6812 * gdbinit.in (break-on-diagnostic): New command.
6814 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6816 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6817 return to indirect jump for EH return if control-flow
6818 protection is enabled. Change explicit 'false' argument in
6819 pro_epilogue_adjust_stack with a value of
6821 * config/i386/i386.md (simple_return_indirect_internal):
6822 Remove SImode restriction to support 64-bit.
6824 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6826 * combine.c (added_notes_insn): New.
6827 (try_combine): Handle added_notes_insn like added_links_insn.
6828 Rewrite return value code.
6829 (distribute_notes): Set added_notes_insn to the earliest insn we added
6832 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6834 PR rtl-optimization/82621
6835 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6836 dest of one of those SETs is unused.
6838 2017-11-17 Richard Biener <rguenther@suse.de>
6841 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6842 * tree-pretty-print.c (dump_generic_node): Handle
6843 annot_expr_parallel_kind.
6844 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6845 * gimplify.c (gimple_boolify): Likewise.
6847 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6849 * config.gcc (extra_headers): Add cet.h for x86 targets.
6850 * config/i386/cet.h: New file.
6851 * doc/install.texi: Add --enable-cet/--disable-cet.
6853 2017-11-17 Richard Biener <rguenther@suse.de>
6855 PR tree-optimization/83017
6856 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6857 (gen_parallel_loop): Properly count iterations.
6858 (parallelize_loops): Handle loop->can_be_parallel independent
6859 of flag_loop_parallelize_all. Make static profitability test match
6861 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6862 * invoke.texi (parloops-min-per-thread): Document.
6864 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6866 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6867 upstreaming review comments.
6869 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6871 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6872 with fast unaligned access.
6873 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6875 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6877 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6878 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6880 2017-11-17 Richard Biener <rguenther@suse.de>
6882 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6883 folding of references.
6885 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6888 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6889 of replacing call to strncmp with corresponding call to strcmp when
6892 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6894 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6895 option prefer-avx256 for skylake-avx512 configuration.
6896 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6897 (get_builtin_code_for_version): Ditto.
6899 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6900 Monk Chiang <sh.chiang04@gmail.com>
6902 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6903 (FIXED_REGISTERS): Reserve more register numbers.
6904 (CALL_USED_REGISTERS): Likewise.
6905 (REG_ALLOC_ORDER): Likewise.
6906 (REG_CLASS_CONTENTS): Likewise.
6907 (REGISTER_NAMES): Likewise.
6909 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6910 Kito Cheng <kito.cheng@gmail.com>
6912 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6914 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6916 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6918 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6919 parameter list for vec_splats.
6921 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6923 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6925 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6927 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6930 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6931 with the required version of automake 1.11.6.
6933 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6935 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6937 (DU_C2_3_power9): Correct reservation combinations.
6938 (FP_DIV_power9, VEC_DIV_power9): New.
6939 (power9-alu): Split out rotate/shift...
6940 (power9-rot): ...to here, correct dispatch resource.
6941 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6943 (power9-fp): Correct latency.
6944 (power9-sdiv): Add div/sqrt resource.
6945 (power9-ddiv): Correct latency, add div/sqrt resource.
6946 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6947 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6949 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6951 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6953 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6954 switch statement mapping KF built-ins to TF built-ins if we don't
6955 have the proper ISA 3.0 assembler support.
6957 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6959 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6960 (lower_emutls_function_body): Do not compute it.
6962 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6964 * ipa-split.c (split_bb_info): Turn time to sreal.
6965 (split_point): Likewise.
6966 (dump_split_point): Likewise.
6967 (fine_split_points): Likewise.
6968 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6970 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6972 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6975 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6977 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6979 (expensive_function_p): Remove useless assert.
6980 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6982 2017-11-16 Martin Sebor <msebor@redhat.com>
6984 PR tree-optimization/82588
6985 PR tree-optimization/82583
6986 * tree-vrp.c (check_array_ref): Handle flexible array members,
6987 string literals, and inner indices.
6988 (search_for_addr_array): Add detail to diagnostics.
6990 2017-11-16 Nathan Sidwell <nathan@acm.org>
6994 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6995 (SET_DECL_ASSEMBLER_NAME): Forward to
6996 overwrite_decl_assembler_name.
6997 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6998 (overwrite_decl_assembler_name): Declare.
6999 * tree.c (overwrite_decl_assembler_name): New.
7000 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7001 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7002 (LANG_HOOKS_INITIALIZER): Add it.
7003 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7004 * langhooks.c (lhd_set_decl_assembler_name): Use
7005 SET_DECL_ASSEMBLER_NAME.
7006 (lhd_overwrite_decl_assembler_name): Default implementation.
7008 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7009 Jackson Woodruff <jackson.woodruff@arm.com>
7011 PR tree-optimization/71026
7012 * match.pd: Canonicalize constant multiplies in division.
7014 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7016 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7017 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7018 Remove OPT_fomit_frame_pointer entry.
7019 * common/config/arc/arc-common.c: Likewise.
7020 * common/config/arm/arm-common.c: Likewise.
7021 * common/config/avr/avr-common.c: Likewise.
7022 * common/config/c6x/c6x-common.c: Likewise.
7023 * common/config/cr16/cr16-common.c: Likewise.
7024 * common/config/cris/cris-common.c: Likewise.
7025 * common/config/epiphany/epiphany-common.c: Likewise.
7026 * common/config/fr30/fr30-common.c: Likewise.
7027 * common/config/frv/frv-common.c: Likewise.
7028 * common/config/ia64/ia64-common.c: Likewise.
7029 * common/config/iq2000/iq2000-common.c: Likewise.
7030 * common/config/lm32/lm32-common.c: Likewise.
7031 * common/config/m32r/m32r-common.c: Likewise.
7032 * common/config/mcore/mcore-common.c: Likewise.
7033 * common/config/microblaze/microblaze-common.c: Likewise.
7034 * common/config/mips/mips-common.c: Likewise.
7035 * common/config/mmix/mmix-common.c: Likewise.
7036 * common/config/mn10300/mn10300-common.c: Likewise.
7037 * common/config/nios2/nios2-common.c: Likewise.
7038 * common/config/pa/pa-common.c: Likewise.
7039 * common/config/pdp11/pdp11-common.c: Likewise.
7040 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7041 * common/config/riscv/riscv-common.c: Likewise.
7042 * common/config/rs6000/rs6000-common.c: Likewise.
7043 * common/config/rx/rx-common.c: Likewise.
7044 * common/config/s390/s390-common.c: Likewise.
7045 * common/config/sh/sh-common.c: Likewise.
7046 * common/config/sparc/sparc-common.c: Likewise.
7047 * common/config/tilegx/tilegx-common.c: Likewise.
7048 * common/config/tilepro/tilepro-common.c: Likewise.
7049 * common/config/v850/v850-common.c: Likewise.
7050 * common/config/visium/visium-common.c: Likewise.
7051 * common/config/xstormy16/xstormy16-common.c: Likewise.
7052 * common/config/xtensa/xtensa-common.c: Likewise.
7053 * invoke.texi (-fomit-frame-pointer): Update documentation.
7055 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7057 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7059 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7061 * cfg.c (scale_bbs_frequencies_int,
7062 cale_bbs_frequencies_gcov_type): Remove.
7063 * cfg.h (scale_bbs_frequencies_int,
7064 cale_bbs_frequencies_gcov_type): Remove.
7066 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7068 * tree-ssa-loop-manip.c
7069 (scale_dominated_blocks_in_loop): Update to profile counts.
7070 (tree_transform_and_unroll_loop): Likewise.
7072 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7074 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7075 scale_bbs_frequencies_int.
7077 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7079 * final.c (compute_alignments): Use counts rather than frequencies.
7081 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7083 * cfgloopanal.c: Include sreal.h
7084 (average_num_loop_insns): Use counts and sreal for accounting.
7086 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7088 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7091 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7092 Kito Cheng <kito.cheng@gmail.com>
7094 * config/nds32/constraints.md: Provide more constraints.
7095 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7096 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7097 support constraints usage.
7099 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7101 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7103 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7104 Kito Cheng <kito.cheng@gmail.com>
7106 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7107 * config/nds32/nds32.opt: Refine the layout.
7108 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7109 TARGET_EXT_STRING): Support new options.
7110 * config/nds32/nds32.h: Likewise.
7111 * config/nds32/nds32.md: Likewise.
7112 * config/nds32/nds32-predicates.c: Likewise.
7113 * config/nds32/constraints.md: Likewise.
7114 * common/config/nds32/nds32-common.c: Likewise.
7116 2017-11-16 Julia Koval <julia.koval@intel.com>
7119 * config/i386/gfniintrin.h: Add sse check.
7120 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7122 2017-11-16 Julia Koval <julia.koval@intel.com>
7124 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7125 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7126 (ix86_handle_option): Handle -mavx512vbmi2.
7127 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7128 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7129 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7130 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7131 (ix86_valid_target_attribute_inner_p): Ditto.
7132 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7133 * config/i386/i386.opt (mavx512vbmi2): New option.
7134 * doc/invoke.texi: Add new option.
7136 2017-11-16 Julia Koval <julia.koval@intel.com>
7138 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7139 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7140 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7141 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7142 _mm512_gf2p8mul_epi8): New intrinsics.
7143 * config/i386/i386-builtin-types.def
7144 (V64QI_FTYPE_V64QI_V64QI): New type.
7145 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7146 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7147 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7148 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7149 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7150 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7152 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7154 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7155 explicitly as a stream of bytes.
7157 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7159 * config/rs6000/altivec.h (vec_xst_be): New #define.
7160 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7161 and externalize from *altivec_vperm_<mode>_internal.
7162 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7164 (XL_BE_V8HI): Likewise.
7165 (XL_BE_V4SI): Likewise.
7166 (XL_BE_V4SI): Likewise.
7167 (XL_BE_V2DI): Likewise.
7168 (XL_BE_V4SF): Likewise.
7169 (XL_BE_V2DF): Likewise.
7170 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7171 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7172 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7173 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7174 VSX_BUILTIN_VEC_XST_BE.
7175 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7176 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7178 (altivec_init_builtins): Replace conditional calls to def_builtin
7179 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7180 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7181 with unconditional calls. Remove calls to def_builtin for
7182 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7183 __builtin_vec_xst_be.
7184 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7185 to define_expand, and add alternate RTL generation for P8.
7186 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7187 vsx_ld_elemrev_v8hi.
7188 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7189 add alternate RTL generation for P8.
7190 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7191 vsx_ld_elemrev_v16qi.
7192 (vsx_st_elemrev_v8hi): Convert define_insn
7193 to define_expand, and add alternate RTL generation for P8.
7194 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7195 vsx_st_elemrev_v8hi.
7196 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7197 add alternate RTL generation for P8.
7198 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7199 vsx_st_elemrev_v16qi.
7201 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7204 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7205 TARGET_AVX512ER check.
7206 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7207 neither -mzeroupper nor -mno-zeroupper is used and
7208 TARGET_EMIT_VZEROUPPER is set.
7209 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7210 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7212 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7214 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7215 folding of vector compares.
7216 (fold_build_vec_cmp): New helper function.
7217 (fold_compare_helper): New helper function.
7218 (builtin_function_type): Add compare builtins to the list of functions
7219 having unsigned arguments. Cosmetic updates to comment indentation.
7220 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7221 the not+eq combination.
7223 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7225 PR tree-optimization/82726
7226 PR tree-optimization/70754
7227 * tree-predcom.c (order_drefs_by_pos): New function.
7228 (combine_chains): Move code setting has_max_use_after to...
7229 (try_combine_chains): ...here. New parameter. Sort combined chains
7230 according to position information.
7231 (tree_predictive_commoning_loop): Update call to above function.
7232 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7234 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7236 PR tree-optimization/82726
7238 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7240 PR tree-optimization/70754
7241 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7242 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7243 combined stmt before it if not NULL.
7244 (combine_chains): Process refs reversely and compute dominance point
7248 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7250 PR tree-optimization/79663
7251 * tree-predcom.c (combine_chains): Process refs in reverse order
7252 only for ZERO length chains, and add explaining comment.
7254 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7256 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7257 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7258 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7260 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7262 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7264 2017-11-15 Martin Liska <mliska@suse.cz>
7266 * tree-cfg.c (pass_warn_function_return::execute):
7267 Compare warn_return_type for greater than zero.
7269 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7273 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7274 to return true on Xeon and not on Xeon Phi.
7275 (ix86_check_avx256_register): Changed to ...
7276 (ix86_check_avx_upper_register): ... this. Add extra check for
7277 VALID_AVX512F_REG_OR_XI_MODE.
7278 (ix86_avx_u128_mode_needed): Changed
7279 ix86_check_avx256_register to ix86_check_avx_upper_register.
7280 (ix86_check_avx256_stores): Changed to ...
7281 (ix86_check_avx_upper_stores): ... this. Changed
7282 ix86_check_avx256_register to ix86_check_avx_upper_register.
7283 (ix86_avx_u128_mode_after): Changed
7284 avx_reg256_found to avx_upper_reg_found. Changed
7285 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7286 (ix86_avx_u128_mode_entry): Changed
7287 ix86_check_avx256_register to ix86_check_avx_upper_register.
7288 (ix86_avx_u128_mode_exit): Ditto.
7289 * config/i386/i386.h: (host_detect_local_cpu): New define.
7291 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7293 * config/arm/xgene1.md (xgene1): Split into automatons
7294 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7295 (xgene1_f_load): Adjust reservations and/or types.
7296 (xgene1_f_store): Likewise.
7297 (xgene1_load_pair): Likewise.
7298 (xgene1_store_pair): Likewise.
7299 (xgene1_fp_load1): Likewise.
7300 (xgene1_load1): Likewise.
7301 (xgene1_store1): Likewise.
7302 (xgene1_move): Likewise.
7303 (xgene1_alu): Likewise.
7304 (xgene1_simd): Likewise.
7305 (xgene1_bfm): Likewise.
7306 (xgene1_neon_load1): Likewise.
7307 (xgene1_neon_store1): Likewise.
7308 (xgene1_neon_logic): Likewise.
7309 (xgene1_neon_st1): Likewise.
7310 (xgene1_neon_ld1r): Likewise.
7311 (xgene1_alu_cond): Added.
7312 (xgene1_shift_reg): Likwise.
7313 (xgene1_bfx): Likewise.
7314 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7316 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7319 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7321 (can_widen_mult_without_libcall): New function.
7322 (expand_mul_overflow): If only checking unsigned mul overflow,
7323 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7324 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7325 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7326 (expand_DIVMOD): Formatting fix.
7327 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7328 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7329 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7331 PR tree-optimization/82977
7332 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7333 constructed temporary to strlen_to_stridx.put.
7335 2017-11-15 Martin Liska <mliska@suse.cz>
7337 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7338 * configure: Regenerate.
7340 2017-11-15 Martin Liska <mliska@suse.cz>
7343 * config/sh/sh-mem.cc: Use proper probability for
7346 2017-11-14 Jeff Law <law@redhat.com>
7348 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7349 the red zone for stack_clash_protection_final_dynamic_probe targets
7350 when the total dynamic stack size is zero bytes.
7352 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7353 blocks is post order.
7355 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7357 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7358 * final.c (rest_of_clean_state): Set it for the
7359 -fcompare-debug dump.
7360 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7361 class when TDF_COMPARE_DEBUG is set.
7363 * dwarf2out.c (gen_producer_string): Discard
7366 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7369 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7370 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7371 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7372 (__TGMATH_REAL_2_3): Remove macros.
7373 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7374 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7376 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7378 (remquo): Define using __TGMATH_REAL_3.
7380 2017-11-14 Jeff Law <law@redhat.com>
7382 * vr-values.c: New file with contents extracted from tree-vrp.c.
7383 * Makefile.in (OBJS): Add vr-values.o
7384 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7385 (set_value_range, set_and_canonicalize_value_range): Likewise.
7386 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7387 (value_range_constant_singleton, symbolic_range_p): Likewise.
7388 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7389 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7390 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7391 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7392 (find_case_label_range, find_case_label_index): Likewise.
7393 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7394 (range_int_cst_singleton_p, value_inside_range): Likewise.
7395 (get_single_symbol): Likewise.
7396 (switch_update): Move structure definition here.
7397 (to_remove_edges, to_update_switch_stmts): Provide externs.
7398 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7399 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7400 (vrp_val_is_min, set_value_range): Likewise.
7401 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7402 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7403 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7404 (range_is_nonnull, range_int_cst_p): Likewwise.
7405 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7406 (get_single_symbol, operand_less_p): Likewise
7407 (compare_values_warnv, compare_values): Likewise.
7408 (value_inside_range, value_range_constant_singleton): Likewise.
7409 (zero_nonzero_bitgs_from_vr): Likewise.
7410 (extract_range_from_binary_expr_1): Likewise.
7411 (overflow_comparison_p): Likewise.
7412 (to_remove_edges, to_update_switch_stmts): Likewise.
7413 (find_case_label-index, find_case_label_range): Likewise.
7414 (switch_update, set_value_range_to_nonnegative): Remove.
7415 (set_value_range_to_truthvalue): Likewise.
7416 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7417 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7418 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7419 (find_case_label_ranges, test_for_singularity): Likewise.
7420 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7421 (x_vr_values): Move to its remaining use site.
7423 2017-11-10 Jeff Law <law@redhat.com>
7425 * vr-values.h (VR_INITIALIZER): Move #define here.
7426 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7427 * Makefile.in (OBJS): Add tree-evrp.o
7428 * tree-vrp.h (assert_info): Move structure definition here.
7429 (set_value_range_to_varying): Prototype.
7430 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7431 (infer_value_range, register_edge_assert_for): Likewise.
7432 (stmt_interesting_for_vrp): Likewise.
7433 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7434 (set_value_range_to_varying): No longer static.
7435 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7436 (infer_value_range, register_edge_assert_for): Likewise.
7438 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7440 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7441 generate the XXBRD instruction.
7443 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7444 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7445 as being compatible if -mabi=ieeelongdouble.
7446 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7447 to setup float128 built-ins with hardware support.
7448 (BU_FLOAT128_HW_2): Likewise.
7449 (BU_FLOAT128_HW_3): Likewise.
7450 (BU_FLOAT128_HW_VSX_1): Likewise.
7451 (BU_FLOAT128_HW_VSX_2): Likewise.
7452 (scalar_extract_expq): Change float128 built-in functions to
7453 accommodate having both KFmode and TFmode functions. Use the
7454 KFmode variant as the default.
7455 (scalar_extract_sigq): Likewise.
7456 (scalar_test_neg_qp): Likewise.
7457 (scalar_insert_exp_q): Likewise.
7458 (scalar_insert_exp_qp): Likewise.
7459 (scalar_test_data_class_qp): Likewise.
7460 (sqrtf128_round_to_odd): Delete processing the round to odd
7461 built-in functions as special built-in functions, and define them
7462 as float128 built-ins. Use the KFmode variant as the default.
7463 (truncf128_round_to_odd): Likewise.
7464 (addf128_round_to_odd): Likewise.
7465 (subf128_round_to_odd): Likewise.
7466 (mulf128_round_to_odd): Likewise.
7467 (divf128_round_to_odd): Likewise.
7468 (fmaf128_round_to_odd): Likewise.
7469 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7470 support for KFmode and TFmode xststdcqp calls.
7471 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7472 point, switch the built-in handlers for the get/set float128
7473 exponent, get float128 mantissa, float128 test built-ins, and the
7474 float128 round to odd built-in functions. Eliminate creating the
7475 float128 round to odd built-in functions as special built-ins.
7476 (rs6000_init_builtins): Eliminate special creation of the float128
7477 round to odd built-in functions.
7478 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7479 function insns to support both TFmode and KFmode variants.
7480 (xsxsigqp_<mode>): Likewise.
7481 (xsiexpqpf_<mode>): Likewise.
7482 (xsiexpqp_<mode>): Likewise.
7483 (xststdcqp_<mode>): Likewise.
7484 (xststdcnegqp_<mode>): Likewise.
7485 (xststdcqp_<mode>): Likewise.
7487 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7489 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7490 unused path_in_freq_ptr parameter.
7491 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7493 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7495 * ipa-inline.c (edge_badness): Dump sreal frequency.
7496 (compute_inlined_call_time): Match natural implementaiton ...
7497 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7498 forgotten division by CGRAPH_FREQ_BASE.
7500 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7502 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7503 Solaris 11. Update comment.
7504 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7506 * config/sol2.h (STARTFILE_SPEC): Likewise.
7507 * configure: Regenerate.
7509 2017-11-14 Carl Love <cel@us.ibm.com>
7511 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7512 le_ and be_ prefixes to swap* variables. Remove
7513 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7516 2017-11-14 Jason Merrill <jason@redhat.com>
7518 Support GTY((cache)) on hash_map.
7519 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7520 (ggc_cache_remove): Override it instead of ggc_mx.
7521 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7522 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7523 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7524 (simple_cache_map_traits): Override maybe_mx.
7525 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7526 (hash_map): Friend gt_cleare_cache.
7527 (gt_cleare_cache): New.
7528 * tree.h (tree_cache_traits): New hash_map traits class.
7529 (tree_cache_map): New typedef.
7531 2017-11-14 Richard Biener <rguenther@suse.de>
7533 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7534 paramter and handling.
7535 (cleanup_control_flow_bb): Likewise.
7536 (cleanup_control_flow_pre): New helper performing a DFS walk
7537 to call cleanup_control_flow_bb in PRE order.
7538 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7539 via cleanup_control_flow_pre.
7541 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7543 * config/aarch64/aarch64-simd.md
7544 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7545 (*aarch64_simd_bsl<mode>_alt): Likewise.
7546 (aarch64_simd_bsldi_internal): New.
7547 (aarch64_simd_bsldi_alt): Likewise.
7549 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7551 * tracer.c (better_p): Do not compare frequencies.
7552 * reg-stack.c (better_edge): Likewise.
7553 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7556 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7558 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7559 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7560 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7561 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7562 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7563 * predict.c (maybe_hot_frequency_p): Inline to ...
7564 (maybe_hot_count_p): ... here; rewrite to counts.
7565 (counts_to_freqs): Rename to ...
7566 (update_max_bb_count): ... this one.
7567 (expensive_function_p): Use counts.
7568 (estimate_bb_frequencies): Update.
7569 (rebuild_frequencies): Update.
7570 * predict.h (counts_to_freqs): Rename to ...
7571 (update_max_bb_count): ... this one.
7572 * profile.c (compute_branch_probabilities): Add debug info
7573 * tree-inline.c (expand_call_inline): Update debug info.
7574 (optimize_inline_calls): Use update_max_bb_count..
7575 (tree_function_versioning): Use update_max_bb_count..
7576 * value-prof.c (gimple_value_profile_transformations):
7577 Do not use update_max_bb_count.
7579 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7581 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7582 always use frequencies.
7584 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7586 * bb-reorder.c: Remove frequencies from comments.
7587 (better_edge_p): Use profile counts.
7588 (find_traces): Dump profile counts.
7589 (rotate_loop): Use profile counts.
7590 (find_traces_1_round): Likewise.
7591 (connect_better_edge_p): Use counts instead of probabilities for
7593 (copy_bb_p): Drop early check for non-0 frequency.
7594 (sanitize_hot_paths): Update comments.
7596 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7598 * ipa-split.c (struct split_point): Add count.
7599 (consider_split): Do not compute incoming frequency; compute incoming
7600 count and store it to split_point.
7601 (split_function): Set count of the call to split part correctly.
7603 2017-11-13 Carl Love <cel@us.ibm.com>
7605 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7607 2017-11-13 Tom Tromey <tom@tromey.com>
7609 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7611 2017-11-13 Carl Love <cel@us.ibm.com>
7613 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7614 Add support for builtins:
7615 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7616 vector {un,}signed {char,int,short},
7617 vector {un,}signed {char,int,short}) arguments.
7618 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7619 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7620 Add BU_P9V_AV_2 expansions for the builtins.
7621 * config/rs6000/altivec.h (vec_first_match_index,
7622 vec_first_mismatch_index, vec_first_match_or_eos_index,
7623 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7624 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7625 new extern declaration.
7626 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7627 * config/rs6000/vsx.md (first_match_index_<mode>,
7628 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7629 first_mismatch_or_eos_index_<mode>): Add define expand.
7630 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7631 * doc/extend.texi: Update the built-in documenation file for the new
7634 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7636 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7637 into the min/max operations for _Float<N> and _Float<N>X types.
7639 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7642 * dwarf2out.c (do_eh_frame): New static variable.
7643 (dwarf2out_begin_prologue): Set it.
7644 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7646 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7648 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7650 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7652 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7655 * debug.h (dwarf2out_do_eh_frame): Declare.
7656 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7657 (dwarf2out_do_frame): Use it.
7658 (dwarf2out_do_cfi_asm): Likewise.
7659 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7660 (dwarf2out_assembly_start): Likewise.
7661 (dwarf2out_begin_prologue): Fix comment.
7662 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7663 if the target needs either debug or unwind DWARF2 info.
7664 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7665 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7666 -ftrapping-math, -ftrapv and -fwrapv.
7668 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7670 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7671 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7672 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7673 (estimate_edge_size_and_time): Likewise.
7674 (ipa_merge_fn_summary_after_inlining): Likewise.
7675 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7676 (compute_uninlined_call_time): Use sreal_frequency.
7677 (compute_inlined_call_time): Likewise.
7678 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7679 * profile-count.c: Include sreal.h.
7680 (profile_count::to_sreal_scale): New.
7681 * profile-count.h: Forward declare sreal.
7682 (profile_count::to_sreal_scale): Declare.
7684 2017-11-13 Nathan Sidwell <nathan@acm.org>
7686 * diagnostic.c (maybe_line_and_column): New.
7687 (diagnostic_get_location_text): Use it.
7688 (diagnostic_report_current_module): Likewise.
7689 (test_diagnostic_get_location_text): Add tests.
7691 2017-11-13 Luis Machado <luis.machado@linaro.org>
7693 * doc/md.texi (Specifying processor pipeline description): Fix
7694 incorrect latency for the div instruction example.
7696 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7698 PR tree-optimization/78821
7699 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7700 that bit_not_p is the same.
7701 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7702 (split_group): Count precisely bit_not_p bits in each statement.
7703 (invert_op): New function.
7704 (imm_store_chain_info::output_merged_store): Use invert_op to
7705 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7706 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7708 2017-11-13 Martin Liska <mliska@suse.cz>
7710 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7711 (struct source_info): Likewise.
7712 (add_branch_counts): Likewise.
7713 (add_line_counts): Likewise.
7714 (function_summary): Likewise.
7715 (output_intermediate_line): Likewise.
7716 (generate_results): Likewise.
7718 2017-11-13 Martin Liska <mliska@suse.cz>
7720 * gcov.c (struct block_info): Remove typedef for block_t.
7721 (struct line_info): Likewise.
7722 (line_info::has_block): Likewise.
7723 (EXIT_BLOCK): Likewise.
7724 (unblock): Likewise.
7725 (circuit): Likewise.
7726 (get_cycles_count): Likewise.
7727 (process_file): Likewise.
7728 (read_graph_file): Likewise.
7729 (solve_flow_graph): Likewise.
7730 (find_exception_blocks): Likewise.
7731 (add_line_counts): Likewise.
7732 (accumulate_line_info): Likewise.
7733 (output_line_details): Likewise.
7735 2017-11-13 Martin Liska <mliska@suse.cz>
7737 * gcov.c (struct arc_info): Remove typedef for arc_t.
7738 (struct line_info): Likewise.
7739 (add_branch_counts): Likewise.
7740 (output_branch_count): Likewise.
7741 (function_info::~function_info): Likewise.
7742 (circuit): Likewise.
7743 (output_intermediate_line): Likewise.
7744 (read_graph_file): Likewise.
7745 (solve_flow_graph): Likewise.
7746 (find_exception_blocks): Likewise.
7747 (add_line_counts): Likewise.
7748 (accumulate_line_info): Likewise.
7749 (output_line_details): Likewise.
7750 (output_function_details): Likewise.
7752 2017-11-13 Martin Liska <mliska@suse.cz>
7754 * gcov.c (struct function_info): Remove typedef for function_t.
7755 (struct source_info): Likewise.
7756 (source_info::get_functions_at_location): Likewise.
7757 (solve_flow_graph): Likewise.
7758 (find_exception_blocks): Likewise.
7759 (add_line_counts): Likewise.
7760 (output_intermediate_file): Likewise.
7761 (process_file): Likewise.
7762 (generate_results): Likewise.
7763 (release_structures): Likewise.
7764 (read_graph_file): Likewise.
7765 (read_count_file): Likewise.
7766 (accumulate_line_counts): Likewise.
7767 (output_lines): Likewise.
7769 2017-11-13 Martin Liska <mliska@suse.cz>
7771 * gcov.c (function_info::function_info): Remove num_counts
7772 and add vector<gcov_type>.
7773 (function_info::~function_info): Use the vector.
7774 (process_file): Likewise.
7775 (read_graph_file): Likewise.
7776 (read_count_file): Likewise.
7777 (solve_flow_graph): Likewise.
7779 2017-11-13 Martin Liska <mliska@suse.cz>
7781 * gcov.c (function_info::is_artificial): New function.
7782 (process_file): Erase all artificial early.
7783 (generate_results): Skip as all artificial are already
7786 2017-11-13 Martin Liska <mliska@suse.cz>
7788 * gcov.c (read_graph_file): Store to global vector of functions.
7789 (read_count_file): Iterate the vector.
7790 (process_file): Likewise.
7791 (generate_results): Likewise.
7792 (release_structures): Likewise.
7794 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7796 PR tree-optimization/82954
7797 * gimple-ssa-store-merging.c
7798 (imm_store_chain_info::coalesce_immediate_stores): If
7799 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7801 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7803 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7804 Upddate call to ENDIAN_LANE_N.
7805 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7806 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7807 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7808 and use aarch64_endian_lane_rtx.
7809 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7811 2017-11-12 Tom de Vries <tom@codesourcery.com>
7813 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7815 2017-11-12 Tom de Vries <tom@codesourcery.com>
7817 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7819 2017-11-12 Tom de Vries <tom@codesourcery.com>
7821 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7823 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7824 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7825 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7827 2017-11-11 Martin Sebor <msebor@redhat.com>
7830 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7833 * prefic.c (translate_name): Replace strncpy with memcpy to
7834 avoid -Wstringop-truncation.
7836 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7838 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7840 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7842 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7844 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7846 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7849 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7851 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7852 enable generating XXBRH if the value is in a vector register.
7853 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7854 value is in a vector register.
7855 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7856 register to register bswap64's instead of doing the GPR sequence
7857 used on previous machines.
7858 (bswapdi2_xxbrd): New insn.
7859 (bswapdi2_reg): Disallow on ISA 3.0.
7860 (register to register bswap64 splitter): Do not split the insn on
7861 ISA 3.0 systems that use XXBRD.
7863 2017-11-10 Martin Sebor <msebor@redhat.com>
7866 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7868 (find_subframework_file): Same.
7870 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7872 * auto-profile.c (afdo_indirect_call): Drop frequency.
7873 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7874 (cgraph_node::create_edge): Drop frequency argument.
7875 (cgraph_node::create_indirect_edge): Drop frequency argument.
7876 (cgraph_edge::make_speculative): Drop frequency arguments.
7877 (cgraph_edge::resolve_speculation): Do not update frequencies
7878 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7879 (cgraph_node::dump): Check consistency in IPA mode.
7880 (cgraph_edge::maybe_hot_p): Use IPA counter.
7881 (cgraph_edge::verify_count_and_frequency): Rename to ...
7882 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7883 (cgraph_node::verify_node): Update.
7884 * cgraph.h (struct cgraph_edge): Drop frequency.
7885 (cgraph_edge::frequency): New function.
7886 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7888 (cgraph_edge::rebuild_edges): Likewise.
7889 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7890 (duplicate_thunk_for_node): Do not pass frequency.
7891 (cgraph_node::create_clone): Scale only counts.
7892 (cgraph_node::create_virtual_clone): Do not pass frequency.
7893 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7894 (cgraph_node::create_version_clone): Do not pass frequency.
7895 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7896 (cgraph_node::expand_thunk): Do not pass frequency.
7897 (cgraph_node::create_wrapper): Do not pass frequency.
7898 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7900 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7901 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7902 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7903 (gather_caller_stats): Use frequency function.
7904 (ipcp_cloning_candidate_p): Use frequency function.
7905 (ipcp_propagate_stage): Use frequency function.
7906 (get_info_about_necessary_edges): Use frequency function.
7907 (update_profiling_info): Update only IPA profile.
7908 (update_specialized_profile): Use frequency functoin.
7909 (perhaps_add_new_callers): Update only IPA profile.
7910 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7911 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7912 (dump_ipa_call_summary): Use frequency function.
7913 (estimate_edge_size_and_time): Use frequency function.
7914 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7915 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7916 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7917 (update_noncloned_counts): ... ths one; scale counts only.
7918 (clone_inlined_nodes): Do not scale frequency.
7919 (inline_call): Do not pass frequency.
7920 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7921 (compute_inlined_call_time): Use IPA profile.
7922 (want_inline_small_function_p): Use IPA profile.
7923 (want_inline_self_recursive_call_p): Use IPA profile.
7924 (edge_badness): Use IPA profile.
7925 (lookup_recursive_calls): Use IPA profile.
7926 (recursive_inlining): Do not pass frequency.
7927 (resolve_noninline_speculation): Do not update frequency.
7928 (inline_small_functions): Collect max of IPA profile.
7929 (dump_overall_stats): Dump IPA porfile.
7930 (dump_inline_stats): Dump IPA porfile.
7931 (ipa_inline): Collect IPA stats.
7932 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7933 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7934 (ipa_propagate_frequency): Use frequency function.
7935 (ipa_profile): Cleanup.
7936 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7937 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7938 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7939 (input_node): Do not stream frequency.
7940 (input_edge): Do not stream frequency.
7941 (merge_profile_summaries): Scale only IPA profiles.
7942 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7943 * predict.c (drop_profile): Do not recompute frequency.
7944 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7945 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7946 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7947 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7948 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7949 * tree-inline.c (copy_bb): Do not scale frequency.
7950 (expand_call_inline): Do not scale frequency.
7951 (tree_function_versioning): Do not scale frequency.
7952 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7954 2017-11-10 Julia Koval <julia.koval@intel.com>
7956 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7957 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7958 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7959 (_mm256_maskz_gf2p8affine_epi64_epi8)
7960 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7961 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7962 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7963 (__builtin_ia32_vgf2p8affineqb_v32qi)
7964 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7965 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7967 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7971 (arm_option_override): Refactor.
7972 (arm_option_reconfigure_globals): New.
7973 (arm_options_perform_arch_sanity_checks): New.
7974 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7976 (arm_options_perform_arch_sanity_checks): Likewise
7978 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7980 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7981 (power9-qpmul): New.
7982 * rs6000/rs6000.md ("type" attr): Add qmul.
7983 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7984 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7985 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7987 2017-11-10 Martin Sebor <msebor@redhat.com>
7990 * builtins.c (compute_objsize): Handle arrays that
7991 compute_builtin_object_size likes to fail for. Make extern.
7992 * builtins.h (compute_objsize): Declare.
7993 (check_strncpy_sizes): New function.
7994 (expand_builtin_strncpy): Call check_strncpy_sizes.
7995 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7996 -Wstringop-truncation.
7997 (gimple_fold_builtin_strncat): Same.
7998 * gimple.c (gimple_build_call_from_tree): Set call location.
7999 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8000 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8001 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8002 (handle_builtin_strlen): Use strlen_to_stridx.
8003 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8005 Use strlen_to_stridx.
8006 (pass_strlen::execute): Release strlen_to_stridx.
8007 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8008 (-Wstringop-truncation): Document new option.
8010 2017-11-10 Martin Liska <mliska@suse.cz>
8012 PR gcov-profile/82702
8013 * gcov.c (main): Handle intermediate files in a different
8015 (get_gcov_intermediate_filename): New function.
8016 (output_gcov_file): Remove support of intermediate files.
8017 (generate_results): Allocate intermediate file.
8018 (release_structures): Clean-up properly fn_end.
8019 (output_intermediate_file): Start iterating with line 1.
8021 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8023 PR tree-optimization/82929
8024 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8025 ops_swapped_p non-static data member.
8026 (store_immediate_info::store_immediate_info): Clear it.
8027 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8028 ops set ops_swapped_p.
8029 (count_multiple_uses): Handle ops_swapped_p.
8031 2017-11-10 Martin Liska <mliska@suse.cz>
8033 * coverage.c (coverage_init): Stream information about
8034 support of has_unexecuted_blocks.
8035 * doc/gcov.texi: Document that.
8036 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8037 * gcov.c (read_graph_file): Likewise.
8038 (output_line_beginning): Fix a small issue with
8041 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8043 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8044 reference of trivial component.
8046 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8049 * gimple-ssa-store-merging.c
8050 (pass_store_merging::terminate_all_aliasing_chains): For
8051 gimple_store_p stmts also call refs_output_dependent_p.
8053 PR rtl-optimization/82913
8054 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8057 2017-11-09 Jeff Law <law@redhat.com>
8059 * vr-values.h: New file with vr_values class.
8060 * tree-vrp.c: Include vr-values.h
8061 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8062 data objects into the vr_values class.
8063 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8064 (get_value_range): Make it a member function within vr_values class.
8065 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8066 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8067 (op_with_constant_singleton_value_range): Likewise.
8068 (extract_range_for_var_from_comparison_expr): Likewise.
8069 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8070 (extract_range_from_binary_expr): Likewise.
8071 (extract_range_from_unary_expr): Likewise.
8072 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8073 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8074 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8075 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8076 (compare_name_with_value, compare_names): Likewise.
8077 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8078 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8079 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8080 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8081 (extract_range_from_phi_node): Likewise.
8082 (simplify_truth_ops_using_ranges): Likewise.
8083 (simplify_div_or_mod_using_ranges): Likewise.
8084 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8085 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8086 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8087 (simplify_float_conversion_using_ranges): Likewise.
8088 (simplify_internal_call_using_ranges): Likewise.
8089 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8090 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8091 poitner via x_vr_values for calls into gimple folder.
8092 (vrp_initialize_lattice): Make this the vr_values ctor.
8093 (vrp_free_lattice): Make this the vr_values dtor.
8094 (set_vr_value): New function.
8095 (class vrp_prop): Add vr_values data member. Add various member
8096 functions as well as member functions that delegate to vr_values.
8097 (check_array_ref): Make a member function within vrp_prop class.
8098 (search_for_addr_array, vrp_initialize): Likewise.
8099 (vrp_finalize): Likewise. Revamp to avoid direct access to
8100 vr_value, values_propagated, etc.
8101 (check_array_bounds): Extract vrp_prop class instance pointer from
8102 walk info structure. Use it to call member functions.
8103 (check_all_array_refs): Make a member function within vrp_prop class.
8104 Smuggle class instance pointer via walk info structure.
8105 (x_vr_values): New local static.
8106 (vrp_valueize): Use x_vr_values to get class instance.
8107 (vr_valueize_1): Likewise.
8108 (class vrp_folder): Add vr_values data member. Add various member
8109 functions as well as member functions that delegate to vr_values.
8110 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8111 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8112 class instance from vr_values. Use it to call member functions.
8113 (vrp_dom_walker): Add vr_values data member.
8114 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8115 instance via x_vr_values.
8116 (identify_jump_threads): Accept vr_values as argument. Store
8117 it into the walker structure.
8118 (evrp_dom_walker): Add vr_values class data member. Add various
8120 (evrp_dom_walker::try_find_new_range): Use vr_values data
8121 member to access the memory allocator.
8122 (evrp_dom_walker::before_dom_children): Store vr_values class
8123 instance into the vrp_folder class.
8124 (evrp_dom_walker::push_value_range): Rework to avoid direct
8125 access to num_vr_values and vr_value.
8126 (evrp_dom_walker::pop_value_range): Likewise.
8127 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8128 Use vr_values to get to dump_all_value_ranges member function.
8129 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8130 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8131 Pass vr_values class instance down to identify_jump_threads.
8132 Remove call to vrp_free_lattice.
8133 (debug_all_value_ranges): Remove.
8135 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8136 (vrp_folder): Likewise.
8138 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8139 Get it from the existing bitmap instead.
8140 (vrp_intersect_ranges_1): Likewise.
8142 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8144 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8146 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8147 set bit_not_p to it.
8148 (imm_store_chain_info::coalesce_immediate_stores): Break group
8149 if bit_not_p is different.
8150 (count_multiple_uses, split_group,
8151 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8152 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8153 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8154 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8155 to store_immediate_info ctor.
8157 2017-11-09 Jim Wilson <jimw@sifive.com>
8159 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8160 (scan_prog_file): Likewise.
8162 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8164 * bb-reorder.c (max_entry_frequency): Remove.
8165 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8166 connect_traces, push_to_next_round_p): Remove prototypes.
8167 (find_traces_1_round): Use counts only.
8168 (push_to_next_round_p): Likewise.
8169 (find_traces): Likewise.
8170 (rotate_loop): Likewise.
8171 (find_traces_1_round): Likewise.
8172 (connect_traces): Likewise.
8173 (edge_order): Likewise.
8175 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8177 * config/arm/arm.c (output_return_instruction): Add comments to
8178 indicate requirement for cmse_nonsecure_entry return to account
8179 for the size of clearing instruction output here.
8180 (thumb_exit): Likewise.
8181 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8182 return in hardfloat mode.
8184 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8186 * config/rs6000/rs6000.c (machine_function): Add a bool,
8187 "toc_is_wrapped_separately".
8188 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8189 if it wasn't explicitly set or unset, we are optimizing for speed, and
8190 doing separate shrink-wrapping.
8191 (rs6000_get_separate_components): Enable the TOC component if
8192 saving the TOC register in the prologue.
8193 (rs6000_components_for_bb): Handle the TOC component.
8194 (rs6000_emit_prologue_components): Store the TOC register where needed.
8195 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8196 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8198 2017-11-09 Martin Jambor <mjambor@suse.cz>
8200 * ipa-param-manipulation.c: New file.
8201 * ipa-param-manipulation.h: Likewise.
8202 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8203 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8204 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8205 (ipa_parm_adjustment): Likewise.
8206 (ipa_parm_adjustment_vec): Likewise.
8207 (ipa_get_vector_of_formal_parms): Moved declaration to
8208 ipa-param-manipulation.h.
8209 (ipa_get_vector_of_formal_parm_types): Likewise.
8210 (ipa_modify_formal_parameters): Likewise.
8211 (ipa_modify_call_arguments): Likewise.
8212 (ipa_combine_adjustments): Likewise.
8213 (ipa_dump_param_adjustments): Likewise.
8214 (ipa_modify_expr): Likewise.
8215 (ipa_get_adjustment_candidate): Likewise.
8216 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8217 ipa-param-manipulation.c.
8218 (ipa_get_vector_of_formal_parm_types): Likewise.
8219 (ipa_modify_formal_parameters): Likewise.
8220 (ipa_modify_call_arguments): Likewise.
8221 (ipa_modify_expr): Likewise.
8222 (get_ssa_base_param): Likewise.
8223 (ipa_get_adjustment_candidate): Likewise.
8224 (index_in_adjustments_multiple_times_p): Likewise.
8225 (ipa_combine_adjustments): Likewise.
8226 (ipa_dump_param_adjustments): Likewise.
8227 * tree-sra.c: Also include ipa-param-manipulation.h
8228 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8231 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8232 Alan Hayward <alan.hayward@arm.com>
8233 David Sherwood <david.sherwood@arm.com>
8235 * doc/sourcebuild.texi (vect_masked_store): Document.
8237 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8238 Alan Hayward <alan.hayward@arm.com>
8239 David Sherwood <david.sherwood@arm.com>
8241 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8243 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8244 Alan Hayward <alan.hayward@arm.com>
8245 David Sherwood <david.sherwood@arm.com>
8247 * doc/sourcebuild.texi (vect_variable_length): Document.
8249 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8250 Alan Hayward <alan.hayward@arm.com>
8251 David Sherwood <david.sherwood@arm.com>
8253 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8255 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8256 Alan Hayward <alan.hayward@arm.com>
8257 David Sherwood <david.sherwood@arm.com>
8259 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8261 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8262 Alan Hayward <alan.hayward@arm.com>
8263 David Sherwood <david.sherwood@arm.com>
8265 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8266 previously undocumented selectors.
8267 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8269 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8271 * doc/rtl.texi (const_vector): Say that elements can be
8272 const_wide_ints too.
8273 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8274 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8275 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8276 * optabs.c (expand_vector_broadcast): Likewise.
8278 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8279 Alan Hayward <alan.hayward@arm.com>
8280 David Sherwood <david.sherwood@arm.com>
8282 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8283 scaled index even if the unscaled address was invalid.
8284 Don't increase the complexity of using a scale in that case.
8286 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8287 Alan Hayward <alan.hayward@arm.com>
8288 David Sherwood <david.sherwood@arm.com>
8290 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8291 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8292 UNITS_PER_WORD bytes.
8293 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8294 based on REGMODE_NATURAL_SIZE of the inner mode.
8295 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8296 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8297 * expmed.c (lowpart_bit_field_p): Divide the value up into
8298 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8299 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8300 whether something is likely to occupy more than one register.
8302 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8305 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8306 profile_count::adjust_for_ipa_scaling.
8307 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8308 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8310 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8312 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8314 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8315 (split_group): Add total_orig and total_new arguments, estimate the
8316 number of statements related to the store group without store merging
8317 and with store merging.
8318 (imm_store_chain_info::output_merged_store): Adjust split_group
8319 callers, punt if estimated number of statements with store merging
8320 is not smaller than estimated number of statements without it.
8322 (handled_load): Remove has_single_use checks.
8323 (pass_store_merging::process_store): Likewise.
8325 2017-11-09 Richard Biener <rguenther@suse.de>
8327 PR tree-optimization/82902
8328 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8330 2017-11-09 Martin Liska <mliska@suse.cz>
8333 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8336 2017-11-09 Martin Liska <mliska@suse.cz>
8338 PR tree-optimization/82669
8339 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8341 2017-11-09 Martin Liska <mliska@suse.cz>
8343 PR gcov-profile/48463
8344 * coverage.c (coverage_begin_function): Output also end locus
8345 of a function and information whether the function is
8347 * gcov-dump.c (tag_function): Parse and print the information.
8348 * gcov.c (INCLUDE_MAP): Add include.
8349 (INCLUDE_SET): Likewise.
8350 (struct line_info): Move earlier in the source file because
8351 of vector<line_info> in function_info structure.
8352 (line_info::line_info): Likewise.
8353 (line_info::has_block): Likewise.
8354 (struct source_info): Add new member index.
8355 (source_info::get_functions_at_location): New function.
8356 (function_info::group_line_p): New function.
8357 (output_intermediate_line): New function.
8358 (output_intermediate_file): Use the mentioned function.
8359 (struct function_start): New.
8360 (struct function_start_pair_hash): Likewise.
8361 (process_file): Add code that identifies group functions.
8362 Assign lines either to global or function scope.
8363 (generate_results): Skip artificial functions.
8364 (find_source): Assign index for each source file.
8365 (read_graph_file): Read new flag artificial and end_line.
8366 (add_line_counts): Assign it either to global of function scope.
8367 (accumulate_line_counts): Isolate core of the function to
8368 accumulate_line_info and call it for both function and global
8370 (accumulate_line_info): New function.
8371 (output_line_beginning): Fix GNU coding style.
8372 (print_source_line): New function.
8373 (output_line_details): Likewise.
8374 (output_function_details): Likewise.
8375 (output_lines): Iterate both source (global) scope and function
8377 (struct function_line_start_cmp): New class.
8378 * doc/gcov.texi: Reflect changes in documentation.
8380 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8383 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8384 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8385 and similarly for not instead of neg.
8387 2017-11-08 Andi Kleen <ak@linux.intel.com>
8389 * config/i386/i386.opt: Add -mforce-indirect-call.
8390 * config/i386/predicates.md: Check for flag_force_indirect_call.
8391 * doc/invoke.texi: Document -mforce-indirect-call
8393 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8395 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8397 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8398 config/riscv/riscv.c (predict.h): New include.
8399 (riscv_slow_unaligned_access_p): No longer static.
8400 (riscv_block_move_straight): Add require.
8401 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8403 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8406 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8407 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8408 nonimmediate_operand predicate for operand 1 instead of
8411 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8413 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8415 * config/aarch64/constraints.md (Uml): New constraint.
8416 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8419 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8421 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8422 of two vec_duplicates into a vec_concat.
8424 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8426 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8427 Simplify vec_merge of vec_duplicate and vec_concat.
8428 * config/aarch64/constraints.md (Utq): New constraint.
8429 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8432 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8434 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8435 Simplify vec_merge of vec_duplicate and const_vector.
8436 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8438 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8439 mode iterator. Update predicate on operand 1 to
8440 handle non-const_vec constants. Delete constraints.
8441 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8443 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8445 PR tree-optimization/78821
8446 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8448 (store_operand_info::store_operand_info): Initialize it to false.
8449 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8450 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8451 store in the group, and if chain_info is non-NULL, to ignore altogether
8453 (compatible_load_p): Fail if bit_not_p does not match.
8454 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8455 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8456 (pass_store_merging::process_store): Adjust
8457 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8458 call terminate_all_aliasing_chains newly when adding a store into
8459 a chain with non-NULL chain_info.
8461 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8463 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8465 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8467 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8469 (aarch64_layout_frame): Initialise emit_frame_chain.
8470 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8471 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8473 2017-11-08 Martin Liska <mliska@suse.cz>
8475 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8476 of gimple_call_internal_p.
8478 2017-11-07 Tom de Vries <tom@codesourcery.com>
8480 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8482 2017-11-07 Tom de Vries <tom@codesourcery.com>
8484 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8487 2017-11-08 Martin Liska <mliska@suse.cz>
8490 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8492 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8494 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8496 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8499 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8500 (*cmp<mode>_ccz_1): New insn with $k alternative.
8503 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8504 RTX_COMM_COMPARE as commutative as well.
8505 (ix86_binary_operator_ok): Formatting fix.
8506 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8507 *<code><mode>3<mask_name><round_saeonly_name>,
8508 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8509 *<s>mul<mode>3_highpart<mask_name>,
8510 *vec_widen_umult_even_v16si<mask_name>,
8511 *vec_widen_umult_even_v8si<mask_name>,
8512 *vec_widen_umult_even_v4si<mask_name>,
8513 *vec_widen_smult_even_v16si<mask_name>,
8514 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8515 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8516 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8517 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8518 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8519 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8520 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8521 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8522 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8523 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8524 ix86_binary_operator_ok. Formatting fixes.
8525 (*<plusminus_insn><mode>3<mask_name><round_name>,
8526 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8529 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8531 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8532 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8533 eq and ne if TARGET_ISEL.
8534 (cmp): New code_iterator.
8535 (UNS, UNSU_, UNSIK): New code_attrs.
8536 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8537 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8539 ("eq<mode>3"): ... this.
8540 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8542 ("ne<mode>3"): ... this.
8544 2017-11-07 Julia Koval <julia.koval@intel.com>
8547 * common/config/i386/i386-common.c
8548 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8549 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8550 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8551 * config/i386/i386.opt: Ditto.
8552 * config/i386/i386.c (ix86_target_string): Ditto.
8553 (ix86_option_override_internal): Ditto.
8554 (ix86_init_mpx_builtins): Move MPX to args2.
8555 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8556 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8557 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8558 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8559 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8560 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8561 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8563 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8566 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8567 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8568 (zero-extendsidi peephole2): Remove peephole.
8570 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8573 * stor-layout.c: Include attribs.h.
8574 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8575 explicit lookup of "aligned" attribute.
8577 2017-11-07 Andrew Waterman <andrew@sifive.com>
8579 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8580 (riscv_expand_block_move): Likewise.
8581 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8583 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8584 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8585 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8586 (riscv_adjust_block_mem): Likewise.
8587 (riscv_block_move_loop): Likewise.
8588 (riscv_expand_block_move): Likewise.
8589 * config/riscv/riscv.md (movmemsi): New pattern.
8591 2017-11-07 Michael Clark <michaeljclark@mac.com>
8593 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8594 (MUSL_DYNAMIC_LINKER): Likewise.
8596 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8598 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8599 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8602 2017-11-07 Richard Biener <rguenther@suse.de>
8604 * match.pd: Fix build.
8606 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8607 Jackson Woodruff <jackson.woodruff@arm.com>
8609 PR tree-optimization/71026
8610 * match.pd: Canonicalize negate in division.
8612 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8615 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8617 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8619 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8620 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8622 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8624 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8625 non-scalar integral types.
8626 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8627 (-(A-B), -(~A)): New transformations.
8629 2017-11-07 Tom de Vries <tom@codesourcery.com>
8631 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8632 semicolon after "do {} while (0)".
8633 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8634 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8635 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8636 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8637 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8639 2017-11-07 Tom de Vries <tom@codesourcery.com>
8641 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8642 after "do {} while (0)".
8643 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8644 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8645 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8646 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8647 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8649 2017-11-07 Tom de Vries <tom@codesourcery.com>
8652 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8654 (arm_rtx_costs_internal): Add missing semicolon after
8655 HANDLE_NARROW_SHIFT_ARITH call.
8657 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8659 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8660 disable isel if it was not set explicitly.
8662 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8664 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8665 (add_type_attribute) likewise.
8667 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8669 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8670 of struct ix86_frame.
8671 (ix86_initial_elimination_offset): Likewise.
8672 (ix86_expand_split_stack_prologue): Likewise.
8674 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8676 * tree-vrp.h (enum value_range_type): Update stale comment.
8678 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8679 Alan Hayward <alan.hayward@arm.com>
8680 David Sherwood <david.sherwood@arm.com>
8682 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8683 (aarch64_expand_vec_perm_const): Take the number of units too.
8684 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8685 (aarch64_expand_vec_perm_const): Likewise.
8686 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8687 (vec_perm<mode>): Update accordingly.
8689 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8690 Alan Hayward <alan.hayward@arm.com>
8691 David Sherwood <david.sherwood@arm.com>
8693 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8694 Take the number of units too.
8695 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8696 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8697 but check for a vector mode before rather than after the call.
8698 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8699 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8700 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8701 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8702 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8703 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8704 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8705 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8706 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8707 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8708 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8709 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8710 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8711 (aarch64_sqdmull2_n<mode>): Update accordingly.
8713 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8714 Alan Hayward <alan.hayward@arm.com>
8715 David Sherwood <david.sherwood@arm.com>
8717 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8718 the number of units too.
8719 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8720 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8721 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8722 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8723 (vec_store_lanesxi<mode>): Update accordingly.
8725 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8726 Alan Hayward <alan.hayward@arm.com>
8727 David Sherwood <david.sherwood@arm.com>
8729 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8730 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8731 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8732 of units rather than the mode.
8733 * config/aarch64/iterators.md (nunits): New mode attribute.
8734 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8735 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8736 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8737 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8738 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8739 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8740 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8741 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8742 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8743 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8744 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8745 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8746 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8747 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8748 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8749 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8750 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8751 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8752 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8753 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8754 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8755 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8756 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8757 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8758 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8759 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8760 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8761 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8762 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8763 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8764 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8765 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8766 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8767 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8768 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8769 (aarch64_simd_vec_setv2di): Likewise.
8771 2017-11-06 Carl Love <cel@us.ibm.com>
8773 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8775 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8776 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8778 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8780 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8781 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8782 Add power 8 macro expansions.
8783 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8784 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8785 power 8 instructions. (VSX_XXBR): Add iterator.
8787 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8789 * config/arm/arm.md (predicable_short_it): Change default to "no",
8790 improve documentation, remove uses that are identical to the default.
8791 (enabled_for_depr_it): Rename to enabled_for_short_it.
8792 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8793 Remove default uses.
8794 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8795 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8796 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8797 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8799 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8802 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8803 float128 helper macros, which are no longer used after deleting
8804 the old 'q' built-in functions, and moving the round to odd
8805 built-in functions to being special built-in functions.
8806 (BU_FLOAT128_2): Likewise.
8807 (BU_FLOAT128_1_HW): Likewise.
8808 (BU_FLOAT128_2_HW): Likewise.
8809 (BU_FLOAT128_3_HW): Likewise.
8810 (FABSQ): Delete old 'q' built-in functions.
8811 (COPYSIGNQ): Likewise.
8812 (SQRTF128_ODD): Move round to odd built-in functions to be
8813 special built-in functions, so that we can handle
8814 -mabi=ieeelongdouble.
8815 (TRUNCF128_ODD): Likewise.
8816 (ADDF128_ODD): Likewise.
8817 (SUBF128_ODD): Likewise.
8818 (MULF128_ODD): Likewise.
8819 (DIVF128_ODD): Likewise.
8820 (FMAF128_ODD): Likewise.
8821 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8822 built-in names to 'f128'.
8823 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8824 old 'q' built-in functions, as the machine independent code for
8825 'f128' built-in functions handles this.
8826 (rs6000_expand_builtin): Add expansion for float128 round to odd
8827 functions, keying off on -mabi=ieeelongdouble of whether to use
8828 the KFmode or TFmode variant.
8829 (rs6000_init_builtins): Initialize the _Float128 round to odd
8831 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8832 _Float128 'q' built-in functions are now mapped into the new
8833 'f128' built-in functions.
8835 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8837 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8838 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8840 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8843 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8844 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8845 * toplev.c: Include "ipa-fnsummary.h".
8846 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8848 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8850 PR tree-optimization/82838
8851 * gimple-ssa-store-merging.c
8852 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8853 on a separate gimple_seq which is then appended to seq.
8855 2017-11-06 Jeff Law <law@redhat.com>
8858 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8859 (get_probe_interval): New functions.
8860 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8861 (ix86_adjust_stack_and_probe): Likewise.
8862 (output_adjust_stack_and_probe): Likewise.
8863 (ix86_emit_probe_stack_range): Likewise.
8864 (ix86_expand_prologue): Likewise.
8866 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8868 PR tree-optimization/82816
8869 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8870 if the modes of the two types are the same.
8871 (convert_plusminus_to_widen): Likewise.
8873 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8875 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8877 (usadv16qi): New define_expand.
8878 (usadv8hi): New define_expand.
8880 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8883 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8885 (inline_transform): Likewise.
8886 * predict.c (counts_to_freqs): Remove useless conditional.
8887 * profile-count.h (profile_count::apply_scale): Move sanity check.
8888 * tree-inline.c (copy_bb): Always scale.
8889 (copy_cfg_body): Likewise.
8891 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8894 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8896 (*cmp_ite0): Add enabled_for_depr_it attribute.
8897 (*cmp_ite1): Likewise.
8899 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8901 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8904 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8906 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8907 return the value by pointer.
8908 (extract_range_from_multiplicative_op_1): Update accordingly.
8909 Return as soon as an operation fails.
8911 2017-11-05 Tom de Vries <tom@codesourcery.com>
8914 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8915 (DEF_SANITIZER_BUILTIN): ... here.
8916 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8917 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8919 2017-11-05 Tom de Vries <tom@codesourcery.com>
8922 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8924 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8925 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8926 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8928 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8929 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8930 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8932 2017-11-05 Tom de Vries <tom@codesourcery.com>
8935 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8938 2017-11-04 Michael Clark <michaeljclark@mac.com>
8940 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8941 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8943 (*addsi3_extended): Likewise.
8944 (*addsi3_extended2): Likewise.
8945 (<optab>si3): Likewise.
8946 (<optab>di3): Likewise.
8947 (<optab><mode>3): Likewise.
8948 (<*optabe>si3_internal): Likewise.
8949 (zero_extendqi<SUPERQI:mode>2): Likewise.
8950 (*add<mode>hi3): Likewise.
8951 (*xor<mode>hi3): Likewise.
8952 (<optab>di3): Likewise.
8953 (*<optab>si3_extend): Likewise.
8954 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8955 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8956 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8958 2017-11-04 Andrew Waterman <andrew@sifive.com>
8960 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8961 TARGET_STRICT_ALIGN based upon -mtune argument.
8963 2017-11-04 Andrew Waterman <andrew@sifive.com>
8965 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8967 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8969 * config/i386/i386.c (choose_basereg): Use optional scratch
8970 register and add assertion.
8971 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8972 needed, and don't allocate stack.
8973 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8974 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8975 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8977 2017-11-03 Jeff Law <law@redhat.com>
8979 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8980 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8981 to probe at the start of a noreturn function.
8983 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8985 PR tree-optimization/78821
8986 * gimple-ssa-store-merging.c: Update the file comment.
8987 (MAX_STORE_ALIAS_CHECKS): Define.
8988 (struct store_operand_info): New type.
8989 (store_operand_info::store_operand_info): New constructor.
8990 (struct store_immediate_info): Add rhs_code and ops data members.
8991 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8992 and op1r arguments to the ctor, initialize corresponding data members.
8993 (struct merged_store_group): Add load_align_base and load_align
8995 (merged_store_group::merged_store_group): Initialize them.
8996 (merged_store_group::do_merge): Update them.
8997 (merged_store_group::apply_stores): Pick the constant for
8998 encode_tree_to_bitpos from one of the two operands, or skip
8999 encode_tree_to_bitpos if neither operand is a constant.
9000 (class pass_store_merging): Add process_store method decl. Remove
9001 bool argument from terminate_all_aliasing_chains method decl.
9002 (pass_store_merging::terminate_all_aliasing_chains): Remove
9003 var_offset_p argument and corresponding handling.
9004 (stmts_may_clobber_ref_p): New function.
9005 (compatible_load_p): New function.
9006 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9007 if there is overlap and rhs_code is not INTEGER_CST. For
9008 non-overlapping stores terminate group if rhs is not mergeable.
9009 (get_alias_type_for_stmts): Change first argument from
9010 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9011 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9012 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9014 (get_location_for_stmts): Change first argument from
9015 auto_vec<gimple *> & to vec<gimple *> &.
9016 (struct split_store): Remove orig_stmts data member, add orig_stores.
9017 (split_store::split_store): Create orig_stores rather than orig_stmts.
9018 (find_constituent_stmts): Renamed to ...
9019 (find_constituent_stores): ... this. Change second argument from
9020 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9021 to info structures rather than the statements.
9022 (split_group): Rename ALLOW_UNALIGNED argument to
9023 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9024 it. Adjust find_constituent_stores caller.
9025 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9026 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9027 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9028 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9029 (mem_valid_for_store_merging): New function.
9030 (handled_load): New function.
9031 (pass_store_merging::process_store): New method.
9032 (pass_store_merging::execute): Use process_store method. Adjust
9033 terminate_all_aliasing_chains caller.
9035 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9037 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9038 Return true for more constants, symbols and label references.
9039 (aarch64_valid_floating_const): Remove unused function.
9041 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9044 * config/aarch64/aarch64.c (aarch64_layout_frame):
9045 Undo forcing of LR at bottom of frame.
9047 2017-11-03 Jeff Law <law@redhat.com>
9050 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9051 for int_registers_saved.
9053 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9054 extracted from tree-ssa-dom.c.
9055 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9056 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9057 (record_equivalences_from_incoming_edge): Add additional argument
9058 to single_pred_edge_ignoring_loop_edges call.
9059 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9060 (uncprop_dom_walker::before_dom_children): Add additional argument
9061 to single_pred_edge_ignoring_loop_edges call.
9062 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9063 single_pred_edge_ignoring_loop_edges rather than open coding.
9064 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9066 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9068 * match.pd (-(-A)): Rewrite.
9070 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9072 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9073 (rs6000_emit_int_cmove): New declaration.
9074 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9075 (rs6000_emit_sISEL): Delete.
9076 (rs6000_emit_int_cmove): Make non-static.
9077 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9078 instead of rs6000_emit_sISEL.
9080 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9082 * asan.c (create_cond_insert_point): Maintain profile.
9083 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9085 * basic-block.h (struct basic_block_def): Remove frequency.
9086 (EDGE_FREQUENCY): Use to_frequency
9087 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9089 (find_traces): Update to use to_frequency.
9090 (find_traces_1_round): Likewise; use only IPA counts.
9091 (bb_to_key): Likewise.
9092 (connect_traces): Use IPA counts only.
9093 (copy_bb_p): Update to use to_frequency.
9094 (fix_up_crossing_landing_pad): Likewise.
9095 (sanitize_hot_paths): Likewise.
9096 * bt-load.c (basic_block_freq): Likewise.
9097 * cfg.c (init_flow): Set count_max to uninitialized.
9098 (check_bb_profile): Remove frequencies; check counts.
9099 (dump_bb_info): Do not dump frequencies.
9100 (update_bb_profile_for_threading): Update counts only.
9101 (scale_bbs_frequencies_int): Likewise.
9102 (MAX_SAFE_MULTIPLIER): Remove.
9103 (scale_bbs_frequencies_gcov_type): Update counts only.
9104 (scale_bbs_frequencies_profile_count): Update counts only.
9105 (scale_bbs_frequencies): Update counts only.
9106 * cfg.h (struct control_flow_graph): Add count-max.
9107 (update_bb_profile_for_threading): Update prototype.
9108 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9109 (find_many_sub_basic_blocks): Likewise.
9110 * cfgcleanup.c (try_forward_edges): Likewise.
9111 (try_crossjump_to_edge): Likewise.
9112 * cfgexpand.c (expand_gimple_cond): Likewise.
9113 (expand_gimple_tailcall): Likewise.
9114 (construct_init_block): Likewise.
9115 (construct_exit_block): Likewise.
9116 * cfghooks.c (verify_flow_info): Check consistency of counts.
9117 (dump_bb_for_graph): Do not dump frequencies.
9118 (split_block_1): Do not update frequencies.
9119 (split_edge): Do not update frequencies.
9120 (make_forwarder_block): Do not update frequencies.
9121 (duplicate_block): Do not update frequencies.
9122 (account_profile_record): Do not update frequencies.
9123 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9124 for global heuristics.
9125 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9126 (expected_loop_iterations_unbounded): Use counts only.
9127 * cfgloopmanip.c (scale_loop_profile): Simplify.
9128 (create_empty_loop_on_edge): Simplify
9130 (duplicate_loop_to_header_edge): Simplify
9131 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9132 (update_br_prob_note): Take care of removing note when profile
9134 (relink_block_chain): Do not dump frequency.
9135 (rtl_account_profile_record): Use to_frequency.
9136 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9137 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9138 (cgraph_update_edges_for_call_stmt_node): Likewise.
9139 (cgraph_edge::verify_count_and_frequency): Update.
9140 (cgraph_node::verify_node): Temporarily disable frequency verification.
9141 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9142 to_cgraph_frequency.
9143 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9144 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9146 (cgraph_node::expand_thunk): Update profile.
9147 * except.c (dw2_build_landing_pads): Do not update frequency.
9148 * final.c (compute_alignments): Use to_frequency.
9149 (dump_basic_block_info): Do not dump frequency.
9150 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9151 (dump_gimple_bb_header): Do not dump frequency.
9152 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9154 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9155 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9156 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9157 (init_before_recovery): Do not update frequency.
9158 (sched_create_recovery_edges): Do not update frequency.
9159 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9160 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9161 (ipa_cp_c_finalize): Set max_count to uninitialized.
9162 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9163 (param_change_prob): Use counts.
9164 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9166 * ipa-split.c (consider_split): Use to_frequency.
9167 (split_function): Use to_frequency.
9168 * ira-build.c (loop_compare_func): Likewise.
9169 (mark_loops_for_removal): Likewise.
9170 (mark_all_loops_for_removal): Likewise.
9171 * loop-doloop.c (doloop_modify): Do not update frequency.
9172 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9174 * lto-streamer-in.c (input_function): Update count_max.
9175 * omp-expand.c (expand_omp_taskreg): Update count_max.
9176 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9177 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9178 (maybe_hot_count_p): Use ipa counts only.
9179 (maybe_hot_bb_p): Simplify.
9180 (maybe_hot_edge_p): Simplify.
9181 (probably_never_executed): Do not take frequency argument.
9182 (probably_never_executed_bb_p): Do not pass frequency.
9183 (probably_never_executed_edge_p): Likewise.
9184 (combine_predictions_for_bb): Check that profile is nonzero.
9185 (propagate_freq): Do not set frequency.
9186 (drop_profile): Simplify.
9187 (counts_to_freqs): Simplify.
9188 (expensive_function_p): Use to_frequency.
9189 (propagate_unlikely_bbs_forward): Simplify.
9190 (determine_unlikely_bbs): Simplify.
9191 (estimate_bb_frequencies): Add hack to silence graphite issues.
9192 (compute_function_frequency): Use ipa counts.
9193 (pass_profile::execute): Update.
9194 (rebuild_frequencies): Use counts only.
9195 (force_edge_cold): Use counts only.
9196 * profile-count.c (profile_count::dump): Dump new count types.
9197 (profile_count::differs_from_p): Check compatiblity.
9198 (profile_count::to_frequency): New function.
9199 (profile_count::to_cgraph_frequency): New function.
9200 * profile-count.h (struct function): Declare.
9201 (enum profile_quality): Add profile_guessed_local and
9202 profile_guessed_global0.
9203 (class profile_proability): Decrease number of bits to 29;
9204 update from_reg_br_prob_note and to_reg_br_prob_note.
9205 (class profile_count: Update comment; decrease number of bits
9206 to 61. Check compatibility.
9207 (profile_count::compatible_p): New private member function.
9208 (profile_count::ipa_p): New member function.
9209 (profile_count::operator<): Handle global zero correctly.
9210 (profile_count::operator>): Handle global zero correctly.
9211 (profile_count::operator<=): Handle global zero correctly.
9212 (profile_count::operator>=): Handle global zero correctly.
9213 (profile_count::nonzero_p): New member function.
9214 (profile_count::force_nonzero): New member function.
9215 (profile_count::max): New member function.
9216 (profile_count::apply_scale): Handle IPA scalling.
9217 (profile_count::guessed_local): New member function.
9218 (profile_count::global0): New member function.
9219 (profile_count::ipa): New member function.
9220 (profile_count::to_frequency): Declare.
9221 (profile_count::to_cgraph_frequency): Declare.
9222 * profile.c (OVERLAP_BASE): Delete.
9223 (compute_frequency_overlap): Delete.
9224 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9225 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9226 * sched-ebb.c (rank): Use counts only.
9227 * shrink-wrap.c (handle_simple_exit): Use counts only.
9228 (try_shrink_wrapping): Use counts only.
9229 (place_prologue_for_one_component): Use counts only.
9230 * tracer.c (find_best_predecessor): Use to_frequency.
9231 (find_trace): Use to_frequency.
9232 (tail_duplicate): Use to_frequency.
9233 * trans-mem.c (expand_transaction): Do not update frequency.
9234 * tree-call-cdce.c: Do not update frequency.
9235 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9236 (gimple_merge_blocks): Likewise.
9237 (gimple_split_edge): Likewise.
9238 (gimple_duplicate_sese_region): Likewise.
9239 (gimple_duplicate_sese_tail): Likewise.
9240 (move_sese_region_to_fn): Likewise.
9241 (gimple_account_profile_record): Likewise.
9242 (insert_cond_bb): Likewise.
9243 * tree-complex.c (expand_complex_div_wide): Likewise.
9244 * tree-eh.c (lower_resx): Update profile.
9245 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9247 (initialize_cfun): Do not initialize frequencies
9248 (freqs_to_counts): Delete.
9249 (copy_cfg_body): Ignore count parameter.
9250 (copy_body): Update.
9251 (expand_call_inline): Update count_max.
9252 (optimize_inline_calls): Update count_max.
9253 (tree_function_versioning): Update count_max.
9254 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9255 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9257 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9258 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9259 (try_peel_loop): Likewise.
9260 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9262 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9263 (tree_transform_and_unroll_loop): Do not use frequencies
9264 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9265 Use reliable prediction only.
9266 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9267 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9268 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9269 probability scaling.
9270 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9271 not update frequency
9272 (any_remaining_duplicated_blocks): Likewise.
9273 (update_profile): Likewise.
9274 (estimated_freqs_path): Delete.
9275 (freqs_to_counts_path): Delete.
9276 (clear_counts_path): Delete.
9277 (ssa_fix_duplicate_block_edges): Likewise.
9278 (duplicate_thread_path): Likewise.
9279 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9280 * tree-tailcall.c (decrease_profile): Do not update frequency.
9281 (eliminate_tail_call): Likewise.
9282 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9283 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9284 (optimize_mask_stores): Likewise.
9285 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9286 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9287 (ubsan_expand_ptr_ifn): Update profile.
9288 * value-prof.c (gimple_ic): Simplify.
9289 * value-prof.h (gimple_ic): Update prototype.
9290 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9291 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9293 (want_inline_self_recursive_call_p): Likewise.
9294 (resolve_noninline_speculation): Only cummulate defined counts.
9295 (inline_small_functions): Use nonzero_p.
9296 (ipa_inline): Do not access freed node.
9298 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9300 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9301 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9303 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9305 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9306 non-legitimate address.
9308 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9310 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9311 (*lt0_<mode>di, *lt0_<mode>si): New.
9313 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9315 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9316 TARGET_PAIRED_FLOAT.
9318 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9319 Jim Wilson <jim.wilson@linaro.org>
9321 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9322 * config/aarch64/aarch64-tune.md: Regenerated.
9323 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9324 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9326 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9328 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9329 (arc_expand_prologue): Restore blink for millicode.
9330 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9332 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9335 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9336 gen_vec_duplicate after forcing the scalar into a register.
9338 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9340 * combine (try_combine): Print the insns input to try_combine to the
9343 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9346 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9347 Remove second argument from aarch64_process_target_attr call.
9348 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9350 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9352 (aarch64_handle_attr_arch): Remove second argument.
9353 (aarch64_handle_attr_cpu): Ditto.
9354 (aarch64_handle_attr_tune): Ditto.
9355 (aarch64_handle_attr_isa_flags): Ditto.
9356 (aarch64_process_one_target_attr): Ditto.
9357 (aarch64_process_target_attr): Ditto.
9358 (aarch64_option_valid_attribute_p): Remove second argument.
9359 on aarch64_process_target_attr call.
9361 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9363 * diagnostic.c: Include "selftest-diagnostic.h".
9364 (selftest::assert_location_text): New function.
9365 (selftest::test_diagnostic_get_location_text): New function.
9366 (selftest::diagnostic_c_tests): Call it.
9368 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9370 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9371 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9372 (class selftest::test_diagnostic_context): Move to...
9373 * selftest-diagnostic.c: New file.
9374 * selftest-diagnostic.h: New file.
9376 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9378 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9379 offset range for FT32B.
9380 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9381 * config/ft32/ft32.md: Add TARGET_NOPM.
9382 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9383 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9385 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9387 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9389 2017-11-02 Jeff Law <law@redhat.com>
9391 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9392 virtual keyword on FINAL OVERRIDE members.
9394 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9395 virtuals together. Add virtual destructor.
9396 (substitute_and_fold_engine): Similarly.
9398 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9400 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9402 2017-11-02 Richard Biener <rguenther@suse.de>
9404 PR tree-optimization/82795
9405 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9407 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9409 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9411 (gcc_GAS_CHECK_FEATURE): Remove.
9412 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9413 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9414 * configure: Regenerate.
9416 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9418 * config/arc/arc.c (hwloop_optimize): Account for empty
9421 2017-11-02 Richard Biener <rguenther@suse.de>
9424 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9425 Truncate ARRAY_REF index and element size.
9427 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9429 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9431 2017-11-01 Jeff Law <law@redhat.com>
9433 * tree-ssa-ccp.c (ccp_folder): New class derived from
9434 substitute_and_fold_engine.
9435 (ccp_folder::get_value): New member function.
9436 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9437 (ccp_fold_stmt): Remove prototype.
9438 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9439 * tree-ssa-copy.c (copy_folder): New class derived from
9440 substitute_and_fold_engine.
9441 (copy_folder::get_value): Renamed from get_value.
9442 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9443 * tree-vrp.c (vrp_folder): New class derived from
9444 substitute_and_fold_engine.
9445 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9446 (vrp_folder::get_value): New member function.
9447 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9448 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9449 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9450 provide a class interface to folder/substitute routines.
9451 (ssa_prop_fold_stmt_fn): Remove typedef.
9452 (ssa_prop_get_value_fn): Likewise.
9453 (subsitute_and_fold): Remove prototype.
9454 (replace_uses_in): Likewise.
9455 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9456 Renamed from replace_uses_in. Call the virtual member function
9457 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9458 (substitute_and_fold_dom_walker): Remove initialization of
9459 data member entries for calbacks. Add substitute_and_fold_engine
9460 member and initialize it.
9461 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9462 member functions for get_value, replace_phi_args_in c
9463 replace_uses_in, and fold_stmt calls.
9464 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9465 substitute_and_fold. Remove assert. Update ctor call.
9467 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9468 (ssa_prop_visit_phi_fn): Likewise.
9469 (class ssa_propagation_engine): New class to provide an interface
9471 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9473 (ssa_prop_visit_phi): Likewise.
9474 (ssa_propagation_engine::simulate_stmt): Moved into class.
9475 Call visit_phi/visit_stmt from the class rather than via
9476 file scoped static variables.
9477 (ssa_propagation_engine::simulate_block): Moved into class.
9478 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9479 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9480 set file scoped statics for the visit_stmt/visit_phi callbacks.
9481 * tree-complex.c (complex_propagate): New class derived from
9482 ssa_propagation_engine.
9483 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9484 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9485 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9487 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9488 ssa_propagation_engine.
9489 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9490 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9491 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9492 * tree-ssa-copy.c (copy_prop): New class derived from
9493 ssa_propagation_engine.
9494 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9495 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9496 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9497 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9498 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9499 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9500 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9502 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9504 PR rtl-optimization/82778
9505 PR rtl-optimization/82597
9506 * compare-elim.c (struct comparison): Add in_a_setter field.
9507 (find_comparison_dom_walker::before_dom_children): Remove killed
9508 bitmap and df_simulate_find_defs call, instead walk the defs.
9509 Compute last_setter and initialize in_a_setter. Merge definitions
9510 with first initialization for a few variables.
9511 (try_validate_parallel): Use insn_invalid_p instead of
9512 recog_memoized. Return insn rather than just the pattern.
9513 (try_merge_compare): Fix up comment. Don't uselessly test if
9514 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9516 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9519 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9520 Alan Hayward <alan.hayward@arm.com>
9521 David Sherwood <david.sherwood@arm.com>
9523 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9524 aarch64_hard_regno_nregs to get the number of registers
9527 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9528 Alan Hayward <alan.hayward@arm.com>
9529 David Sherwood <david.sherwood@arm.com>
9531 * config/aarch64/constraints.md (Upl): Rename to...
9533 * config/aarch64/aarch64.md
9534 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9537 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9538 Alan Hayward <alan.hayward@arm.com>
9539 David Sherwood <david.sherwood@arm.com>
9541 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9542 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9545 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9546 Alan Hayward <alan.hayward@arm.com>
9547 David Sherwood <david.sherwood@arm.com>
9549 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9550 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9551 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9553 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9554 of a single element.
9555 * config/aarch64/iterators.md: Add a comment above the permute
9556 unspecs to say that they are generated directly by
9557 aarch64_expand_vec_perm_const.
9558 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9560 2017-11-01 Nathan Sidwell <nathan@acm.org>
9562 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9564 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9566 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9567 and medany code models, and describe what they do.
9569 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9571 Revert accidental duplicate:
9573 * combine.c (can_change_dest_mode): Reject changes in
9574 REGMODE_NATURAL_SIZE.
9576 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9578 PR rtl-optimization/64682
9579 PR rtl-optimization/69567
9580 PR rtl-optimization/69737
9581 PR rtl-optimization/82683
9582 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9583 register mentioned in the note, drop the note, unless it came from I3,
9584 in which case it should go to I3 again.
9586 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9588 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9589 and return false if not.
9590 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9592 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9594 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9595 range is known to be empty.
9597 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9598 Alan Hayward <alan.hayward@arm.com>
9599 David Sherwood <david.sherwood@arm.com>
9601 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9602 and CONST_VECTOR_NUNITS instead of computing the number of units from
9603 the byte sizes of the vector and element.
9604 (simplify_binary_operation_1): Likewise.
9605 (simplify_const_binary_operation): Likewise.
9606 (simplify_ternary_operation): Likewise.
9608 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9609 Alan Hayward <alan.hayward@arm.com>
9610 David Sherwood <david.sherwood@arm.com>
9612 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9613 (int_mem_offset): ...this new function.
9614 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9615 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9616 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9619 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9620 Alan Hayward <alan.hayward@arm.com>
9621 David Sherwood <david.sherwood@arm.com>
9623 * lower-subreg.c (interesting_mode_p): New function.
9624 (compute_costs, find_decomposable_subregs, decompose_register)
9625 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9626 (resolve_clobber, dump_choices): Use it.
9628 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9629 Alan Hayward <alan.hayward@arm.com>
9630 David Sherwood <david.sherwood@arm.com>
9632 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9634 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9635 Alan Hayward <alan.hayward@arm.com>
9636 David Sherwood <david.sherwood@arm.com>
9638 * alias.c (find_base_value, find_base_term): Only process integer
9639 truncations. Check the precision rather than the size.
9641 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9642 Alan Hayward <alan.hayward@arm.com>
9643 David Sherwood <david.sherwood@arm.com>
9645 * machmode.h (is_narrower_int_mode): New function
9646 * optabs.c (expand_float, expand_fix): Use it.
9647 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9649 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9650 Alan Hayward <alan.hayward@arm.com>
9651 David Sherwood <david.sherwood@arm.com>
9653 * rtl.h (narrower_subreg_mode): New function.
9654 * ira-color.c (update_costs_from_allocno): Use it.
9656 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9657 Alan Hayward <alan.hayward@arm.com>
9658 David Sherwood <david.sherwood@arm.com>
9660 * optabs-query.h (convert_optab_p): New function, split out from...
9661 (convert_optab_handler): ...here.
9662 (widening_optab_handler): Delete.
9663 (find_widening_optab_handler): Remove permit_non_widening parameter.
9664 (find_widening_optab_handler_and_mode): Likewise. Provide an
9665 override that operates on mode class wrappers.
9666 * optabs-query.c (widening_optab_handler): Delete.
9667 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9668 parameter. Assert that the two modes are the same class and that
9669 the "from" mode is narrower than the "to" mode. Use
9670 convert_optab_handler instead of widening_optab_handler.
9671 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9672 instead of widening_optab_handler.
9673 * expr.c (expand_expr_real_2): Update calls to
9674 find_widening_optab_handler.
9675 * optabs.c (expand_widen_pattern_expr): Likewise.
9676 (expand_binop_directly): Take the insn_code as a parameter.
9677 (expand_binop): Only call find_widening_optab_handler for
9678 conversion optabs; use optab_handler otherwise. Update calls
9679 to find_widening_optab_handler and expand_binop_directly.
9680 Use convert_optab_handler instead of widening_optab_handler.
9681 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9682 find_widening_optab_handler and use scalar_mode rather than
9684 (convert_plusminus_to_widen): Likewise.
9686 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9687 Alan Hayward <alan.hayward@arm.com>
9688 David Sherwood <david.sherwood@arm.com>
9690 * machmode.h (fixed_size_mode): New class.
9691 * rtl.h (get_pool_mode): Return fixed_size_mode.
9692 * gengtype.c (main): Add fixed_size_mode.
9693 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9694 (get_raw_arg_mode): Likewise.
9695 * doc/tm.texi: Regenerate.
9696 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9697 * targhooks.c (default_get_reg_raw_mode): Likewise.
9698 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9699 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9700 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9701 (msp430_get_raw_result_mode): Likewise.
9702 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9703 * dbxout.c (dbxout_parms): Require fixed-size modes.
9704 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9705 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9706 * omp-low.c (lower_oacc_reductions): Likewise.
9707 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9708 (simplify_subreg): Update accordingly.
9709 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9710 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9711 that aren't fixed-size.
9712 (get_pool_mode): Return a fixed_size_mode.
9713 (output_constant_pool_2): Take a fixed_size_mode.
9715 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9716 Alan Hayward <alan.hayward@arm.com>
9717 David Sherwood <david.sherwood@arm.com>
9719 * doc/rtl.texi (vec_series): Document.
9720 (const): Say that the operand can be a vec_series.
9721 * rtl.def (VEC_SERIES): New rtx code.
9722 * rtl.h (const_vec_series_p_1): Declare.
9723 (const_vec_series_p): New function.
9724 * emit-rtl.h (gen_const_vec_series): Declare.
9725 (gen_vec_series): Likewise.
9726 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9727 (gen_vec_series): Likewise.
9728 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9729 * simplify-rtx.c (simplify_unary_operation): Handle negations
9731 (simplify_binary_operation_series): New function.
9732 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9733 (test_vector_ops_series): New function.
9734 (test_vector_ops): Call it.
9735 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9736 gen_const_vec_series.
9737 (altivec_lvsr): Likewise.
9738 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9740 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9741 Alan Hayward <alan.hayward@arm.com>
9742 David Sherwood <david.sherwood@arm.com>
9744 * doc/rtl.texi (const): Update description of address constants.
9745 Say that vector constants are allowed too.
9746 * common.md (E, F): Use CONSTANT_P instead of checking for
9748 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9749 checking for CONST_VECTOR.
9750 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9752 * expr.c (expand_expr_real_2): Check for vector modes instead
9753 of checking for CONST_VECTOR.
9754 * rtl.h (const_vec_p): New function.
9755 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9756 (unwrap_const_vec_duplicate): Handle them here too.
9758 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9759 David Malcolm <dmalcolm@redhat.com>
9760 Alan Hayward <alan.hayward@arm.com>
9761 David Sherwood <david.sherwood@arm.com>
9763 * rtl.h (vec_duplicate_p): New function.
9764 * selftest-rtl.c (assert_rtx_eq_at): New function.
9765 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9766 (assert_rtx_eq_at): Declare.
9767 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9768 * selftest-run-tests.c (selftest::run_tests): Call it.
9769 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9770 (simplify_unary_operation_1): Recursively handle vector duplicates.
9771 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9773 (simplify_subreg): Handle subregs of vector duplicates.
9774 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9775 (selftest::simplify_rtx_c_tests): New functions.
9777 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9778 Alan Hayward <alan.hayward@arm.com>
9779 David Sherwood <david.sherwood@arm.com>
9781 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9782 (gen_vec_duplicate): Likewise.
9783 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9785 (gen_const_vector): ...here.
9786 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9787 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9788 whose elements are all equal.
9789 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9790 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9791 (simplify_relational_operation): Likewise.
9792 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9794 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9795 (aarch64_expand_vector_init): Likewise.
9796 * config/arm/arm.c (neon_vdup_constant): Likewise.
9797 (neon_expand_vector_init): Likewise.
9798 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9799 (arm_block_set_unaligned_vect): Likewise.
9800 (arm_block_set_aligned_vect): Likewise.
9801 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9802 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9803 (expand_vec_perm_even_odd_pack): Likewise.
9804 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9805 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9806 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9807 gen_const_vec_duplicate.
9808 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9809 * config/mips/mips.c (mips_gen_const_int_vector): Use
9810 gen_const_vec_duplicate.
9811 (mips_expand_vector_init): Use CONST0_RTX.
9812 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9813 (define_split): Use gen_const_vec_duplicate.
9814 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9815 (define_split): Use gen_const_vec_duplicate.
9816 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9817 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9818 * config/spu/spu.c (spu_const): Likewise.
9820 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9821 Alan Hayward <alan.hayward@arm.com>
9822 David Sherwood <david.sherwood@arm.com>
9824 * combine.c (can_change_dest_mode): Reject changes in
9825 REGMODE_NATURAL_SIZE.
9827 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9829 * configure.ac (--enable-libssp): New.
9830 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9831 trying to determine target-specific default. Adjust indentation.
9832 * configure: Regenerated.
9833 * doc/install.texi (Configuration): Expand --disable-libssp
9836 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9838 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9841 2017-10-31 Martin Jambor <mjambor@suse.cz>
9844 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9846 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9848 * auto-profile.c (autofdo_source_profile::read): Use
9849 UNKNOWN_LOCATION rather than 0.
9850 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9851 (warning_at): ...this overload.
9852 (warning_at_rich_loc_n): Rename to...
9853 (warning_n): ...this overload.
9854 (error_at_rich_loc): Rename to...
9855 (error_at): ...this overload.
9856 (pedwarn_at_rich_loc): Rename to...
9857 (pedwarn): ...this overload.
9858 (permerror_at_rich_loc): Rename to...
9859 (permerror): ...this overload.
9860 (inform_at_rich_loc): Rename to...
9861 (inform): ...this overload.
9862 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9863 (diagnostic_n_impl_richloc): Rename to...
9864 (diagnostic_n_impl): ...this rich_location *-based decl.
9865 (inform_at_rich_loc): Rename to...
9866 (inform): ...this, and add an assertion.
9867 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9868 (warning_at_rich_loc): Rename to...
9869 (warning_at): ...this, and add an assertion.
9870 (warning_at_rich_loc_n): Rename to...
9871 (warning_n): ...this, and add an assertion.
9872 (warning_n): Update location_t-based implementation for removal of
9873 location_t-based diagnostic_n_impl.
9874 (pedwarn_at_rich_loc): Rename to...
9875 (pedwarn): ...this, and add an assertion.
9876 (permerror_at_rich_loc): Rename to...
9877 (permerror): ...this, and add an assertion.
9878 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9879 (error_at_rich_loc): Rename to...
9880 (error_at): ...this, and add an assertion.
9881 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9882 (driver::do_spec_on_infiles): Likewise.
9883 * substring-locations.c (format_warning_va): Update for renaming
9884 of inform_at_rich_loc.
9886 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9888 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9889 _Float<N>X built-in functions so that the variant without the
9890 "__builtin_" prefix is only enabled for the GNU C and Objective C
9891 languages when they are in non-strict ANSI/ISO mode.
9892 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9893 * target.def (floatn_builtin_p): Add a target hook to control
9894 whether _Float<N> and _Float<N>X built-in functions without the
9895 "__builtin_" prefix are enabled, and return true for C and
9896 Objective C in the default hook. Include langhooks.h in
9898 * targhooks.h (default_floatn_builtin_p): Likewise.
9899 * targhooks.c (default_floatn_builtin_p): Likewise.
9900 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9901 floatn_builtin_p target hook.
9902 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9904 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9905 Eric Botcazou <ebotcazou@adacore.com>
9907 PR rtl-optimization/81803
9908 * lra-constraints.c (curr_insn_transform): Also reload the whole
9909 register for a strict subreg no wider than a word if this is for
9910 a WORD_REGISTER_OPERATIONS target.
9912 2017-10-31 Jason Merrill <jason@redhat.com>
9914 * gdbinit.in: Skip over inlines from timevar.h.
9916 2017-10-31 Martin Liska <mliska@suse.cz>
9918 * doc/gcov.texi: Document new option.
9919 * gcov.c (print_usage): Likewise print it.
9920 (process_args): Support the argument.
9921 (format_count): New function.
9922 (format_gcov): Use the function.
9924 2017-10-31 Martin Liska <mliska@suse.cz>
9926 * gcov.c (struct name_map): do not use typedef.
9927 Define operator== and operator<.
9928 (name_search): Remove.
9929 (name_sort): Remove.
9930 (main): Do not allocate names.
9931 (process_file): Add vertical space.
9932 (generate_results): Use std::find.
9933 (release_structures): Do not release memory.
9934 (find_source): Use std::find.
9936 2017-10-31 Martin Liska <mliska@suse.cz>
9938 * gcov.c (struct line_info): Remove it's typedef.
9939 (line_info::line_info): Add proper ctor.
9940 (line_info::has_block): Do not use a typedef.
9941 (struct source_info): Do not use typedef.
9942 (circuit): Likewise.
9943 (get_cycles_count): Likewise.
9944 (output_intermediate_file): Iterate via vector iterator.
9945 (add_line_counts): Use std::vector methods.
9946 (accumulate_line_counts): Likewise.
9947 (output_lines): Likewise.
9949 2017-10-31 Martin Liska <mliska@suse.cz>
9951 * gcov.c (struct source_info): Remove typedef.
9952 (source_info::source_info): Add proper ctor.
9953 (accumulate_line_counts): Use struct, not it's typedef.
9954 (output_gcov_file): Likewise.
9955 (output_lines): Likewise.
9956 (main): Do not allocate an array.
9957 (output_intermediate_file): Use size of vector container.
9958 (process_file): Resize the vector.
9959 (generate_results): Do not preallocate, use newly added vector
9961 (release_structures): Do not release sources.
9962 (find_source): Use vector methods.
9963 (add_line_counts): Do not use typedef.
9965 2017-10-31 Martin Liska <mliska@suse.cz>
9967 * doc/gcov.texi: Document that.
9968 * gcov.c (add_line_counts): Mark lines with a non-executed
9970 (output_line_beginning): Handle such lines.
9971 (output_lines): Pass new argument.
9972 (output_intermediate_file): Print it in intermediate format.
9974 2017-10-31 Martin Liska <mliska@suse.cz>
9976 * color-macros.h: New file.
9977 * diagnostic-color.c: Factor out color related to macros to
9979 * doc/gcov.texi: Document -k option.
9980 * gcov.c (INCLUDE_STRING): Include string.h.
9981 (print_usage): Add -k option.
9982 (process_args): Parse it.
9983 (pad_count_string): New function.
9984 (output_line_beginning): Likewise.
9985 (DEFAULT_LINE_START): New macro.
9986 (output_lines): Support color output.
9988 2017-10-31 Martin Liska <mliska@suse.cz>
9990 PR gcov-profile/82633
9991 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9992 their interaction with GCOV infrastructure.
9993 * configure.ac: Add -fkeep-{inline,static}-functions to
9995 * configure: Regenerate.
9997 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10000 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10002 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10005 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10006 into a register if it does not fit into an immediate offset field.
10008 2017-10-31 Olivier Hainque <hainque@adacore.com>
10010 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10012 2017-10-31 Julia Koval <julia.koval@intel.com>
10014 * config.gcc: Add gfniintrin.h.
10015 * config/i386/gfniintrin.h: New.
10016 * config/i386/i386-builtin-types.def
10017 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10018 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10019 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10020 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10021 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10022 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10023 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10024 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10025 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10026 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10027 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10028 * config/i386/immintrin.h: Include gfniintrin.h.
10029 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10031 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10033 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10035 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10037 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10038 (arm_ashldi3_1bit): Remove pattern.
10039 (ashrdi3): Remove shift by 1 expansion.
10040 (arm_ashrdi3_1bit): Remove pattern.
10041 (lshrdi3): Remove shift by 1 expansion.
10042 (arm_lshrdi3_1bit): Remove pattern.
10043 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10044 cost of ashldi3 by 1.
10045 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10046 (<shift>di3_neon): Likewise.
10048 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10050 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10051 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10052 and (*aarch64_simd_mov<VQ:mode>).
10053 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10054 pattern alternative.
10055 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10056 attributes to match pattern alternative.
10058 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10060 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10061 * config/rs6000/emmintrin.h: New file.
10062 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10064 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10066 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10067 * (movdi_vfp_cortexa8): Remove pattern.
10069 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10071 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10073 (Specific, alpha*-dec-osf5.1): Remove.
10074 (Specific, mips-sgi-irix5): Remove.
10075 (Specific, mips-sgi-irix6): Remove.
10077 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10079 PR middle-end/22141
10080 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10081 arguments to clear_bit_region_be.
10083 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10085 * gimplify.c: Include memmodel.h.
10087 2017-10-30 Martin Jambor <mjambor@suse.cz>
10089 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10090 condition whether loop should be executed at all.
10092 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10094 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10095 gimple folding of vec_madd() intrinsics.
10096 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10097 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10098 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10100 2017-10-30 Richard Biener <rguenther@suse.de>
10102 PR tree-optimization/82762
10104 2017-10-23 Richard Biener <rguenther@suse.de>
10106 PR tree-optimization/82129
10108 2017-08-01 Richard Biener <rguenther@suse.de>
10110 PR tree-optimization/81181
10111 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10112 (compute_antic): ... end of iteration here.
10114 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10116 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10117 -std=iso9899:2017 and -std=gnu17.
10118 * doc/standards.texi (C Language): Document C17 support.
10119 * doc/cpp.texi (Overview): Mention -std=c17.
10120 (Standard Predefined Macros): Document C11 and C17 values of
10121 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10122 * doc/extend.texi (Inline): Do not list individual options for
10123 standards newer than C99.
10124 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10126 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10129 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10131 * asan.c (asan_finish_file): Align asan globals array by shadow
10134 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10136 PR middle-end/22141
10137 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10138 (struct store_immediate_info): Add bitregion_start and bitregion_end
10140 (store_immediate_info::store_immediate_info): Add brs and bre
10141 arguments and initialize bitregion_{start,end} from those.
10142 (struct merged_store_group): Add bitregion_start, bitregion_end,
10143 align_base and mask fields. Drop unnecessary struct keyword from
10144 struct store_immediate_info. Add do_merge method.
10145 (clear_bit_region_be): Use memset instead of loop storing zeros.
10146 (merged_store_group::do_merge): New method.
10147 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10148 stores as long as the surrounding bitregions have no gaps.
10149 (merged_store_group::merge_overlapping): Use do_merge.
10150 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10151 is byte aligned, rather than requiring that start and width are
10152 byte aligned. Drop unnecessary struct keyword from
10153 struct store_immediate_info. Allocate and populate also mask array.
10154 Make start of the arrays relative to bitregion_start rather than
10155 start and size them according to bitregion_{end,start} difference.
10156 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10157 struct store_immediate_info.
10158 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10159 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10160 struct keyword from struct store_immediate_info.
10161 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10162 between stores as long as the surrounding bitregions have no gaps.
10164 (struct split_store): Add orig non-static data member.
10165 (split_store::split_store): Initialize orig to false.
10166 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10167 if there is exactly a single original stmt. Change stmts argument
10168 to pointer from reference, if NULL, don't push anything to it. Add
10169 first argument, use it to optimize skipping over orig stmts that
10170 are known to be before bitpos already. Simplify.
10171 (split_group): Return unsigned int count how many stores are or
10172 would be needed rather than a bool. Add allow_unaligned argument.
10173 Change split_stores argument from reference to pointer, if NULL,
10174 only do a dry run computing how many stores would be produced.
10175 Rewritten algorithm to use both alignment and misalign if
10176 !allow_unaligned and handle bitfield stores with gaps.
10177 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10178 from bitregion_start instead of start. Compute allow_unaligned
10179 here, if true, do 2 split_group dry runs to compute which one
10180 produces fewer stores and prefer aligned if equal. Punt if
10181 new count is bigger or equal than original before emitting any
10182 statements, rather than during that. Remove no longer needed
10183 new_ssa_names tracking. Replace num_stmts with
10184 split_stores.length (). Use 32-bit stack allocated entries
10185 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10186 if possible. Handle bitfields with gaps.
10187 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10188 Compute bitregion_{start,end} for the stores and construct
10189 store_immediate_info with that. Formatting fixes.
10191 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10194 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10195 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10197 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10199 * gimplify.c: Include tm_p.h.
10201 * common.opt (gcoff): Re-add as ignored option.
10202 (gcoff1, gcoff2, gcoff3): Likewise.
10204 * Makefile.in (OBJS): Delete sdbout.o.
10205 (GTFILES): Delete $(srcdir)/sdbout.c.
10206 * debug.h: Delete sdb_debug_hooks.
10207 * final.c: Delete sdbout.h include.
10208 (final_scan_insn): Delete SDB_DEBUG check.
10209 (rest_of_clean_state): Likewise.
10210 * output.h: Delete sdb_begin_function_line.
10211 * sdbout.c: Delete.
10212 * sdbout.h: Delete.
10213 * toplev.c: Delete sdbout.h include.
10214 (process_options): Delete SDB_DEBUG check.
10215 * tree-core.h (tree_type_common): Delete pointer field of
10217 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10218 TYPE_SYMTAB_POINTER.
10219 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10220 (TYPE_SYMTAB_IS_POINTER): Delete.
10221 (TYPE_SYMTAB_IS_DIE): Renumber.
10222 * xcoffout.c: Refer to former sdbout.c file.
10223 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10225 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10226 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10227 (Debugging Options): Delete -gcoff.
10228 (-gstabs): Delete SDB reference.
10230 (-gcoff@var{level}): Delete.
10231 * doc/passes.texi (Debugging information output): Delete SDB and
10232 sdbout.c references.
10233 * doc/tm.texi: Regenerate.
10234 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10235 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10237 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10238 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10239 (SDB_DEBUGGING_INFO): Delete.
10240 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10241 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10242 * target.def (output_source_filename): Delete COFF reference.
10244 * common.opt (gcoff): Delete.
10245 (gxcoff+): Update Negative chain.
10246 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10248 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10249 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10250 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10251 and SDB references.
10252 (expand_function_start): Change sdb reference to past tense.
10253 (expand_function_end): Change sdb reference to past tense.
10254 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10255 * opts.c (debug_type_names): Delete coff entry.
10256 (common_handle_option): Delete OPT_gcoff case.
10257 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10259 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10260 * config/cris/cris.h: Delete SDB reference in comment.
10261 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10262 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10263 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10264 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10266 (ix86_expand_prologue): Likewise.
10267 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10268 * config/ia64/ia64.h: Likewise.
10269 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10270 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10272 * config/mmix/mmix.h: Likewise.
10273 * config/nds32/nds32.c: Likewise.
10274 * config/stormy/storym16.h: Likewise.
10275 * config/visium/visium.h: Likewise.
10276 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10278 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10280 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10281 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10282 FRAME_POINTER_REGNUM point at high end of local var area.
10284 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10286 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10287 Move comment around. Do not reset best_edge for a copiable
10288 destination if the copy would cause a partition change.
10289 (better_edge_p): Remove redundant check.
10291 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10293 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10295 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10297 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10298 for math functions that have _Float<N> and _Float<N>X variants.
10299 (mathfn_built_in_2): Add support for math functions that have
10300 _Float<N> and _Float<N>X variants.
10301 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10302 (expand_builtin_mathfn_ternary): Add support for fma with
10303 _Float<N> and _Float<N>X variants.
10304 (expand_builtin): Likewise.
10305 (fold_builtin_3): Likewise.
10306 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10307 create math function _Float<N> and _Float<N>X variants as external
10309 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10310 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10311 the __builtin_ prefix and if not strict ansi, without the prefix.
10312 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10313 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10314 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10315 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10316 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10317 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10318 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10319 function signatures for fma _Float<N> and _Float<N>X variants.
10320 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10321 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10322 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10323 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10324 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10325 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10326 * gencfn-macros.c (print_case_cfn): Add support for math functions
10327 that have _Float<N> and _Float<N>X variants.
10328 (print_define_operator_list): Likewise.
10329 (fltfn_suffixes): Likewise.
10331 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10332 for math functions that have _Float<N> and _Float<N>X variants.
10333 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10334 and _Float<N>X variants.
10335 (COPYSIGN): Likewise.
10338 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10339 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10341 (integer_valued_read_call_p): Likewise.
10342 * fold-const-call.c (fold_const_call_ss): Likewise.
10343 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10344 _Float<N> and _Float<N>X variants.
10345 (fold_const_call_ssss): Add support for fma _Float<N> and
10346 _Float<N>X variants.
10347 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10348 support for copysign and fma _Float<N> and _Float<N>X variants.
10349 (backprop::process_builtin_call_use): Likewise.
10350 * tree-call-cdce.c (can_test_argument_range); Add support for
10351 sqrt _Float<N> and _Float<N>X variants.
10352 (edom_only_function): Likewise.
10353 (get_no_error_domain): Likewise.
10354 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10355 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10356 copysign _Float<N> and _Float<N>X variants.
10357 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10358 handled by machine independent code.
10359 (FMAF128): Likewise.
10360 * doc/cpp.texi (Common Predefined Macros): Document defining
10361 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10362 fma _Float<N> and _Float<N>X variants.
10364 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10367 * config/i386/i386-modes.def (CCFPU): Remove definition.
10368 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10369 (ix86_cc_modes_compatible): Ditto.
10370 (ix86_expand_carry_flag_compare): Ditto.
10371 (ix86_expand_int_movcc): Ditto.
10372 (ix86_expand_int_addcc): Ditto.
10373 (ix86_reverse_condition): Ditto.
10374 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10375 Return true/false for unordered/ordered fp comparisons.
10376 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10377 (ix86_prepare_fp_compare_args): Update for rename.
10378 (ix86_expand_fp_compare): Update for rename. Generate unordered
10379 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10380 (ix86_expand_sse_compare_and_jump): Ditto.
10381 * config/i386/predicates.md (fcmov_comparison_operator):
10382 Remove CCFPU mode handling.
10383 (ix86_comparison_operator): Ditto.
10384 (ix86_carry_flag_operator): Ditto.
10385 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10386 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10387 (*cmpu<mode>_cc_i387): Ditto.
10388 (FPCMP): Remove mode iterator.
10389 (unord): Remove mode attribute.
10390 (unord_subst): New define_subst transformation
10391 (unord): New define_subst attribute.
10392 (unordered): Ditto.
10393 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10394 (*cmpi<unord>xf_i387): Ditto.
10395 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10396 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10397 using unord_subst transformation.
10398 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10399 (round_saeonly): Also handle CCFP mode.
10400 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10401 Remove UNSPEC_SAHF unspec handling.
10403 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10405 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10407 2017-10-27 Jeff Law <law@redhat.com>
10409 * gimple-ssa-sprintf.c: Include domwalk.h.
10410 (class sprintf_dom_walker): New class, derived from dom_walker.
10411 (sprintf_dom_walker::before_dom_children): New function.
10412 (struct call_info): Moved into sprintf_dom_walker class
10413 (compute_formath_length, handle_gimple_call): Likewise.
10414 (sprintf_length::execute): Call the dominator walker rather
10415 than walking the statements.
10417 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10418 gimple statement locations.
10419 (check_array_bounds): Corresponding changes. Get the statement's
10420 location directly from wi->stmt.
10422 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10425 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10427 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10429 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10430 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10432 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10435 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10436 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10437 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10438 maybe_get_pool_constant.
10439 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10442 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10444 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10445 2.26 caveat. Update gas and gld versions.
10446 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10449 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10451 * cgraph.h (set_malloc_flag): Declare.
10452 * cgraph.c (set_malloc_flag_1): New function.
10453 (set_malloc_flag): Likewise.
10454 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10455 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10457 (read_ipa_call_summary): Add support for reading is_return_callee.
10458 (write_ipa_call_summary): Stream is_return_callee.
10459 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10460 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10461 ipa-prop.h, ipa-fnsummary.h.
10462 (pure_const_names): Change to static.
10463 (malloc_state_e): Define.
10464 (malloc_state_names): Define.
10465 (funct_state_d): Add field malloc_state.
10466 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10467 (check_retval_uses): New function.
10468 (malloc_candidate_p): Likewise.
10469 (analyze_function): Add support for malloc attribute.
10470 (pure_const_write_summary): Stream malloc_state.
10471 (pure_const_read_summary): Add support for reading malloc_state.
10472 (dump_malloc_lattice): New function.
10473 (propagate_malloc): New function.
10474 (warn_function_malloc): New function.
10475 (ipa_pure_const::execute): Call propagate_malloc and
10476 ipa_free_fn_summary.
10477 (pass_local_pure_const::execute): Add support for malloc attribute.
10478 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10479 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10481 2017-10-27 Martin Liska <mliska@suse.cz>
10483 PR gcov-profile/82457
10484 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10485 for fork-like functions to be properly instrumented.
10487 2017-10-27 Richard Biener <rguenther@suse.de>
10489 PR middle-end/81659
10490 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10491 info when we redirected EH.
10493 2017-10-26 Michael Collison <michael.collison@arm.com>
10495 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10497 (<optab>_trunchf<GPI:mode>2: New pattern.
10498 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10499 * config/aarch64/iterators.md (wv): New mode attribute.
10500 (vf, VF): New mode attributes.
10501 (vgp, VGP): New mode attributes.
10502 (s): Update attribute with SImode and DImode prefixes.
10504 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10506 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10507 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10508 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10509 (nios2_option_overide): Initialize it. Don't allow R0-relative
10510 addressing with PIC.
10511 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10512 (nios2_symbolic_constant_p): Likewise.
10513 (nios2_legitimate_address_p): Likewise.
10514 (nios2_r0rel_section_name_p): New.
10515 (nios2_symbol_ref_in_r0rel_data_p): New.
10516 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10517 (r0rel_constant_p): New.
10518 (nios2_print_operand_address): Handle r0rel_constant_p.
10519 (nios2_cdx_narrow_form_p): Likewise.
10520 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10521 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10522 (Nios II Options): Document -mr0rel-sec.
10524 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10526 * config/nios2/nios2.c: Include xregex.h.
10527 (nios2_gprel_sec_regex): New.
10528 (nios2_option_overide): Initialize it. Don't allow GP-relative
10529 addressing with PIC.
10530 (nios2_small_section_name_p): Check for regex match.
10531 * config/nios2/nios2.opt (mgprel-sec=): New option.
10532 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10533 (Nios II Options): Document -mgprel-sec.
10535 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10537 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10539 2017-10-26 Tom de Vries <tom@codesourcery.com>
10541 PR tree-optimization/82707
10542 * gimple.c (gimple_copy): Fix unsharing of
10543 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10545 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10547 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10548 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10549 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10550 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10551 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10552 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10553 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10554 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10555 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10556 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10557 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10558 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10559 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10560 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10561 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10562 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10563 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10565 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10567 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10569 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10570 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10571 warning to rs6000.c. Remove the Undocumented flag, since it has
10573 (-mabi=ibmlongdouble): Likewise.
10574 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10575 already set, set the default format for long double.
10576 (rs6000_debug_reg_global): Print whether long double is IBM or
10578 (rs6000_option_override_internal): Rework setting long double
10579 format. Only warn if the user is changing the long double default
10580 and they did not use -Wno-psabi.
10581 * doc/invoke.texi (PowerPC options): Update the documentation for
10582 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10584 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10585 Alan Hayward <alan.hayward@arm.com>
10586 David Sherwood <david.sherwood@arm.com>
10588 * rtl.h (wider_subreg_mode): New function.
10589 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10590 rather than an unsigned int *.
10591 * ira-color.c (regno_max_ref_width): Replace with...
10592 (regno_max_ref_mode): ...this new variable.
10593 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10594 Use wider_subreg_mode.
10595 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10596 rather than an unsigned int *.
10597 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10598 (process_alt_operands): Likewise.
10599 (invariant_p): Likewise.
10600 * lra-spills.c (assign_mem_slot): Likewise.
10601 (add_pseudo_to_slot): Likewise.
10602 * lra.c (collect_non_operand_hard_regs): Likewise.
10603 (add_regs_to_insn_regno_info): Likewise.
10604 * reload1.c (regno_max_ref_width): Replace with...
10605 (regno_max_ref_mode): ...this new variable.
10606 (reload): Update accordingly. Update call to
10607 ira_sort_regnos_for_alter_reg.
10608 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10609 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10610 (scan_paradoxical_subregs): Likewise.
10612 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10614 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10615 (aarch64_frame): Add emit_frame_chain boolean.
10616 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10617 Move eh_return case to aarch64_layout_frame.
10618 (aarch64_layout_frame): Initialize emit_frame_chain.
10619 (aarch64_expand_prologue): Use emit_frame_chain.
10621 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10623 * config/aarch64/aarch64.c (aarch64_layout_frame):
10624 Ensure LR is always stored at the bottom of the callee-saves.
10625 Remove rarely used frame layout which saves callee-saves at top of
10626 frame, so the store of LR can be used as a valid probe in all cases.
10628 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10630 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10631 Improve unaligned TImode/TFmode base/offset split.
10633 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10634 Alan Hayward <alan.hayward@arm.com>
10635 David Sherwood <david.sherwood@arm.com>
10637 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10638 * combine.c (find_single_use_1): Likewise.
10639 (expand_field_assignment): Likewise.
10640 (move_deaths): Likewise.
10641 * lra-constraints.c (simplify_operand_subreg): Likewise.
10642 (curr_insn_transform): Likewise.
10643 * lra.c (collect_non_operand_hard_regs): Likewise.
10644 (add_regs_to_insn_regno_info): Likewise.
10645 * rtlanal.c (reg_referenced_p): Likewise.
10646 (covers_regno_no_parallel_p): Likewise.
10648 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10650 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10651 Don't print any bits outside the precision of the value.
10652 * wide-int.cc (test_printing): Add some new tests.
10654 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10656 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10657 supports -xbrace_comment option.
10658 * configure: Regenerate.
10659 * config.in: Regenerate.
10660 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10661 (ASM_CPU_SPEC): Use it.
10663 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10665 * target.def (static_rtx_alignment): New hook.
10666 * targhooks.h (default_static_rtx_alignment): Declare.
10667 * targhooks.c (default_static_rtx_alignment): New function.
10668 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10669 * doc/tm.texi: Regenerate.
10670 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10671 instead of targetm.constant_alignment. Remove call to
10672 set_mem_attributes.
10673 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10674 (cris_preferred_mininum_alignment): New function, split out from...
10675 (cris_constant_alignment): ...here.
10676 (cris_static_rtx_alignment): New function.
10677 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10679 (ix86_constant_alignment): ...here.
10680 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10681 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10682 (mmix_static_rtx_alignment): New function.
10683 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10684 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10686 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10689 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10690 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10692 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10694 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10695 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10696 mark as a sign-extending load.
10697 (local_pic_load_u): Define.
10699 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10701 PR middle-end/82062
10702 * fold-const.c (operand_equal_for_comparison_p): Also return true
10703 if ARG0 is a simple variant of ARG1 with narrower precision.
10704 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10706 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10708 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10710 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10711 scatter_static, scatter_per_elt.
10712 * x86-tune-costs.h: Add new cost entries.
10714 2017-10-25 Richard Biener <rguenther@suse.de>
10716 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10717 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10718 class pass_fre): Move to ...
10719 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10720 class pass_fre): ... here and adjust for statistics.
10722 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10725 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10726 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10727 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10728 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10731 2017-10-25 Richard Biener <rguenther@suse.de>
10733 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10734 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10735 eliminate_push_avail, eliminate_insert): Move inside...
10736 (class eliminate_dom_walker): ... this class in preparation
10738 (fini_eliminate): Remove by merging with ...
10739 (eliminate): ... this function. Adjust for class changes.
10740 (pass_pre::execute): Remove fini_eliminate call.
10741 (pass_fre::execute): Likewise.
10743 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10746 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10747 (VPERMI2, VPERMI2I): New mode iterators.
10748 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10749 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10751 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10752 mode iterator. Remove 3 old define_insn patterns.
10753 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10754 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10755 VPERMI2 mode iterator, remove the other two expanders.
10756 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10757 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10758 instructions, remove the other two patterns.
10759 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10760 mode iterator, remove the other two patterns.
10761 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10762 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10763 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10764 and adjust argument order accordingly.
10765 (ix86_expand_vec_perm): Adjust caller.
10766 (expand_vec_perm_1): Likewise.
10767 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10768 (expand_vec_perm_vpermt2_vpshub2): ... this.
10769 (ix86_expand_vec_perm_const_1): Adjust caller.
10770 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10773 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10774 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10775 (vec_shl_<mode>): Remove unused expander.
10776 (avx512bw_<shift_insn><mode>3): New define_insn.
10777 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10778 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10780 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10783 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10786 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10788 PR rtl-optimization/82396
10789 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10790 (autopref_multipass_init): Simplify initialization.
10791 (autopref_rank_data): Simplify sort order.
10792 * gcc/sched-int.h (autopref_multipass_data_): Remove
10793 multi_mem_insn_p, min_offset and max_offset.
10795 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10797 PR middle-end/60580
10798 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10799 Check special value of flag_omit_frame_pointer.
10800 (aarch64_can_eliminate): Likewise.
10801 (aarch64_override_options_after_change_1): Simplify handling of
10802 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10804 2017-10-24 Richard Biener <rguenther@suse.de>
10806 PR tree-optimization/82697
10807 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10808 zero for conditional load and unconditional store.
10810 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10812 * doc/install.texi: Document bootstrap-cet.
10814 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10817 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10818 ENDBR instruction at function entrance if function is only
10821 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10824 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10825 patterns to better describe from which operation the CF is computed.
10826 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10827 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10828 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10829 is 0, use _0 suffixed expanders instead of emitting a comparison
10832 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10834 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10835 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10837 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10839 PR middle-end/82569
10840 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10841 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10842 * loop-iv.c (iv_get_reaching_def): Likewise.
10843 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10844 variable is promoted and the partition contains undefined values.
10846 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10848 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10850 (nios2_address_cost): Define.
10851 (nios2_legitimize_address): Recognize (exp + constant) directly.
10852 (TARGET_ADDRESS_COST): Define.
10854 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10856 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10857 (nios2_symbolic_memory_operand_p): Declare.
10858 (nios2_split_large_constant): Declare.
10859 (nios2_split_symbolic_memory_operand): Declare.
10860 * config/nios2/nios2.c: Adjust includes.
10861 (nios2_symbolic_constant_allowed): New.
10862 (nios2_symbolic_constant_p): New.
10863 (nios2_plus_symbolic_constant_p): New.
10864 (nios2_valid_addr_expr_p): Recognize addresses involving
10865 symbolic constants.
10866 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10867 (nios2_symbolic_memory_operand_p): New.
10868 (nios2_large_constant_p): New.
10869 (nios2_split_large_constant): New.
10870 (nios2_split_plus_large_constant): New.
10871 (nios2_split_symbolic_memory_operand): New.
10872 (nios2_legitimize_address): Code refactoring. Handle addresses
10873 involving symbolic constants.
10874 (nios2_emit_move_sequence): Likewise.
10875 (nios2_print_operand): Improve error output.
10876 (nios2_print_operand_address): Handle LO_SUM.
10877 (nios2_cdx_narrow_form_p): Likewise.
10878 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10879 operands involving symbolic constants.
10880 (movhi_internal, movsi_internal): Likewise.
10881 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10882 (extendhisi2, extendqi<mode>2): Likewise.
10884 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10886 * tree-pass.h (PROP_rtl_split_insns): Define.
10887 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10889 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10891 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10893 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10896 * target.def (const_not_ok_for_debug_p): Default to
10897 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10898 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10899 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10900 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10901 which targetm.const_not_ok_for_debug_p returned true.
10902 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10904 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10906 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10907 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10908 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10910 (ix86_const_not_ok_for_debug_p): New function.
10911 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10912 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10914 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10917 * system.h: Conditionally include "unique-ptr.h" if
10918 INCLUDE_UNIQUE_PTR is defined.
10919 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10920 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10922 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10924 * config/rl78/rl78.md: New define_expand "subdi3".
10926 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10929 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10930 DF_REF_INSN if DF_REF_INSN_INFO is false.
10932 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10934 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10935 xmm_move instead of sse_move.
10936 (sse_store_index): New function.
10937 (ix86_register_move_cost): Be more sensible about mismatch stall;
10938 model AVX moves correctly; make difference between sse->integer and
10940 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10941 moves; make difference between SSE and AVX.
10942 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10943 and zmm_move. Increase size of sse load and store tables;
10944 add unaligned load and store tables; add ssemmx_to_integer.
10945 * x86-tune-costs.h: Update all entries according to real
10946 move latencies from Agner Fog's manual and chip documentation.
10948 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10951 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10952 * config/i386/constraints.md (Wf): New constraint.
10953 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10954 (cmp<dwi>_doubleword): Removed.
10955 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10956 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10957 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10958 expand with cmp<dwi>_doubleword. For LTU and GEU use
10959 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10961 * common.opt (gcolumn-info): Enable by default.
10962 * doc/invoke.texi (gcolumn-info): Document new default.
10964 2017-10-23 Richard Biener <rguenther@suse.de>
10966 PR tree-optimization/82672
10967 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10968 Fold the stmt if we propagated into it.
10970 2017-10-23 Richard Biener <rguenther@suse.de>
10972 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10973 (bitmap_remove_expr_from_set): ... this. All callers call this
10974 for non-constant values.
10975 (bitmap_set_subtract): Rename to...
10976 (bitmap_set_subtract_expressions): ... this. Adjust and
10978 (bitmap_set_contains_value): Remove superfluous check.
10979 (bitmap_set_replace_value): Inline into single caller ...
10980 (bitmap_value_replace_in_set): ... here and simplify.
10981 (dependent_clean): Merge into ...
10982 (clean): ... this using an overload. Adjust.
10983 (prune_clobbered_mems): Adjust.
10984 (compute_antic_aux): Likewise.
10985 (compute_partial_antic_aux): Likewise.
10987 2017-10-23 Richard Biener <rguenther@suse.de>
10989 PR tree-optimization/82129
10991 2017-08-01 Richard Biener <rguenther@suse.de>
10993 PR tree-optimization/81181
10994 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10995 (compute_antic): ... end of iteration here.
10997 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10999 * target.def (starting_frame_offset): New hook.
11000 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11001 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11002 * doc/tm.texi.in: Regenerate.
11003 * hooks.h (hook_hwi_void_0): Declare.
11004 * hooks.c (hook_hwi_void_0): New function.
11005 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11006 STARTING_FRAME_OFFSET.
11007 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11008 * reload1.c (reload): Likewise.
11009 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11010 instead of STARTING_FRAME_OFFSET.
11011 * function.c (try_fit_stack_local): Likewise.
11012 (assign_stack_local_1): Likewise
11013 (instantiate_virtual_regs): Likewise.
11014 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11015 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11016 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11017 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11018 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11019 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11020 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11021 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11022 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11023 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11024 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11025 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11026 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11027 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11028 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11029 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11030 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11031 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11032 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11033 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11034 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11035 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11036 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11037 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11038 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11039 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11040 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11041 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11042 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11043 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11044 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11045 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11046 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11047 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11048 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11049 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11050 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11051 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11052 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11053 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11054 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11055 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11056 return a HOST_WIDE_INT.
11057 (avr_builtin_setjmp_frame_value): Use it instead of
11058 STARTING_FRAME_OFFSET.
11059 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11060 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11061 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11063 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11064 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11065 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11066 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11067 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11068 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11069 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11070 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11071 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11072 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11073 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11074 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11076 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11077 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11078 * config/mips/mips.c (mips_compute_frame_info): Refer to
11079 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11080 (mips_starting_frame_offset): New function.
11081 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11082 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11083 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11084 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11085 and return a HOST_WIDE_INT.
11086 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11087 (mmix_initial_elimination_offset): Refer to
11088 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11089 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11090 * config/pa/pa.c (pa_starting_frame_offset): New function.
11091 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11092 (pa_expand_prologue): Likewise.
11093 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11094 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11095 !FRAME_GROWS_DOWNWARD handling to...
11096 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11097 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11098 !FRAME_GROWS_DOWNWARD handling to...
11099 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11100 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11101 !FRAME_GROWS_DOWNWARD handling to...
11102 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11103 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11105 (rs6000_starting_frame_offset): New function.
11106 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11107 !FRAME_GROWS_DOWNWARD handling to...
11108 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11109 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11110 !FRAME_GROWS_DOWNWARD handling to...
11111 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11112 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11113 !FRAME_GROWS_DOWNWARD handling to...
11114 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11115 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11116 (rs6000_starting_frame_offset): New function.
11117 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11118 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11119 * config/vax/vax.c (vax_starting_frame_offset): New function.
11120 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11121 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11122 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11123 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11124 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11125 * system.h (STARTING_FRAME_OFFSET): Poison.
11127 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11129 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11130 SCALAR_TYPE_MODE instead of TYPE_MODE.
11132 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11133 Alan Hayward <alan.hayward@arm.com>
11134 David Sherwood <david.sherwood@arm.com>
11136 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11138 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11139 Alan Hayward <alan.hayward@arm.com>
11140 David Sherwood <david.sherwood@arm.com>
11142 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11144 2017-10-23 Richard Biener <rguenther@suse.de>
11146 PR tree-optimization/82129
11147 * tree-ssa-pre.c (bitmap_set_and): Remove.
11148 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11149 canonicalizing expressions in the set to those with lowest
11150 ID rather than taking that from the first edge.
11152 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11154 * combine.c (rtx_equal_for_field_assignment_p): Use
11155 byte_lowpart_offset.
11157 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11158 Alan Hayward <alan.hayward@arm.com>
11159 David Sherwood <david.sherwood@arm.com>
11161 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11162 to a SUBREG_PROMOTED_VAR.
11164 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11165 Alan Hayward <alan.hayward@arm.com>
11166 David Sherwood <david.sherwood@arm.com>
11168 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11169 (expand_debug_source_expr): Likewise.
11170 * combine.c (combine_simplify_rtx): Likewise.
11171 * cse.c (fold_rtx): Likewise.
11172 * optabs.c (expand_float): Likewise.
11173 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11174 (simplify_binary_operation_1): Likewise.
11176 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11177 Alan Hayward <alan.hayward@arm.com>
11178 David Sherwood <david.sherwood@arm.com>
11180 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11181 (record_promoted_value): Likewise.
11182 * expr.c (expand_expr_real_2): Likewise.
11183 * ree.c (update_reg_equal_equiv_notes): Likewise.
11184 (combine_set_extension): Likewise.
11185 * rtlanal.c (low_bitmask_len): Likewise.
11186 * simplify-rtx.c (neg_const_int): Likewise.
11187 (simplify_binary_operation_1): Likewise.
11189 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11190 Alan Hayward <alan.hayward@arm.com>
11191 David Sherwood <david.sherwood@arm.com>
11193 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11194 * regcprop.c (maybe_mode_change): Likewise.
11195 * reload1.c (alter_reg): Likewise.
11197 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11199 * inchash.h (inchash::hash::add_wide_int): New function.
11200 * lto-streamer-out.c (hash_tree): Use it.
11202 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11204 * inchash.h (inchash::hash::add_wide_int): Rename to...
11205 (inchash::hash::add_hwi): ...this.
11206 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11207 (polymorphic_call_target_hasher::hash): Likewise.
11208 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11209 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11210 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11211 * lto-streamer-out.c (hash_tree): Likewise.
11212 * optc-save-gen.awk: Likewise.
11213 * tree.c (add_expr): Likewise.
11215 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11218 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11219 for ordered inequality comparisons even with TARGET_IEEE_FP.
11221 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11224 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11225 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11226 Expand with cmp<dwi>_doubleword.
11228 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11230 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11231 List CET intrinsics.
11232 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11233 specific to -fcf-protection option.
11235 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11237 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11238 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11239 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11240 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11241 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11242 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11243 (extra_objs): Add cet.o for Linux/x86 targets.
11244 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11245 * config/i386/cet.c: New file.
11246 * config/i386/cetintrin.h: Likewise.
11247 * config/i386/t-cet: Likewise.
11248 * config/i386/cpuid.h (bit_SHSTK): New.
11249 (bit_IBT): Likewise.
11250 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11251 pass IBT and SHSTK bits.
11252 * config/i386/i386-builtin-types.def
11253 (VOID_FTYPE_UNSIGNED_PVOID): New.
11254 (VOID_FTYPE_UINT64_PVOID): Likewise.
11255 * config/i386/i386-builtin.def: Add CET intrinsics.
11256 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11257 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11258 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11259 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11261 * config/i386/i386.c (rest_of_insert_endbranch): New.
11262 (pass_data_insert_endbranch): Likewise.
11263 (pass_insert_endbranch): Likewise.
11264 (make_pass_insert_endbranch): Likewise.
11265 (ix86_notrack_prefixed_insn_p): Likewise.
11266 (ix86_target_string): Add -mibt, -mshstk flags.
11267 (ix86_option_override_internal): Add flag_cf_protection
11269 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11270 (ix86_print_operand): Add 'notrack' prefix output.
11271 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11272 (ix86_expand_builtin): Expand CET intrinsics.
11273 (x86_output_mi_thunk): Add 'endbranch' instruction.
11274 * config/i386/i386.h (TARGET_IBT): New.
11275 (TARGET_IBT_P): Likewise.
11276 (TARGET_SHSTK): Likewise.
11277 (TARGET_SHSTK_P): Likewise.
11278 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11279 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11280 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11281 (builtin_setjmp_setup): New pattern.
11282 (builtin_longjmp): Likewise.
11283 (rdssp<mode>): Likewise.
11284 (incssp<mode>): Likewise.
11285 (saveprevssp): Likewise.
11286 (rstorssp): Likewise.
11287 (wrss<mode>): Likewise.
11288 (wruss<mode>): Likewise.
11289 (setssbsy): Likewise.
11290 (clrssbsy): Likewise.
11291 (nop_endbr): Likewise.
11292 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11294 * config/i386/immintrin.h: Include <cetintrin.h>.
11295 * config/i386/linux-common.h
11296 (file_end_indicate_exec_stack_and_cet): New prototype.
11297 (TARGET_ASM_FILE_END): New.
11299 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11301 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11302 latencies instead of having separate table; make difference between
11303 integer and float costs.
11304 * i386.h (processor_costs): Remove scalar_stmt_cost,
11305 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11306 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11308 * x86-tune-costs.h: Remove entries which has been removed in
11309 procesor_costs from all tables; make cond_taken_branch_cost
11310 and cond_not_taken_branch_cost COST_N_INSNS based.
11312 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11314 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11316 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11318 * config/i386/i386.md (isa): Remove fma_avx512f.
11319 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11320 <avx512>_fmadd_<mode>_mask3<round_name>,
11321 <avx512>_fmsub_<mode>_mask<round_name>,
11322 <avx512>_fmsub_<mode>_mask3<round_name>,
11323 <avx512>_fnmadd_<mode>_mask<round_name>,
11324 <avx512>_fnmadd_<mode>_mask3<round_name>,
11325 <avx512>_fnmsub_<mode>_mask<round_name>,
11326 <avx512>_fnmsub_<mode>_mask3<round_name>,
11327 <avx512>_fmaddsub_<mode>_mask<round_name>,
11328 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11329 <avx512>_fmsubadd_<mode>_mask<round_name>,
11330 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11331 (*vec_widen_umult_even_v16si<mask_name>,
11332 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11333 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11335 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11337 * extend.texi: Add 'nocf_check' documentation.
11338 * gimple.texi: Add second parameter to
11339 gimple_build_call_from_tree.
11340 * invoke.texi: Add -fcf-protection documentation.
11341 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11343 2017-10-20 Richard Biener <rguenther@suse.de>
11345 PR tree-optimization/82473
11346 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11347 the largest input type.
11349 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11351 * c-attribs.c (handle_nocf_check_attribute): New function.
11352 (c_common_attribute_table): Add 'nocf_check' handling.
11353 * gimple-parser.c: Add second argument NULL to
11354 gimple_build_call_from_tree.
11355 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11356 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11358 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11359 * common.opt: Add fcf-protection flag.
11360 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11361 * flag-types.h: Add enum cf_protection_level.
11362 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11363 Add 'nocf_check' attribute propagation to gimple call.
11364 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11365 (gimple_build_call_from_tree): Update prototype.
11366 (gimple_call_nocf_check_p): New function.
11367 (gimple_call_set_nocf_check): Likewise.
11368 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11369 * ipa-icf.c: Add nocf_check attribute in statement hash.
11370 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11371 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11372 * toplev.c (process_options): Add flag_cf_protection handling.
11374 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11376 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11378 2017-10-20 Richard Biener <rguenther@suse.de>
11380 PR tree-optimization/82603
11381 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11382 remove false predicated stores.
11384 2017-10-20 Richard Biener <rguenther@suse.de>
11386 * graphite-isl-ast-to-gimple.c
11387 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11388 Remove return value and simplify, dump copied stmt after lhs
11390 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11391 Reduce dump verbosity.
11392 (gsi_insert_earliest): Likewise.
11393 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11394 * graphite.c (print_global_statistics): Adjust dumping.
11395 (print_graphite_scop_statistics): Likewise.
11396 (print_graphite_statistics): Do not dump loops here.
11397 (graphite_transform_loops): But here.
11399 2017-10-20 Nicolas Roche <roche@adacore.com>
11401 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11402 * configure: Regenerate.
11404 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11407 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11408 functions when optimizing replace GIMPLE_RETURN stmts with
11409 calls to __builtin_unreachable ().
11412 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11413 for -fsanitize=thread link of executables.
11414 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11415 link of executables.
11418 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11419 New mode iterators.
11420 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11421 define_insns for logical vector shifts to use VI248_AVX512BW
11422 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11423 condition, useless isa and prefix attributes. Change the first
11424 2 of these define_insns to ...
11425 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11426 define_insn for avx512vl.
11427 (<shift_insn><mode>3): ... and this, new define_insn without
11428 masking for non-avx512vl.
11431 * config/i386/sse.md (*andnot<mode>3,
11432 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11433 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11434 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11435 not applied use empty suffix even for TARGET_AVX512VL.
11436 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11437 is applied, supply evex,evex or evex,evex,evex instead of just
11440 2017-10-20 Julia Koval <julia.koval@intel.com>
11442 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11443 (OPTION_MASK_ISA_GFNI_UNSET): New.
11444 (ix86_handle_option): Handle OPT_mgfni.
11445 * config/i386/cpuid.h (bit_GFNI): New.
11446 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11447 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11448 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11449 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11450 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11451 * config/i386/i386.opt: Add mgfni.
11453 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11455 * config/msp430/msp430.c (msp430_option_override): Disable
11456 -fdelete-null-pointer-checks.
11457 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11459 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11461 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11462 of x87 and SSE instructions.
11464 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11466 * asan.c (create_cond_insert_point): Do not update edge count.
11467 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11468 (afdo_propagate_circuit): Likewise.
11469 (afdo_calculate_branch_prob): Likewise.
11470 (afdo_annotate_cfg): Likewise.
11471 * basic-block.h (struct edge_def): Remove count.
11472 (edge_def::count): New accessor.
11473 * bb-reorder.c (rotate_loop): Update.
11474 (find_traces_1_round): Update.
11475 (connect_traces): Update.
11476 (sanitize_hot_paths): Update.
11477 * cfg.c (unchecked_make_edge): Update.
11478 (make_single_succ_edge): Update.
11479 (check_bb_profile): Update.
11480 (dump_edge_info): Update.
11481 (update_bb_profile_for_threading): Update.
11482 (scale_bbs_frequencies_int): Update.
11483 (scale_bbs_frequencies_gcov_type): Update.
11484 (scale_bbs_frequencies_profile_count): Update.
11485 (scale_bbs_frequencies): Update.
11486 * cfganal.c (connect_infinite_loops_to_exit): Update.
11487 * cfgbuild.c (compute_outgoing_frequencies): Update.
11488 (find_many_sub_basic_blocks): Update.
11489 * cfgcleanup.c (try_forward_edges): Update.
11490 (try_crossjump_to_edge): Update
11491 * cfgexpand.c (expand_gimple_cond): Update
11492 (expand_gimple_tailcall): Update
11493 (construct_exit_block): Update
11494 * cfghooks.c (verify_flow_info): Update
11495 (redirect_edge_succ_nodup): Update
11496 (split_edge): Update
11497 (make_forwarder_block): Update
11498 (duplicate_block): Update
11499 (account_profile_record): Update
11500 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11501 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11502 * cfgloopmanip.c (scale_loop_profile): Update.
11504 (lv_adjust_loop_entry_edge): Update.
11505 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11506 (force_nonfallthru_and_redirect): Update.
11507 (purge_dead_edges): Update.
11508 (rtl_flow_call_edges_add): Update.
11509 * cgraphunit.c (init_lowered_empty_function): Update.
11510 (cgraph_node::expand_thunk): Update.
11511 * gimple-pretty-print.c (dump_probability): Update.
11512 (dump_edge_probability): Update.
11513 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11514 * haifa-sched.c (sched_create_recovery_edges): Update.
11515 * hsa-gen.c (convert_switch_statements): Update.
11516 * ifcvt.c (dead_or_predicable): Update.
11517 * ipa-inline-transform.c (inline_transform): Update.
11518 * ipa-split.c (split_function): Update.
11519 * ipa-utils.c (ipa_merge_profiles): Update.
11520 * loop-doloop.c (add_test): Update.
11521 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11522 * lto-streamer-in.c (input_cfg): Update.
11523 (input_function): Update.
11524 * lto-streamer-out.c (output_cfg): Update.
11525 * modulo-sched.c (sms_schedule): Update.
11526 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11527 * predict.c (maybe_hot_edge_p): Update.
11528 (unlikely_executed_edge_p): Update.
11529 (probably_never_executed_edge_p): Update.
11530 (dump_prediction): Update.
11531 (drop_profile): Update.
11532 (propagate_unlikely_bbs_forward): Update.
11533 (determine_unlikely_bbs): Update.
11534 (force_edge_cold): Update.
11535 * profile.c (compute_branch_probabilities): Update.
11536 * reg-stack.c (better_edge): Update.
11537 * shrink-wrap.c (handle_simple_exit): Update.
11538 * tracer.c (better_p): Update.
11539 * trans-mem.c (expand_transaction): Update.
11540 (split_bb_make_tm_edge): Update.
11541 * tree-call-cdce.c: Update.
11542 * tree-cfg.c (gimple_find_sub_bbs): Update.
11543 (gimple_split_edge): Update.
11544 (gimple_duplicate_sese_region): Update.
11545 (gimple_duplicate_sese_tail): Update.
11546 (gimple_flow_call_edges_add): Update.
11547 (insert_cond_bb): Update.
11548 (execute_fixup_cfg): Update.
11549 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11550 * tree-complex.c (expand_complex_div_wide): Update.
11551 * tree-eh.c (lower_resx): Update.
11552 (unsplit_eh): Update.
11553 (cleanup_empty_eh_move_lp): Update.
11554 * tree-inline.c (copy_edges_for_bb): Update.
11555 (freqs_to_counts): Update.
11556 (copy_cfg_body): Update.
11557 * tree-ssa-dce.c (remove_dead_stmt): Update.
11558 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11559 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11560 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11561 (unloop_loops): Update.
11562 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11563 * tree-ssa-loop-split.c (connect_loops): Update.
11564 (split_loop): Update.
11565 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11566 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11567 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11568 * tree-ssa-reassoc.c (branch_fixup): Update.
11569 * tree-ssa-tail-merge.c (replace_block_by): Update.
11570 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11571 (compute_path_counts): Update.
11572 (update_profile): Update.
11573 (recompute_probabilities): Update.
11574 (update_joiner_offpath_counts): Update.
11575 (estimated_freqs_path): Update.
11576 (freqs_to_counts_path): Update.
11577 (clear_counts_path): Update.
11578 (ssa_fix_duplicate_block_edges): Update.
11579 (duplicate_thread_path): Update.
11580 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11581 (case_bit_test_cmp): Update.
11582 (collect_switch_conv_info): Update.
11583 (gen_inbound_check): Update.
11584 (do_jump_if_equal): Update.
11585 (emit_cmp_and_jump_insns): Update.
11586 * tree-tailcall.c (decrease_profile): Update.
11587 (eliminate_tail_call): Update.
11588 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11589 (vect_do_peeling): Update.
11590 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11591 * ubsan.c (ubsan_expand_null_ifn): Update.
11592 (ubsan_expand_ptr_ifn): Update.
11593 * value-prof.c (gimple_divmod_fixed_value): Update.
11594 (gimple_mod_pow2): Update.
11595 (gimple_mod_subtract): Update.
11596 (gimple_ic): Update.
11597 (gimple_stringop_fixed_value): Update.
11599 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11602 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11604 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11606 PR rtl-optimization/82395
11607 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11608 based on non_spilled_static_chain_regno_p.
11610 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11612 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11613 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11614 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11616 2017-10-19 Martin Sebor <msebor@redhat.com>
11618 PR tree-optimization/82596
11619 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11621 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11623 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11624 (get_mem_refs_of_builtin_call): Likewise.
11625 * builtins.c (expand_builtin_apply): Adjust call to
11626 allocate_dynamic_stack_space.
11627 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11628 the third argument to allocate_dynamic_stack_space, otherwise -1.
11629 (expand_builtin): Deal with all alloca variants.
11630 (is_inexpensive_builtin): Likewise.
11631 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11632 * calls.c (special_function_p): Deal with all alloca variants.
11633 (initialize_argument_information): Adjust call to
11634 allocate_dynamic_stack_space.
11635 (expand_call): Likewise.
11636 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11637 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11638 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11639 use it for the stack usage computation.
11640 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11641 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11642 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11643 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11644 (in_loop_p): Remove first argument and useless check.
11645 (pass_walloca::execute): Remove useless test and adjust call to above.
11646 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11647 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11648 (gimplify_call_expr): Deal with all alloca variants.
11649 * hsa-gen.c (gen_hsa_alloca): Likewise.
11650 (gen_hsa_insns_for_call): Likewise.
11651 * ipa-pure-const.c (special_builtin_state): Likewise.
11652 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11653 * tree-object-size.c (alloc_object_size): Likewise.
11654 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11655 (call_may_clobber_ref_p_1): Likewise.
11656 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11657 (ccp_fold_stmt): Likewise.
11658 (optimize_stack_restore): Likewise.
11659 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11660 (mark_all_reaching_defs_necessary_1): Likewise.
11661 (propagate_necessity): Likewise.
11662 (eliminate_unnecessary_stmts): Likewise.
11663 * tree.c (build_common_builtin_nodes): Build
11664 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11665 (build_alloca_call_expr): New function.
11666 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11667 (CASE_BUILT_IN_ALLOCA): Likewise.
11668 (build_alloca_call_expr): Declare.
11669 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11671 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11674 * dwarf2out.c (new_die_raw): New static inline function.
11675 (new_die): Use it to create the DIE.
11676 (add_AT_external_die_ref): Likewise.
11677 (clone_die): Likewise.
11678 (clone_as_declaration): Likewise.
11679 (dwarf2out_vms_debug_main_pointer): Likewise.
11680 (base_type_die): Likewise. Remove early return for corner cases.
11681 Do not call add_pubtype on the DIE here.
11682 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11683 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11684 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11685 native order exists for base types, attach the DIE manually and call
11686 add_pubtype on it. Do not equate a reverse order DIE to the type.
11688 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11690 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11691 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11693 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11695 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11697 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11698 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11699 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11700 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11701 * opts.c (sanitizer_opts): Add builtin.
11702 * ubsan.c (instrument_builtin): New function.
11703 (pass_ubsan::execute): Call it.
11704 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11705 * doc/invoke.texi: Document -fsanitize=builtin.
11707 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11708 builtins, store max (log2 (align), 0) into uchar field instead of
11709 align into uptr field.
11710 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11711 store uchar 0 field instead of uptr 0 field.
11712 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11713 instead of passing one address of struct with 2 locations pass
11714 two addresses of structs with 1 location each.
11715 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11716 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11717 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11718 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11719 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11720 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11721 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11722 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11724 2017-10-19 Martin Liska <mliska@suse.cz>
11727 * file-find.c (remove_prefix): Remove.
11728 * file-find.h (remove_prefix): Likewise.
11729 * gcc-ar.c: Remove smartness of lookup.
11731 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11733 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11734 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11735 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11737 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11740 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11741 (setcc + and to xor + setcc): New peephole2.
11743 2017-10-19 Tom de Vries <tom@codesourcery.com>
11745 * doc/sourcebuild.texi (Test Directives, Variants of
11746 dg-require-support): Add dg-require-stack-size.
11748 2017-10-19 Martin Liska <mliska@suse.cz>
11751 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11752 that have a large alignment.
11753 (gimplify_target_expr): Likewise.
11755 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11757 PR rtl-optimization/82602
11758 * ira.c (rtx_moveable_p): Return false for volatile asm.
11760 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11763 * config/i386/i386-modes.def (CCGZ): New CC mode.
11764 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11765 * config/i386/predicates.md (ix86_comparison_operator):
11767 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11768 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11769 with double-word subtraction.
11770 (put_condition_code): Handle CCGZmode.
11772 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11774 * wide-int.cc (debug (const wide_int &)): New.
11775 (debug (const wide_int *)): New.
11776 (debug (const widest_int &)): New.
11777 (debug (const widest_int *)): New.
11779 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11781 PR middle-end/82556
11782 * lra-constraints.c (curr_insn_transform): Use non-input operand
11783 instead of output one for matched reload.
11785 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11787 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11788 (tree-ssa-loop-ivopts.h): New header file.
11789 (struct builtin_info): New fields.
11790 (classify_builtin_1): Compute and record base and offset parts for
11791 memset builtin partition by calling strip_offset.
11792 (offset_cmp, fuse_memset_builtins): New functions.
11793 (finalize_partitions): Fuse adjacent memset partitions by calling
11795 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11796 Expose the interface.
11797 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11799 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11801 PR tree-optimization/82574
11802 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11803 that data reference must be executed exactly once per iteration
11804 against the outermost loop in nest.
11805 (classify_partition): Update call to above function.
11807 2017-10-18 Richard Biener <rguenther@suse.de>
11809 PR tree-optimization/82591
11810 * graphite.c (graphite_transform_loops): Move code gen message
11812 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11813 Here. Handle scop_to_isl_ast failing.
11814 (scop_to_isl_ast): Limit the number of ISL operations.
11816 2017-10-18 Richard Biener <rguenther@suse.de>
11818 * graphite-isl-ast-to-gimple.c
11819 (translate_isl_ast_to_gimple::set_rename): Simplify.
11820 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11821 (graphite_copy_stmts_from_block): ... here.
11822 (copy_bb_and_scalar_dependences): Simplify.
11823 (add_parameters_to_ivs_params): Canonicalize.
11824 (generate_entry_out_of_ssa_copies): Simplify.
11825 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11826 by passing in ISL dimension.
11827 (parameter_index_in_region_1): Rename to ...
11828 (parameter_index_in_region): ... this.
11829 (extract_affine): Adjust assert, pass down parameter index.
11830 (add_param_constraints): Use range-info when available.
11831 (build_scop_context): Adjust.
11832 * sese.c (new_sese_info): Adjust.
11833 (free_sese_info): Likewise.
11834 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11835 Remove unused typedefs.
11836 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11838 2017-10-18 Martin Liska <mliska@suse.cz>
11840 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11842 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11844 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11845 (s390_sched_init): Do not reset s390_sched_state if we entered the
11846 current basic block via a fallthru edge and all others are unlikely.
11848 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11850 * config/s390/s390.c (NUM_SIDES): New variable.
11851 (LONGRUNNING_THRESHOLD): New variable.
11852 (LATENCY_FACTOR): New variable.
11853 (s390_sched_score): Decrease score for long-running instructions on
11855 (s390_sched_variable_issue): Perform bookkeeping for long-running
11858 2017-10-18 Richard Biener <rguenther@suse.de>
11860 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11861 Simplify with removal of the parameter rename map.
11862 (set_rename): Likewise.
11863 (should_copy_to_new_region): Likewise.
11864 (graphite_copy_stmts_from_block): Likewise.
11865 (copy_bb_and_scalar_dependences): Remove initialization of
11866 unused copied_bb_map.
11867 (copy_def): Remove.
11868 (copy_internal_parameters): Likewise.
11869 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11870 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11871 Use INTEGRAL_TYPE_P.
11872 (parameter_index_in_region_1): Rename to ...
11873 (assign_parameter_index_in_region): ... this. Assert we have
11874 a parameter we handle.
11875 (scan_tree_for_params): Adjust.
11876 * sese.h (parameter_rename_map_t): Remove.
11877 (struct sese_info_t): Remove unused parameter_rename_map and
11878 copied_bb_map members.
11879 * sese.c (new_sese_info): Adjust.
11880 (free_sese_info): Likewise.
11882 2017-10-18 Martin Liska <mliska@suse.cz>
11885 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11886 on an abnormal edge.
11888 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11890 * doc/invoke.texi (ffunction-sections and fdata-sections):
11893 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11895 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11896 the use statement can throw internally.
11898 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11900 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11901 any RTX present on the RHS of a SET.
11902 * compare-elim.c (try_eliminate_compare): Restore comment.
11904 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11906 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11907 may be also called on tcc_type nodes.
11908 * langhooks.c (lhd_tree_size): Likewise.
11910 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11912 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11913 format_warning_at_substring.
11914 (maybe_warn): Convert source_range * param to a location_t. Pass
11915 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11916 (format_directive): Remove code to extract source_ranges and
11917 source_range * in favor of just a location_t.
11918 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11920 * substring-locations.c (format_warning_va): Convert
11921 source_range * param to a location_t.
11922 (format_warning_at_substring): Likewise.
11923 * substring-locations.h (format_warning_va): Likewise.
11924 (format_warning_at_substring): Likewise.
11926 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11928 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11930 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11931 and scatter/gather ops.
11933 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11934 vec_gather_load and vec_scatter_store.
11935 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11936 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11937 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11938 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11939 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11940 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11942 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11944 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11945 Detect FTST instruction and handle its register pops. Only pop
11946 second operand if can_pop_second_op is true.
11947 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11948 set can_pop_second_op to false in the compare_for_stack_reg call.
11950 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11951 output_fp_compare for stack register operands.
11952 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11953 instructions here. Do not emit stack register pops here. Assert
11954 that FCOMPP pops next to top stack register. Rewrite function.
11956 2017-10-17 Nathan Sidwell <nathan@acm.org>
11958 PR middle-end/82577
11959 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11960 use DECL_ASSEMBLER_NAME_RAW.
11962 PR middle-end/82546
11963 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11966 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11967 Wilco Dijkstra <wilco.dijkstra@arm.com>
11969 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11970 converstion to Pmode from the buf_addr.
11972 2017-10-17 Richard Biener <rguenther@suse.de>
11974 * graphite-dependences.c (scop_get_reads_and_writes): Change
11975 output parameters to references.
11977 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11979 PR 71026/tree-optimization
11980 * fold-const.c (distribute_real_division): Removed.
11981 (fold_binary_loc): Remove calls to distribute_real_divison.
11983 2017-10-17 Richard Biener <rguenther@suse.de>
11985 * graphite-scop-detection.c
11986 (scop_detection::stmt_has_simple_data_refs_p): Always use
11987 the full nest as region.
11988 (try_generate_gimple_bb): Likewise.
11989 * sese.c (scalar_evolution_in_region): Simplify now that
11990 SCEV can handle instantiation in regions.
11991 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11992 in the non-loop part of a function if requested.
11994 2017-10-17 Richard Biener <rguenther@suse.de>
11996 PR tree-optimization/82563
11997 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11999 (graphite_regenerate_ast_isl): Call it.
12000 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12002 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12004 PR tree-optimization/82549
12005 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12006 Formatting fixes. Instead of calling make_bit_field_ref with negative
12009 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12011 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12012 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12013 _mm_maskz_reduce_ss): New.
12014 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12015 __builtin_ia32_reducess_mask): Ditto..
12016 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12017 * config/i386/sse.md (reduces<mode>): Renamed to ...
12018 (reduces<mode><mask_scalar_name>): ... this.
12019 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12021 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12022 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12024 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12026 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12027 * selftest-run-tests.c (selftest::run_tests): Call
12028 selftest::unique_ptr_tests_cc_tests.
12029 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12030 * unique-ptr-tests.cc: New file.
12032 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12035 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12037 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12038 (make_hard_regno_born, make_hard_regno_dead): Update
12039 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12041 2017-10-16 Jeff Law <law@redhat.com>
12043 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12045 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12047 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12049 2017-10-16 Olivier Hainque <hainque@adacore.com>
12051 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12052 with_cpu if we were configured for an e500v2 target cpu name.
12054 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12056 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12057 * doc/invoke.texi: Document +nodsp as a valid extension for
12060 2017-10-16 Martin Liska <mliska@suse.cz>
12062 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12063 (test_set_range): Likewise.
12064 (test_range_functions): Rename to ...
12065 (test_bit_in_range): ... this.
12066 (sbitmap_c_tests): Add new test.
12068 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12070 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12072 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12073 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12075 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12077 * config/aarch64/aarch64-builtins.c
12078 (aarch64_types_quadopu_lane_qualifiers): New.
12079 (TYPES_QUADOPU_LANE): New.
12080 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12081 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12082 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12083 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12084 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12085 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12086 (Vdottype, DOTPROD): New.
12087 (sur): Add SDOT and UDOT.
12089 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12091 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12092 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12093 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12094 Add TARGET_DOTPROD.
12095 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12096 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12097 Enable TARGET_DOTPROD.
12098 (cortex-a75.cortex-a55): Likewise.
12099 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12101 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12103 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12104 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12106 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12108 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12109 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12110 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12111 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12112 * config/arm/types.md (neon_dot, neon_dot_q): New.
12113 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12115 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12117 * config/arm/arm.h (TARGET_DOTPROD): New.
12118 * config/arm/arm.c (arm_arch_dotprod): New.
12119 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12120 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12121 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12122 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12123 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12124 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12125 * doc/invoke.texi (armv8.2-a): Document dotprod
12127 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12129 * i386.c (ix86_vec_cost): New function.
12130 (ix86_rtx_costs): Handle vector operations better.
12131 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12132 * x86-tune-costs.h: Add new costs to all tables.
12134 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12136 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12138 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12139 divsd, sqrtss and sqrtsd
12140 * x86-tune-costs.h: Add new entries to all costs.
12141 (znver1_cost): Fix to match real instruction latencies.
12143 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12144 Michael Collison <michael.collison@arm.com>
12146 * compare-elim.c: Include emit-rtl.h.
12147 (can_merge_compare_into_arith): New function.
12148 (try_validate_parallel): Likewise.
12149 (try_merge_compare): Likewise.
12150 (try_eliminate_compare): Call the above when no previous clobber
12152 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12155 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12157 PR middle-end/62263
12158 PR middle-end/82498
12159 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12160 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12162 PR middle-end/62263
12163 PR middle-end/82498
12164 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12165 to be any operand_equal_p operands. For & (B - 1) require
12166 B to be power of 2. Recognize
12167 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12169 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12172 * optabs.c (expand_memory_blockage): Fix call of
12173 targetm.have_memory_blockage.
12175 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12178 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12179 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12180 objects to extra_objs instead of overwriting it.
12182 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12184 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12185 Use any_fp_register_operand as operand[3] predicate. Simplify
12186 equality test for operands[2] and operands[4] memory location.
12187 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12188 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12189 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12190 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12191 any_fp_register_operand as operand[1] predicate. Simplify
12192 equality test for operands[0] and operands[3] memory location.
12193 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12194 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12195 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12197 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12199 * target-insns.def: Add memory_blockage.
12200 * optabs.c (expand_memory_blockage): New function.
12201 (expand_asm_memory_barrier): Rename ...
12202 (expand_asm_memory_blockage): ... to this.
12203 (expand_mem_thread_fence): Call expand_memory_blockage
12204 instead of expand_asm_memory_barrier.
12205 (expand_mem_singnal_fence): Ditto.
12206 (expand_atomic_load): Ditto.
12207 (expand_atomic_store): Ditto.
12208 * doc/md.texi (Standard Pattern Names For Generation):
12209 Document memory_blockage instruction pattern.
12211 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12213 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12214 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12215 * config/rl78/rl78.md: New define_expand "adddi3".
12217 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12219 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12222 2017-10-13 Jeff Law <law@redhat.com>
12224 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12226 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12229 * internal-fn.c (expand_mul_overflow): If both operands have
12230 the same highpart of -1 or 0 and the topmost bit of lowpart
12231 is different, overflow is if res <= 0 rather than res < 0.
12233 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12235 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12236 TARGET_P9_VECTOR code for unaligned_load case.
12238 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12240 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12242 2017-10-13 Nathan Sidwell <nathan@acm.org>
12244 * tree-core.h (tree_contains_struct): Make bool.
12245 * tree.c (tree_contains_struct): Likewise.
12246 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12247 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12248 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12249 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12250 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12252 2017-10-13 Richard Biener <rguenther@suse.de>
12254 * graphite-isl-ast-to-gimple.c
12255 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12256 parameters and dominance check.
12257 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12258 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12259 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12260 Do not update SSA form here or do intermediate IL verification.
12261 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12262 (graphite_initialize): Remove check on the number of loops in
12263 the function and inline into graphite_transform_loops.
12264 (graphite_finalize): Inline into graphite_transform_loops.
12265 (graphite_transform_loops): Perform SSA update and IL verification
12267 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12269 2017-10-13 Richard Biener <rguenther@suse.de>
12271 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12272 graphite_expression_type_precision): Avoid global constructor
12274 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12275 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12276 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12277 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12278 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12279 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12281 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12284 * config/i386/i386.h (ix86_red_zone_size): New.
12285 * config/i386/i386.md (push peephole2s): Replace
12286 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12288 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12289 Alan Hayward <alan.hayward@arm.com>
12290 David Sherwood <david.sherwood@arm.com>
12292 * combine.c (can_change_dest_mode): Reject changes in
12293 REGMODE_NATURAL_SIZE.
12295 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12296 Alan Hayward <alan.hayward@arm.com>
12297 David Sherwood <david.sherwood@arm.com>
12299 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12300 (expand_debug_source_expr): Likewise.
12301 * combine.c (combine_simplify_rtx): Likewise.
12302 * cse.c (fold_rtx): Likewise.
12303 * fwprop.c (canonicalize_address): Likewise.
12304 * targhooks.c (default_shift_truncation_mask): Likewise.
12306 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12307 Alan Hayward <alan.hayward@arm.com>
12308 David Sherwood <david.sherwood@arm.com>
12310 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12311 (widened_mode): Likewise.
12312 (expand_unop): Likewise.
12313 * ree.c (transform_ifelse): Likewise.
12314 (merge_def_and_ext): Likewise.
12315 (combine_reaching_defs): Likewise.
12316 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12318 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12319 Alan Hayward <alan.hayward@arm.com>
12320 David Sherwood <david.sherwood@arm.com>
12322 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12323 * combine.c (gen_lowpart_for_combine): Likewise.
12324 * dwarf2out.c (rtl_for_decl_location): Likewise.
12325 * final.c (alter_subreg): Likewise.
12326 * rtlhooks.c (gen_lowpart_general): Likewise.
12327 (gen_lowpart_if_possible): Likewise.
12329 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12330 Alan Hayward <alan.hayward@arm.com>
12331 David Sherwood <david.sherwood@arm.com>
12333 * calls.c (expand_call): Use subreg_lowpart_offset.
12334 * cse.c (cse_insn): Likewise.
12335 * regcprop.c (copy_value): Likewise.
12336 (copyprop_hardreg_forward_1): Likewise.
12338 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12341 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12342 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12343 =Q constraints to +Q and into insn condition add check
12344 that operands[0] and operands[1] are equal.
12345 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12346 =Q constraints to +Q and into insn condition add check
12347 that operands[0] is equal to either operands[1] or operands[2].
12350 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12351 instead of handling MINUS_EXPR twice (once for each argument),
12352 canonicalize operand order and handle just once, use rtype where
12353 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12356 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12357 any values of __C while still being pattern recognizable as a simple
12358 rotate instruction.
12360 2017-10-13 Richard Biener <rguenther@suse.de>
12362 PR tree-optimization/82451
12364 2017-10-02 Richard Biener <rguenther@suse.de>
12366 PR tree-optimization/82355
12367 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12368 a mapping for the enclosing loop but avoid generating one for
12369 the loop tree root.
12370 (copy_bb_and_scalar_dependences): Remove premature codegen
12371 error on PHIs in blocks duplicated into multiple places.
12372 * graphite-scop-detection.c
12373 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12374 in the region use it as loop and nest to analyze the DR in.
12375 (try_generate_gimple_bb): Likewise.
12376 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12377 (add_loop_constraints): For blocks in a loop not in the region
12378 create a dimension with a single iteration.
12379 * sese.h (gbb_loop_at_index): Remove assert.
12381 * cfgloop.c (loop_preheader_edge): For the loop tree root
12382 return the single successor of the entry block.
12383 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12384 Reset the SCEV hashtable and niters.
12385 * graphite-scop-detection.c
12386 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12387 assert that we only have POLYNOMIAL_CHREC that vary in loops
12388 contained in the region.
12389 (scop_detection::graphite_can_represent_expr): Adjust.
12390 (scop_detection::stmt_has_simple_data_refs_p): For loops
12391 not in the region set loop to NULL. The nest is now the
12392 entry edge to the region.
12393 (try_generate_gimple_bb): Likewise.
12394 * sese.c (scalar_evolution_in_region): Adjust for
12395 instantiate_scev change.
12396 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12397 Make nest parameter the edge into the region.
12398 (create_data_ref): Likewise.
12399 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12400 entry edge into a region and adjust instantiate_scev calls.
12401 (create_data_ref): Likewise.
12402 (graphite_find_data_references_in_stmt): Likewise.
12403 (find_data_references_in_stmt): Pass the loop preheader edge
12404 from the nest argument.
12405 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12406 parameter the edge into the region.
12407 (instantiate_parameters): Use the loop preheader edge as entry.
12408 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12410 (get_instantiated_value_entry): Make instantiate_below parameter
12411 the edge into the region.
12412 (instantiate_scev_name): Likewise. Adjust dominance checks,
12413 when we cannot use loop-based instantiation instantiate by
12414 walking use-def chains.
12415 (instantiate_scev_poly): Adjust.
12416 (instantiate_scev_binary): Likewise.
12417 (instantiate_scev_convert): Likewise.
12418 (instantiate_scev_not): Likewise.
12419 (instantiate_array_ref): Remove.
12420 (instantiate_scev_3): Likewise.
12421 (instantiate_scev_2): Likewise.
12422 (instantiate_scev_1): Likewise.
12423 (instantiate_scev_r): Do not blindly handle N-operand trees.
12424 Do not instantiate array-refs. Handle all constants and invariants.
12425 (instantiate_scev): Make instantiate_below parameter
12426 the edge into the region.
12427 (resolve_mixers): Use the loop preheader edge for the region
12428 parameter to instantiate_scev_r.
12429 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12431 2017-10-13 Richard Biener <rguenther@suse.de>
12433 PR tree-optimization/82525
12434 * graphite-isl-ast-to-gimple.c
12435 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12437 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12438 Fail code generation when we cannot represent the isl integer.
12439 (binary_op_to_tree): Elide modulo operations that are no-ops
12440 in the type we code generate. Remove now superfluous code
12443 2017-10-13 Richard Biener <rguenther@suse.de>
12445 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12446 (scop_detection::harmful_loop_in_region): Remove premature
12447 IV type restriction.
12448 (scop_detection::graphite_can_represent_scev): We can handle
12449 pointer IVs just fine.
12451 2017-10-13 Alan Modra <amodra@gmail.com>
12453 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12454 "Clobbers and Scratch Registers". Add paragraph on
12455 alternative to clobbers for scratch registers and OpenBLAS
12458 2017-10-13 Alan Modra <amodra@gmail.com>
12460 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12461 example of a memory input for a string of known length. Move
12462 commentary out of table. Add a number of new examples
12463 covering array memory inputs.
12465 2017-10-12 Martin Liska <mliska@suse.cz>
12467 PR tree-optimization/82493
12468 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12469 (test_range_functions): New function.
12470 (sbitmap_c_tests): Likewise.
12471 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12472 * selftest.h (sbitmap_c_tests): New function.
12474 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12476 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12478 * config/rs6000/amo.h: Fix spacing issue.
12480 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12483 * config/i386/i386.md (*ashl<mode>3_mask_1,
12484 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12485 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12488 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12490 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12491 (profile_probability): Set max_probability
12492 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12494 * profile-count.c (profile_probability::differs_from_p): Do not
12495 rely on max_probaiblity == 10000
12497 2017-10-12 Jeff Law <law@redhat.com>
12499 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12502 2017-10-12 Martin Sebor <msebor@redhat.com>
12506 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12507 (handle_alias_pairs): Call it.
12508 * common.opt (-Wattribute-alias): New option.
12509 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12510 * doc/invoke.texi (-Wattribute-alias): Document.
12512 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12515 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12517 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12519 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12520 (make_hard_regno_born, make_hard_regno_dead): Update
12521 bb_killed_pseudos and bb_gen_pseudos.
12523 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12525 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12527 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12529 * config/alpha/alpha.c (alpha_split_conditional_move):
12530 Use std::swap instead of manually swapping.
12531 (alpha_stdarg_optimize_hook): Ditto.
12532 (alpha_canonicalize_comparison): Ditto.
12534 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12536 * tree-loop-distribution.c (struct builtin_info): New struct.
12537 (struct partition): Refactor fields into struct builtin_info.
12538 (partition_free): Free struct builtin_info.
12539 (build_size_arg_loc, build_addr_arg_loc): Delete.
12540 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12541 information from struct builtin_info.
12542 (find_single_drs): New function refactored from classify_partition.
12543 Also moved builtin validity checks to this function.
12544 (compute_access_range, alloc_builtin): New functions.
12545 (classify_builtin_st, classify_builtin_ldst): New functions.
12546 (classify_partition): Refactor code into functions find_single_drs,
12547 classify_builtin_st and classify_builtin_ldst.
12548 (distribute_loop): Don't do runtime alias check when distributing
12550 (find_seed_stmts_for_distribution): New function.
12551 (pass_loop_distribution::execute): Refactor code finding seed
12552 stmts into above function. Support distribution for the innermost
12553 two-level loop nest. Adjust dump information.
12555 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12557 * tree-loop-distribution.c: Adjust the general comment.
12558 (NUM_PARTITION_THRESHOLD): New macro.
12559 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12560 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12561 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12562 in call to build_partition_graph.
12563 (finalize_partitions): New parameter. Make loop distribution more
12564 conservative by fusing more partitions.
12565 (distribute_loop): Don't do runtime alias check in case of loop nest
12567 (find_seed_stmts_for_distribution): New function.
12568 (prepare_perfect_loop_nest): New function.
12569 (pass_loop_distribution::execute): Refactor code finding seed stmts
12570 and loop nest into above functions. Support loop nest distribution.
12571 Adjust dump information accordingly.
12573 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12575 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12576 and set PTYPE_SEQUENTIAL for merged partition.
12578 2017-10-12 Richard Biener <rguenther@suse.de>
12580 PR tree-optimization/69728
12582 2017-09-19 Richard Biener <rguenther@suse.de>
12584 PR tree-optimization/69728
12585 * graphite-sese-to-poly.c (schedule_error): New global.
12586 (add_loop_schedule): Handle empty domain by failing the
12588 (build_original_schedule): Handle schedule_error.
12590 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12591 domain by returning an unchanged schedule.
12593 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12595 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12596 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12598 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12600 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12603 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12606 * expr.c (store_field): Don't optimize away bitsize == 0 store
12607 from CALL_EXPR with addressable return type.
12609 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12611 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12612 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12613 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12614 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12615 TARGET_ISEL instead of TARGET_ISEL<sel>.
12617 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12619 * config/rs6000/rs6000.c
12620 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12622 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12624 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12626 (reg_or_cint_operand): Fix comment.
12627 (reg_or_zero_operand): New predicate.
12628 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12629 * config/rs6000/rs6000.c (output_isel): Delete.
12630 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12631 instead of reg_or_cint_operand. Output instruction directly (not via
12633 (isel_unsigned_<mode>): Ditto.
12634 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12635 gpc_reg_operand. Add an instruction alternative for this. Output
12636 instruction directly.
12637 (*isel_reversed_unsigned_<mode>): Ditto.
12639 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12641 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12642 (TARGET_CANONICALIZE_COMPARISON): Define.
12644 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12647 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12648 Check whether the dest is REG before adding REG_EQUIV note.
12650 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12653 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12655 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12656 (make_hard_regno_born, make_hard_regno_dead): Update
12657 bb_killed_pseudos and bb_gen_pseudos.
12659 2017-10-11 Nathan Sidwell <nathan@acm.org>
12661 * incpath.h (enum incpath_kind): Name enum, prefix values.
12662 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12663 * incpath.c (heads, tails): Use INC_MAX.
12664 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12665 (merge_include_chains, split_quote_chain,
12666 register_include_chains): Update incpath_kind names.
12667 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12668 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12670 (add_framework_path, darwin_register_objc_includes): Likewise.
12671 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12673 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12675 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12676 Do not use float_operator operator predicate.
12677 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12678 * config/i386/predicates.md (float_operator): Remove predicate.
12680 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12682 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12683 (*jccxf_i387): Ditto.
12684 (*jcc<mode>_i387): Ditto.
12685 (*jccu<mode>_i387): Ditto.
12686 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12687 (*jcc_*_i387 splitters): Remove.
12688 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12689 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12690 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12693 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12695 * profile-count.h (slow_safe_scale_64bit): New function.
12696 (safe_scale_64bit): New inline.
12697 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12698 * profile-count.c: Include wide-int.h
12699 (slow_safe_scale_64bit): New.
12701 2017-10-11 Nathan Sidwell <nathan@acm.org>
12703 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12704 HAS_DECL_ASSEMBLER_NAME_P.
12705 * gimple-expr.c (gimple_decl_printable_name: Check
12706 HAS_DECL_ASSEMBLER_NAME_P too.
12707 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12708 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12709 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12710 * passes.c (rest_of_decl_compilation): Check
12711 HAS_DECL_ASSEMBLER_NAME_P too.
12712 * recog.c (verify_changes): Likewise.
12713 * tree-pretty-print.c (dump_decl_name): Likewise.
12714 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12716 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12717 (SET_DECL_ASSEMBLER_NAME): Use it.
12718 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12719 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12720 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12722 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12724 * config.gcc (i386, x86_64): Add extra objects.
12725 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12726 (ix86_min_insn_size): Declare.
12727 (ix86_issue_rate): Declare.
12728 (ix86_adjust_cost): Declare.
12729 (ia32_multipass_dfa_lookahead): Declare.
12730 (ix86_macro_fusion_p): Declare.
12731 (ix86_macro_fusion_pair_p): Declare.
12732 (ix86_bd_has_dispatch): Declare.
12733 (ix86_bd_do_dispatch): Declare.
12734 (ix86_core2i7_init_hooks): Declare.
12735 (ix86_atom_sched_reorder): Declare.
12736 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12737 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12738 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12739 (rip_relative_addr_p): Rename to ...
12740 (ix86_rip_relative_addr_p): ... this one; export.
12741 (memory_address_length): Update.
12742 (ix86_issue_rate): Move to x86-tune-sched.c.
12743 (ix86_flags_dependent): Move to x86-tune-sched.c.
12744 (ix86_agi_dependent): Move to x86-tune-sched.c.
12745 (exact_dependency_1): Move to x86-tune-sched.c.
12746 (exact_store_load_dependency): Move to x86-tune-sched.c.
12747 (ix86_adjust_cost): Move to x86-tune-sched.c.
12748 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12749 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12750 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12751 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12752 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12753 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12754 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12755 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12756 (min_insn_size): Rename to ...
12757 (ix86_min_insn_size): ... this one; export.
12758 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12759 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12760 (core2i7_first_cycle_multipass_backtrack): Move to
12761 x86-tune-sched-core.c.
12762 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12763 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12764 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12765 (ix86_avoid_jump_mispredicts): Update.
12766 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12767 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12768 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12769 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12770 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12771 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12772 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12773 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12774 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12775 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12776 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12777 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12778 (BIG): Move to ix86-tune-sched-bd.c.
12779 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12780 (enum insn_path): Move to ix86-tune-sched-bd.c.
12781 (get_mem_group): Move to ix86-tune-sched-bd.c.
12782 (is_cmp): Move to ix86-tune-sched-bd.c.
12783 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12784 (is_branch): Move to ix86-tune-sched-bd.c.
12785 (is_prefetch): Move to ix86-tune-sched-bd.c.
12786 (init_window): Move to ix86-tune-sched-bd.c.
12787 (allocate_window): Move to ix86-tune-sched-bd.c.
12788 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12789 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12790 (process_end_window): Move to ix86-tune-sched-bd.c.
12791 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12792 (find_constant): Move to ix86-tune-sched-bd.c.
12793 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12794 (has_immediate): Move to ix86-tune-sched-bd.c.
12795 (get_insn_path): Move to ix86-tune-sched-bd.c.
12796 (get_insn_group): Move to ix86-tune-sched-bd.c.
12797 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12798 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12799 (add_insn_window): Move to ix86-tune-sched-bd.c.
12800 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12801 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12802 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12803 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12804 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12805 (do_dispatch): Move to ix86-tune-sched-bd.c.
12806 (has_dispatch): Move to ix86-tune-sched-bd.c.
12807 * config/i386/t-i386: Add new object files.
12808 * config/i386/x86-tune-costs.h: New file.
12809 * config/i386/x86-tune-sched-atom.c: New file.
12810 * config/i386/x86-tune-sched-bd.c: New file.
12811 * config/i386/x86-tune-sched-core.c: New file.
12812 * config/i386/x86-tune-sched.c: New file.
12814 2017-10-11 Liu Hao <lh_mouse@126.com>
12816 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12817 the generic version below instead.
12818 (should_colorize): Recognize Windows consoles as terminals
12820 * pretty-print.c [__MINGW32__] (write_all): New function.
12821 [__MINGW32__] (find_esc_head): Likewise.
12822 [__MINGW32__] (find_esc_terminator): Likewise.
12823 [__MINGW32__] (eat_esc_sequence): Likewise.
12824 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12826 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12829 2017-10-11 Richard Biener <rguenther@suse.de>
12831 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12832 Properly call analyze_scalar_evolution with the loop of the stmt.
12834 2017-10-11 Richard Biener <rguenther@suse.de>
12836 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12837 * tree-core.h (tree_base): Add chrec_var union member.
12838 * tree.h (CHREC_VAR): Remove.
12839 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12840 * tree-chrec.h (build_polynomial_chrec): Adjust.
12841 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12842 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12844 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12846 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12847 * match.pd: ... here.
12848 ((T) X == (T) Y): Relax condition.
12850 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12852 PR tree-optimization/82472
12853 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12855 (break_alias_scc_partitions): Update postorder number.
12857 2017-10-11 Martin Liska <mliska@suse.cz>
12860 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12862 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12864 2017-10-11 Martin Liska <mliska@suse.cz>
12866 * print-rtl.c (print_insn): Move declaration of idbuf
12867 to same scope as name.
12869 2017-10-11 Martin Liska <mliska@suse.cz>
12874 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12875 volatile arguments.
12877 2017-10-11 Martin Liska <mliska@suse.cz>
12880 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12881 volatile arguments.
12883 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12885 * config.gcc (default_gnu_indirect_function): Default to yes for
12886 arm*-*-linux* with glibc.
12888 2017-10-11 Richard Biener <rguenther@suse.de>
12890 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12891 default-defs and types we do not want to analyze.
12892 (interpret_loop_phi): Replace unreachable code with an assert.
12893 (compute_scalar_evolution_in_loop): Remove and inline ...
12894 (analyze_scalar_evolution_1): ... here, replacing condition with
12895 what makes the intent clearer. Remove handling of cases
12896 get_scalar_evolution now handles.
12898 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12900 PR rtl-optimization/81434
12901 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12902 comment for main loop. In sched_group_found if, also add checks for
12903 pass and min_cost_group.
12905 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12907 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12908 (rs6000_insn_cost): New function.
12909 * config/rs6000/rs6000.md (cost): New attribute.
12911 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12912 H.J. Lu <hongjiu.lu@intel.com>
12916 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12917 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12918 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12919 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12920 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12921 OPTION_MASK_ISA_MMX and some other bit set require both
12922 mmx and the ISAs without the mmx bit.
12923 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12924 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12925 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12926 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12927 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12928 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12929 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12930 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12931 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12932 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12933 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12934 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12935 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12936 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12937 Add OPTION_MASK_ISA_MMX.
12939 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12941 * config.gcc (armv7*-*-freebsd*): New target.
12942 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12944 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12946 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12947 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12950 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12952 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12953 a template parameter.
12954 (WIDE_INT_REF_FOR): Update accordingly.
12955 * tree.h (wi::int_traits <const_tree>): Delete.
12956 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12957 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12958 (wi::tree_to_wide_ref): New typedef.
12959 (wi::to_wide): New function.
12960 * calls.c (get_size_range): Use wi::to_wide when operating on
12961 trees as wide_ints.
12962 * cgraph.c (cgraph_node::create_thunk): Likewise.
12963 * config/i386/i386.c (ix86_data_alignment): Likewise.
12964 (ix86_local_alignment): Likewise.
12965 * dbxout.c (stabstr_O): Likewise.
12966 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12967 * expr.c (const_vector_from_tree): Likewise.
12968 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12969 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12970 (fold_negate_expr_1, int_const_binop_1, const_binop)
12971 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12972 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12973 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12974 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12975 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12976 (fold_not_const, round_up_loc): Likewise.
12977 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12978 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12979 (alloca_call_type): Likewise.
12980 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12981 * godump.c (go_output_typedef): Likewise.
12982 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12983 * internal-fn.c (get_min_precision): Likewise.
12984 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12985 * ipa-polymorphic-call.c
12986 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12987 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12988 (ipa_modify_call_arguments): Likewise.
12989 * match.pd: Likewise.
12990 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12991 * print-tree.c (print_node_brief, print_node): Likewise.
12992 * stmt.c (expand_case): Likewise.
12993 * stor-layout.c (layout_type): Likewise.
12994 * tree-affine.c (tree_to_aff_combination): Likewise.
12995 * tree-cfg.c (group_case_labels_stmt): Likewise.
12996 * tree-data-ref.c (dr_analyze_indices): Likewise.
12997 (prune_runtime_alias_test_list): Likewise.
12998 * tree-dump.c (dequeue_and_dump): Likewise.
12999 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13000 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13001 * tree-pretty-print.c (dump_generic_node): Likewise.
13002 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13003 (simple_iv_with_niters): Likewise.
13004 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13005 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13006 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13007 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13008 (refine_value_range_using_guard, number_of_iterations_ne_max)
13009 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13010 (get_cst_init_from_scev, record_nonwrapping_iv)
13011 (scev_var_range_cant_overflow): Likewise.
13012 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13013 * tree-ssa-pre.c (compute_avail): Likewise.
13014 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13015 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13016 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13017 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13018 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13019 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13020 (dump_case_nodes, try_switch_expansion): Likewise.
13021 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13022 (vect_do_peeling): Likewise.
13023 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13024 * tree-vect-stmts.c (vectorizable_load): Likewise.
13025 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13026 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13027 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13028 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13029 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13030 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13031 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13032 (evrp_dom_walker::before_dom_children): Likewise.
13033 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13034 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13035 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13036 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13037 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13038 (get_range_pos_neg): Likewise.
13039 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13040 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13041 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13042 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13043 * config/avr/avr.c (avr_fold_builtin): Likewise.
13044 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13045 * config/msp430/msp430.c (msp430_attr): Likewise.
13046 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13047 * config/powerpcspe/powerpcspe-c.c
13048 (altivec_resolve_overloaded_builtin): Likewise.
13049 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13050 (rs6000_expand_ternop_builtin): Likewise.
13051 * config/rs6000/rs6000-c.c
13052 (altivec_resolve_overloaded_builtin): Likewise.
13053 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13054 (rs6000_expand_ternop_builtin): Likewise.
13055 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13057 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13059 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13060 when copying loop nest with only one inner loop.
13062 2017-10-10 Richard Biener <rguenther@suse.de>
13064 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13065 blocks if SCEV is active.
13066 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13068 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13069 (scev_initialize): Assert we are not yet initialized.
13071 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13073 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13074 inner loop's exit stmt by making it always exit the loop, otherwise
13075 we would generate an infinite empty loop.
13077 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13079 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13080 renaming variables in new preheader if it's deleted.
13082 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13084 * tree-loop-distribution.c (struct partition): Remove unused field
13085 loops of the structure.
13086 (partition_alloc, partition_free): Ditto.
13087 (build_rdg_partition_for_vertex): Ditto.
13089 2017-10-09 Jeff Law <law@redhat.com>
13091 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13092 return type to match prototype and documentation.
13094 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13096 * config/rs6000/rs6000.c (processor_costs): Move to ...
13097 * config/rs6000/rs6000.h: ... here.
13098 (rs6000_cost): Declare.
13100 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13102 * except.c (setjmp_fn): New global variable.
13103 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13104 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13105 if DONT_USE_BUILTIN_SETJMP is defined.
13107 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13109 * target.def (insn_cost): New hook.
13110 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13111 * doc/tm.texi: Regenerate.
13112 * rtlanal.c (insn_cost): Use the new hook.
13114 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13116 * combine.c (combine_validate_cost): Compute the new insn_cost,
13117 not just pattern_cost.
13118 (try_combine): Adjust comment.
13120 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13122 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13124 * combine.c (uid_insn_cost): Adjust comment.
13125 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13127 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13128 * dse.c (find_shift_sequence): Ditto.
13129 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13130 (bb_valid_for_noce_process_p): Use pattern_cost.
13131 * rtl.h (insn_rtx_cost): Delete.
13132 (pattern_cost): New prototype.
13133 (insn_cost): New prototype.
13134 * rtlanal.c (insn_rtx_cost): Rename to...
13135 (pattern_cost): ... this.
13138 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13140 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13141 (*jcc<mode>_0_r_i387): Ditto.
13142 (*jccxf_r_i387): Ditto.
13143 (*jcc<mode>_r_i387): Ditto.
13144 (*jccu<mode>_r_i387): Ditto.
13145 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13146 (*jcc): Rename from *jcc_1.
13148 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13150 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13151 deferred rescans after the lvx/stvx recombination pre-pass.
13153 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13155 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13156 memory operation instruction support.
13157 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13158 (rs6000-ibm-aix[789]*): Likewise.
13159 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13160 Document new functions.
13162 2017-10-09 Richard Biener <rguenther@suse.de>
13164 PR tree-optimization/82397
13165 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13166 equality only for semantically equal trees.
13168 2017-10-09 Richard Biener <rguenther@suse.de>
13170 PR tree-optimization/82449
13171 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13172 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13173 allow constant addresses.
13174 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13177 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13179 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13182 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13185 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13188 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13191 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13193 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13196 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13197 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13199 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13201 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13202 (WI_BINARY_PREDICATE_RESULT): Likewise.
13203 (wi::binary_traits::operator_result): New type.
13204 (wi::binary_traits::predicate_result): Likewise.
13205 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13206 (generic_wide_int::operator==, generic_wide_int::operator!=)
13207 (generic_wide_int::operator&, generic_wide_int::and_not)
13208 (generic_wide_int::operator|, generic_wide_int::or_not)
13209 (generic_wide_int::operator^, generic_wide_int::operator+
13210 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13212 (operator~, unary operator-, operator==, operator!=, operator&)
13213 (operator|, operator^, operator+, binary operator-, operator*): New
13215 * expr.c (get_inner_reference): Use wi::bit_and_not.
13216 * fold-const.c (fold_binary_loc): Likewise.
13217 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13218 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13219 (bit_value_binop): Likewise.
13220 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13221 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13222 (extract_range_from_binary_expr_1): Likewise.
13223 (masked_increment): Likewise.
13224 (simplify_bit_ops_using_ranges): Likewise.
13226 2017-10-09 Martin Jambor <mjambor@suse.cz>
13229 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13230 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13231 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13233 (hsa_fixup_mov_insn_type): New function.
13234 (hsa_op_with_type::get_in_type): Use it.
13235 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13236 immediates in an assert.
13237 (hsa_op_with_type::extend_int_to_32bit): New method.
13238 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13239 types. Convert to dest type if necessary.
13240 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13241 (reg_for_gimple_ssa): Pass false as min32int to
13242 hsa_type_for_scalar_tree_type.
13243 (gen_hsa_addr): Fixup type when creating addresable temporary.
13244 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13245 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13246 necessary. Call hsa_fixup_mov_insn_type.
13247 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13248 extend operands and convert to dest type if necessary.
13249 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13250 to dest type if necessary.
13251 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13252 if conversion nt necessary and size matches.
13253 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13254 to dest type if necessary.
13255 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13256 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13258 (gen_hsa_clrsb): Likewise.
13259 (gen_hsa_ffs): Likewise.
13260 (gen_hsa_divmod): Extend operands and convert to dest type if
13262 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13264 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13266 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13267 Remove empty default arguments. Use a brace block as output
13269 (conditional return): Ditto.
13271 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13272 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13274 (group_ending_nop): Ditto.
13275 (doloop_end): Ditto.
13276 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13277 (splitters for those): Ditto.
13279 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13281 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13282 a conditional jump (and the compare for it) so that pc_rtx is the
13284 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13285 for the deleted and renamed ctr<mode>_internal[234] patterns.
13286 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13287 Delete second conditional return pattern.
13288 (ctr<mode>_internal2): Delete this second bdnz pattern.
13289 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13290 (ctr<mode>_internal4): Delete this second bdz pattern.
13292 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13294 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13295 (always_initialized_rtx_for_ssa_name_p): New predicate.
13296 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13297 (finish_out_of_ssa): Free new field of SA.
13298 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13299 * tree-ssa-coalesce.c: Include tree-ssa.h.
13300 (get_parm_default_def_partitions): Remove extern keyword.
13301 (get_undefined_value_partitions): New function.
13302 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13303 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13304 uninitialized bits.
13305 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13307 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13309 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13311 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13313 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13314 for targets that preffer 128bit.
13316 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13318 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13320 2017-10-08 Olivier Hainque <hainque@adacore.com>
13322 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13323 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13325 (thumb_set_return_address): Likewise.
13327 2017-10-08 Olivier Hainque <hainque@adacore.com>
13329 * common/config/arm/arm-common.c (arm_except_unwind_info):
13330 Handle DWARF2_UNWIND_INFO.
13332 2017-10-07 Michael Collison <michael.collison@arm.com>
13334 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13337 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13339 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13340 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13341 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13342 defined, force the creation of a new block for a dispatch label.
13344 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13346 * invoke.texi (Wsuggest-attribute=cold): Document.
13347 * common.opt (Wsuggest-attribute=cold): New
13348 * ipa-pure-const.c (warn_function_cold): New function.
13349 * predict.c (compute_function_frequency): Use it.
13350 * predict.h (warn_function_cold): Declare.
13352 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13354 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13357 2017-10-06 Martin Liska <mliska@suse.cz>
13359 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13360 keyword for member functions.
13361 (struct sanopt_tree_couple): New struct.
13362 (struct sanopt_tree_couple_hash): New function.
13363 (struct sanopt_ctx): Add new hash_map.
13364 (has_dominating_ubsan_ptr_check): New function.
13365 (record_ubsan_ptr_check_stmt): Likewise.
13366 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13367 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13368 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13370 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13373 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13374 aarch64_simd_valid_immediate on CONST_VECTORs.
13375 (aarch64_reg_or_bic_imm): Likewise.
13377 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13379 PR rtl-optimization/82396
13380 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13382 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13384 * graphite-dependences.c (scop_get_reads): Move code to...
13385 (scop_get_must_writes): Move code to...
13386 (scop_get_may_writes): Move code to...
13387 (scop_get_reads_and_writes): ... here.
13388 (scop_get_dependences): Call scop_get_reads_and_writes.
13390 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13392 PR tree-optimization/82434
13393 * fold-const.h (can_native_encode_type_p,
13394 can_native_encode_string_p): Remove.
13395 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13396 don't encode anything, just return what would be otherwise returned.
13397 (native_encode_fixed, native_encode_complex, native_encode_vector):
13399 (native_encode_string): Likewise. Inline by hand
13400 can_native_encode_string_p.
13401 (can_native_encode_type_p): Remove.
13402 (can_native_encode_string_p): Remove.
13403 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13404 STRING_CSTs using can_native_encode_string_p, test all
13405 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13406 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13407 argument from native_encode_expr.
13408 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13409 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13412 2017-10-06 Richard Biener <rguenther@suse.de>
13414 PR tree-optimization/82397
13415 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13416 operand_equal_p but rely on data_ref_compare_tree for detecting
13418 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13419 to match up with dr_group_sort_cmp.
13421 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13424 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13426 * config/s390/s390-builtin-types.def: Regenerate.
13428 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13431 * config/s390/s390-builtin-types.def: Regenerate.
13432 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13433 Change flag from B_VXE to B_VX.
13434 (s390_vec_min_dbl): Remove B_VXE flag.
13436 2017-10-06 Richard Biener <rguenther@suse.de>
13438 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13439 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13440 translate_isl_ast_to_gimple::is_valid_rename,
13441 translate_isl_ast_to_gimple::get_rename,
13442 translate_isl_ast_to_gimple::get_def_bb_for_const,
13443 translate_isl_ast_to_gimple::get_new_name,
13444 translate_isl_ast_to_gimple::collect_all_ssa_names,
13445 translate_isl_ast_to_gimple::copy_loop_phi_args,
13446 translate_isl_ast_to_gimple::collect_all_ssa_names,
13447 translate_isl_ast_to_gimple::copy_loop_phi_args,
13448 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13449 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13450 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13451 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13452 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13453 translate_isl_ast_to_gimple::copy_cond_phi_args,
13454 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13455 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13456 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13457 translate_isl_ast_to_gimple::rename_uses,
13458 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13459 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13460 (set_rename_for_each_def): Likewise.
13461 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13462 here. Handle rewriting SCEV analyzable uses here.
13463 (copy_bb_and_scalar_dependences): Generate code for PHI
13465 (graphite_regenerate_ast_isl): Adjust.
13466 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13467 (add_write, add_read): New functions.
13468 (build_cross_bb_scalars_def): Use it and simplify.
13469 (build_cross_bb_scalars_use): Likewise.
13470 (graphite_find_cross_bb_scalar_vars): Inline into...
13471 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13472 simulating out-of-SSA. Compute liveout and add dependencies.
13473 (build_scops): Force an empty entry block.
13474 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13476 (sese_build_liveouts): Declare.
13477 (sese_trivially_empty_bb_p): Likewise.
13478 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13479 compute liveout and debug_liveout.
13480 (sese_bad_liveouts_use): Remove.
13481 (sese_reset_debug_liveouts_bb): Likewise.
13482 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13483 (sese_build_liveouts): Build liveout and debug_liveout and store
13485 (new_sese_info): Adjust.
13486 (free_sese_info): Likewise.
13487 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13488 do not build liveout here.
13489 (move_sese_in_condition): Adjust region entry.
13490 (scev_analyzable_p): Match up with chrec_apply requirements.
13491 (sese_trivially_empty_bb_p): New.
13492 * tree-into-ssa.c (get_reaching_def): Properly support generating
13493 default-defs for incremental rewrite of anonymous names.
13495 2017-10-06 Richard Biener <rguenther@suse.de>
13497 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13498 precision do not perform modulo reduction.
13500 2017-10-06 Richard Biener <rguenther@suse.de>
13502 PR tree-optimization/82436
13503 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13504 conservatively choose the vectorization factor when checking
13505 whether we can perform the required load permutation.
13506 (vect_transform_slp_perm_load): Assert when we may not fail.
13508 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13510 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13511 message for incompatible -msdata=* and -mcall-* options.
13513 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13515 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13516 rate for post-reload scheduling.
13518 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13520 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13522 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13524 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13525 to improve monte carlo in scimark.
13527 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13529 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13530 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13531 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13532 pentium4_cost, nocona_cost): Set reassociation width to 1.
13533 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13534 width to 2 for fp operations and 1 otherwise.
13535 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13537 (atom_cost): Set reassociation width to 2.
13538 (slm_cost, generic_cost): Set fp reassociation width
13539 to 2 and 1 otherwise.
13540 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13541 (core_cost): Set fp reassociation width to 4 and vector to 2.
13542 (ix86_reassociation_width): Rewrite using cost table; special case
13543 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13544 and TARGET_AVX128_OPTIMAL.
13545 * config/i386/i386.h (processor_costs): Add
13546 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13547 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13548 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13549 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13550 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13551 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13553 2017-10-05 Nathan Sidwell <nathan@acm.org>
13555 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13557 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13559 * config/arm/arm.c (arm_test_fpu_data): New.
13560 (arm_run_selftests): Call arm_test_fpu_data.
13562 2017-10-04 Nathan Sidwell <nathan@acm.org>
13564 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13566 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13568 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13570 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13571 check type for aarch64_simd_valid_immediate.
13572 (aarch64_output_simd_mov_immediate): Update prototype.
13573 (aarch64_simd_valid_immediate): Update prototype.
13574 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13575 support for ORR-immediate.
13576 (and<mode>3): modified pattern to add support for BIC-immediate.
13577 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13578 now checks for valid immediate for BIC and ORR based on new enum
13580 (aarch64_output_simd_mov_immediate): Function now used to output
13581 BIC/ORR imm as well based on new enum argument.
13582 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13584 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13585 (aarch64_reg_or_bic_imm): Likewise.
13587 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13589 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13590 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13592 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13596 PR rtl-optimization/82396
13597 * haifa-sched.c (autopref_multipass_init): Simplify
13599 (autopref_rank_data): Simplify sort order.
13600 * sched-int.h (autopref_multipass_data_): Remove
13601 multi_mem_insn_p, min_offset and max_offset.
13603 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13605 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13607 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13609 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13610 vect_doubleint_cvt.
13612 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13614 * doc/sourcebuild.texi: Document vect_long_mult.
13616 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13618 PR tree-optimization/82413
13619 * fold-const.c (build_range_check): Use widest_int when comparing
13620 the maximum ETYPE value with HIGH.
13622 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13624 PR rtl-optimization/82396
13625 * haifa-sched.c (autopref_multipass_init): Simplify
13627 (autopref_rank_data): Simplify sort order.
13628 * sched-int.h (autopref_multipass_data_): Remove
13629 multi_mem_insn_p, min_offset and max_offset.
13631 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13633 PR tree-optimization/82381
13634 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13635 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13638 PR tree-optimization/82374
13639 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13640 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13641 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13642 current_function_decl to the new decl.
13644 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13646 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13647 helper macro for IEEE float128 hardware built-in functions.
13648 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13650 (TRUNCF128_ODD): Likewise.
13651 (ADDF128_ODD): Likewise.
13652 (SUBF128_ODD): Likewise.
13653 (MULF128_ODD): Likewise.
13654 (DIVF128_ODD): Likewise.
13655 (FMAF128_ODD): Likewise.
13656 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13657 UNSPEC_TRUNC_ROUND_TO_ODD.
13658 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13659 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13660 floating point round to odd instructions.
13661 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13662 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13663 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13664 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13665 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13666 (trunc<mode>sf2_hw): Change the truncate with round to odd
13667 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13668 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13669 to odd hardware instructions.
13670 (sub<mode>3_odd): Likewise.
13671 (mul<mode>3_odd): Likewise.
13672 (div<mode>3_odd): Likewise.
13673 (sqrt<mode>2_odd): Likewise.
13674 (fma<mode>4_odd): Likewise.
13675 (fms<mode>4_odd): Likewise.
13676 (nfma<mode>4_odd): Likewise.
13677 (nfms<mode>4_odd): Likewise.
13678 (trunc<mode>df2_odd): Change the truncate with round to odd
13679 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13681 * doc/extend.texi (PowerPC built-in functions): Update documentation
13682 for existing IEEE float128-bit built-in functions. Add built-in
13683 functions that generate the IEEE 128-bit floating point round to
13686 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13688 PR rtl-optimization/77729
13689 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13690 to (X&(C1&~C2))|C2 transformations.
13692 2017-10-03 Martin Jambor <mjambor@suse.cz>
13694 PR tree-optimization/82363
13695 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13696 mismatch, mark lacc written regardless of racc.
13698 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13700 PR tree-optimization/82381
13701 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13702 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13703 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13704 fallthrough into reassoc_stmt_dominates_stmt_p.
13707 * combine.c (combine_instructions): Don't combine in unreachable
13710 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13713 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13714 function to not use the have_cpu variable. Do not set cpu_index,
13715 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13716 or the default cpu.
13717 (rs6000_valid_attribute_p): Remove duplicate initializations of
13718 old_optimize and func_optimize.
13719 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13720 (rs6000_activate_target_options): Make global.
13721 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13724 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13726 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13727 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13728 Set *poffset to 0 and *psize and *pmax_size to -1 if
13729 *poffset + *psize overflows in HOST_WIDE_INT.
13731 PR tree-optimization/82387
13732 PR tree-optimization/82388
13733 PR tree-optimization/82389
13734 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13735 instead of live_bytes non-NULL.
13737 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13740 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13743 2017-10-02 Richard Biener <rguenther@suse.de>
13745 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13746 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13747 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13749 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13751 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13752 requested precision matches the type's.
13753 * calls.c (alloc_max_size): Calculate the new candidate size as
13754 a widest_int and use wi::to_widest when comparing it with the
13755 current candidate size.
13756 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13757 zero rather than integer_zero_node.
13758 * match.pd: Check for a no-op conversion before using wi::add
13759 rather than after. Use tree_to_uhwi when summing small shift
13760 counts into an unsigned int.
13762 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13763 Alan Hayward <alan.hayward@arm.com>
13764 David Sherwood <david.sherwood@arm.com>
13767 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13768 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13769 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13770 POINTER_AND_FP_REGS.
13772 2017-10-02 Richard Biener <rguenther@suse.de>
13774 PR tree-optimization/82355
13775 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13776 a mapping for the enclosing loop but avoid generating one for
13777 the loop tree root.
13778 (copy_bb_and_scalar_dependences): Remove premature codegen
13779 error on PHIs in blocks duplicated into multiple places.
13780 * graphite-scop-detection.c
13781 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13782 in the region use it as loop and nest to analyze the DR in.
13783 (try_generate_gimple_bb): Likewise.
13784 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13785 (add_loop_constraints): For blocks in a loop not in the region
13786 create a dimension with a single iteration.
13787 * sese.h (gbb_loop_at_index): Remove assert.
13789 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13791 * omp-expand.c (adjust_context_scope): New function.
13792 (expand_parallel_call): Call adjust_context_scope.
13794 2017-10-01 Jeff Law <law@redhat.com>
13796 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13797 dom_opt_dom_walker class with direct access to private members.
13798 Add comments. Call test_for_singularity.
13799 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13800 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13801 m_dummy_cond anymore.
13802 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13804 (pass_dominator:execute): Build the dummy_cond here and pass it
13805 to the dom_opt_dom_walker ctor.
13806 (test_for_singularity): New function.
13808 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13809 Maya Rashish <coypu@sdf.org>
13811 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13812 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13813 (alpha*-*-netbsd*) Use nbsd_tm_file.
13814 (arm*-*-netbsdelf*) Likewise.
13815 (i[34567]86-*-netbsdelf*) Likewise.
13816 (x86_64-*-netbsd*) Likewise.
13817 (mips*-*-netbsd*) Likewise.
13818 (powerpc-*-netbsd*) Likewise.
13819 (sh*-*-netbsd*) Likewise.
13820 (sparc-*-netbsdelf*) Likewise.
13821 (sparc64-*-netbsd*) Likewise.
13822 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13824 (vax-*-netbsdelf*) Likewise.
13825 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13826 (UINT_FAST8_TYPE) Likewise.
13827 (INT_FAST16_TYPE) Check CHAR_FAST16.
13828 (UINT_FAST16_TYPE) Likewise.
13830 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13833 * config/i386/i386.md
13834 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13835 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13836 *divmodsi4_zext_2): New define_insn_and_split.
13837 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13838 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13839 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13840 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13841 New define_insn_and_split.
13842 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13843 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13844 operands[1] having DImode when mode is SImode.
13846 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13847 always SImode for DIV and MOD in REG_EQUAL notes.
13849 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13851 PR middle-end/82319
13852 * match.pd: Fix handling of NaNs in pattern.
13854 2017-09-29 Jeff Law <law@redhat.com>
13856 * sbitmap.c (bitmap_bit_in_range_p): New function.
13857 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13858 * tree-ssa-dse.c (live_bytes_read): New function.
13859 (dse_classify_store): Ignore reads of dead bytes.
13861 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13862 typos and whitespace errors.
13863 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13864 * config/s390/s390.c (s390_emit_prologue): Likewise.
13866 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13869 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13870 with a symbol for LRA.
13872 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13874 PR rtl-optimization/82338
13875 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13877 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13879 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13880 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13881 (qsort_chk): Declare.
13882 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13883 (qsort_chk): New function.
13885 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13887 PR tree-optimization/82337
13888 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13889 phi definition if the PHI result appears in an abnormal PHI.
13890 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13891 basis appears in an abnormal PHI.
13893 2017-09-29 Richard Biener <rguenther@suse.de>
13895 * graphite-isl-ast-to-gimple.c
13896 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13897 (binary_op_to_tree): Use it.
13898 (get_rename_from_scev): Likewise.
13899 (copy_loop_phi_nodes): Likewise.
13900 (copy_bb_and_scalar_dependences): Likewise.
13901 (translate_pending_phi_nodes): Likewise.
13903 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13906 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13907 for movabsq $(i32 << shift), r64.
13909 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13911 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13912 index when encoding %esp as %rsp to avoid 0x67 prefix.
13914 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13916 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13917 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13919 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13921 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13922 Extensions with more than 16 double VFP registers.
13923 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13924 to_clear_mask and all code related to it. Replace the remaining
13925 entry by a sbitmap and adapt code accordingly.
13927 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13929 * brig-builtins.def: Change pure attributes to const.
13931 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13933 * config.gcc (default_gnu_indirect_function): Default to yes for
13934 sparc*-*-linux* with glibc.
13936 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13938 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13939 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13940 when creating .init_array and .fini_array sections with priority
13943 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13946 * config/aarch64/aarch64.c
13947 (aarch64_builtin_support_vector_misalignment): Always return false
13948 when misalignment is unknown.
13950 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13952 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13953 this function to return false if the definition used by the swap
13954 instruction is artificial, or if the memory address from which the
13955 constant value is loaded is not represented by a base address held
13956 in a register or if the base address register is a frame or stack
13957 pointer. Additionally, return false if the base address of the
13958 loaded constant is a SYMBOL_REF but is not considered to be a
13960 (replace_swapped_load_constant): New function.
13961 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13962 loaded constant vector with a load of a swapped constant vector.
13964 2017-09-27 Carl Love <cel@us.ibm.com>
13966 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13967 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13968 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13971 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13973 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13974 first, always call autopref_rank_data otherwise.
13976 2017-09-27 Richard Biener <rguenther@suse.de>
13978 * graphite-scop-detection.c (find_scop_parameters): Move
13979 loop bound handling ...
13980 (gather_bbs::before_dom_children): ... here, avoiding the need
13981 to build scop_info->loop_nest.
13982 (record_loop_in_sese): Remove.
13983 * sese.h (sese_info_t::loop_nest): Remove.
13984 * sese.c (new_sese_info): Do not allocate loop_nest.
13985 (free_sese_info): Do not free loop_nest.
13987 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13990 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13991 lhs from calls if the lhs has addressable type.
13993 2017-09-27 Richard Biener <rguenther@suse.de>
13995 * graphite.h (scop::max_alias_set): New member.
13996 * graphite-scop-detection.c: Remove references to non-existing
13997 --param in comments.
13998 (build_alias_sets): Record the maximum alias set used for drs.
13999 (build_scops): Support zero as unlimited for
14000 --param graphite-max-arrays-per-scop.
14001 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14002 and inline into ...
14003 (build_poly_sr_1): ... here. Compute alias set based on the
14004 maximum alias set used for drs rather than
14005 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14007 2017-09-27 Richard Biener <rguenther@suse.de>
14009 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14010 --param loop-block-tile-size=0 to disable tiling.
14012 2017-09-27 Richard Biener <rguenther@suse.de>
14014 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14015 (graphite-max-nb-scop-params): Document special value zero.
14016 * domwalk.h (dom_walker::STOP): New symbolical constant.
14017 (dom_walker::dom_walker): Add optional parameter for bb to
14019 (dom_walker::~dom_walker): Declare.
14020 (dom_walker::before_dom_children): Document STOP return value.
14021 (dom_walker::m_user_bb_to_rpo): New member.
14022 (dom_walker::m_bb_to_rpo): Likewise.
14023 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14024 mapping here if not provided by the user.
14025 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14026 provided by the user.
14027 (dom_walker::STOP): Define.
14028 (dom_walker::walk): Do not compute bb to RPO mapping here.
14029 Support STOP return value from before_dom_children to stop
14031 * graphite-optimize-isl.c (optimize_isl): If the schedule
14032 is the same still generate code if -fgraphite-identity
14033 or -floop-parallelize-all are given.
14034 * graphite-scop-detection.c: Include cfganal.h.
14035 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14037 (gather_bbs::before_dom_children): Return STOP for BBs
14039 (build_scops): Compute bb to RPO mapping and pass it to
14040 the domwalk. Treat --param graphite-max-nb-scop-params=0
14041 as not limiting the number of params.
14042 * graphite.c (graphite_initialize): Remove limit on the
14043 number of basic-blocks in a function.
14044 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14045 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14046 default value of 10.
14048 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14050 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14051 Adjust code to eliminate needing to do the shift right 32-bits
14052 operation after XSCVDPSPN.
14054 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14056 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14057 ((X / Y) != 0 -> X >= Y): Likewise.
14059 2017-09-26 Carl Love <cel@us.ibm.com>
14061 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14062 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14063 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14064 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14065 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14066 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14067 definitions and overloading.
14068 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14069 statement for P9V_BUILTIN_XST_LEN_R.
14070 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14071 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14072 define_expand and define_insn for the instructions and builtins.
14073 * doc/extend.texi: Update the built-in documentation file for the new
14074 built-in functions.
14075 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14076 define_insn for the instructions
14078 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14081 * gcc/config/netbsd-protos.h: New file.
14082 * gcc/config/netbsd.c: New file.
14083 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14084 * gcc/config/t-netbsd: New file.
14085 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14086 (tmake_file) Add t-netbsd.
14087 (extra_objs) Add netbsd.o.
14089 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14093 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14095 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14097 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14098 sign extension from a vector register to a GPR by doing a 32-bit
14099 direct move and then an EXTSW.
14100 (extendsi<mode>2 splitter): Likewise.
14101 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14102 right or vector extract after doing XSCVDPSPN. Use
14103 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14105 (movdi_from_sf_zero_ext): Likewise.
14106 (reload_gpr_from_vsxsf): Likewise.
14107 (p8_mfvsrd_4_disf): Delete, no longer used.
14108 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14109 then needing to move the SFmode to a GPR to use the XSCVDPSP
14110 instruction instead of FRSP and XSCVDPSPN.
14111 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14112 it is adjacent to the other XSCVSPDP insns.
14113 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14114 SFmode to be in traditional Altivec registers.
14115 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14116 (vsx_xscvspdpn): Likewise.
14117 (vsx_xscvspdpn_scalar): Likewise.
14119 2017-09-26 Martin Jambor <mjambor@suse.cz>
14121 * tree-sra.c (compare_access_positions): Put integral types first,
14122 stabilize sorting of integral types, remove conditions putting
14123 non-full-precision integers last.
14124 (sort_and_splice_var_accesses): Disable scalarization if a
14125 non-integert would be represented by a non-full-precision integer.
14127 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14129 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14130 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14131 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14132 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14133 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14134 conditionals inside the function instead of around it. Call
14135 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14136 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14138 2017-09-26 Richard Biener <rguenther@suse.de>
14140 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14142 (scop_detection::build_scop_breadth): ... this. Removed.
14143 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14144 (scop_detection::harmful_stmt_in_bb): Likewise.
14145 (scop_detection::graphite_can_represent_stmt): Likewise.
14146 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14147 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14148 (scop_detection::can_represent_loop_1): ... this. Removed.
14149 (scop_detection::harmful_loop_in_region): Simplify after inlining
14150 the above and remove more quadraticness.
14151 (build_scops): Adjust.
14152 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14155 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14158 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14159 REGNO (base) == SP_REG if base is a REG.
14161 PR middle-end/35691
14162 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14163 if it is different SSA_NAME.
14164 (optimize_range_tests_cmp_bitwise): New function.
14165 (optimize_range_tests): Call it.
14167 2017-09-26 Richard Biener <rguenther@suse.de>
14169 PR tree-optimization/82321
14170 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14171 for the def being inside the loop.
14173 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14175 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14177 * config/s390/s390-builtins.def: Fix constraint on op4.
14179 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14181 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14182 independent expanders.
14183 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14184 ("vec_ordered", "vec_unordered"): New expanders.
14186 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14188 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14191 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14193 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14194 vec_unpacks_lo_v16qi.
14195 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14197 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14199 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14200 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14201 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14203 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14205 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14207 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14209 ("*vec_slb<mode>"): New insn pattern.
14210 ("vec_shr_<mode>"): New expander.
14211 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14212 and force the shift count operand to V16QImode.
14213 ("vec_srb<mode>"): Set shift count mode to V16QI.
14215 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14217 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14218 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14219 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14221 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14224 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14226 2017-09-26 Richard Biener <rguenther@suse.de>
14228 PR tree-optimization/82320
14229 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14232 2017-09-25 Jeff Law <law@redhat.com>
14234 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14235 prototype for new argument.
14236 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14237 mostly extracted from rs6000_emit_allocate_stack.
14238 (rs6000_emit_probe_stack_range_stack_clash): New function.
14239 (rs6000_emit_allocate_stack): Call
14240 rs6000_emit_probe_stack_range_stack_clash as needed.
14241 (rs6000_emit_probe_stack_range): Add additional argument
14242 to call to gen_probe_stack_range{si,di}.
14243 (output_probe_stack_range): New.
14244 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14245 (output_probe_stack_range_stack_clash): New.
14246 (rs6000_emit_prologue): Emit notes into dump file as requested.
14247 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14248 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14249 Add additional operand and pass it to output_probe_stack_range.
14251 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14253 PR tree-optimization/82163
14254 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14255 (checking_verify_loop_closed_ssa): New parameter.
14256 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14257 (check_loop_closed_ssa_stmt): Delete.
14258 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14259 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14260 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14263 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14265 * brig-builtins.def: Treat HSAIL barrier builtins as
14266 setjmp/longjump style functions.
14268 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14270 * target.def (constant_alignment): New hook.
14271 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14272 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14273 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14274 * doc/tm.texi: Regenerate.
14275 * targhooks.h (default_constant_alignment): Declare.
14276 (constant_alignment_word_strings): Likewise.
14277 * targhooks.c (default_constant_alignment): New function.
14278 (constant_alignment_word_strings): Likewise.
14279 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14280 instead of CONSTANT_ALIGNMENT.
14281 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14282 (force_const_mem): Likewise.
14283 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14284 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14285 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14286 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14287 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14289 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14290 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14291 constant_alignment_word_strings.
14292 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14293 (CONSTANT_ALIGNMENT): Likewise.
14294 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14295 (arm_constant_alignment): New function.
14296 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14297 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14298 constant_alignment_word_strings.
14299 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14300 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14301 constant_alignment_word_strings.
14302 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14303 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14304 (cris_constant_alignment): New function.
14305 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14306 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14307 (epiphany_constant_alignment): New function.
14308 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14309 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14310 constant_alignment_word_strings.
14311 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14312 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14313 constant_alignment_word_strings.
14314 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14315 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14316 constant_alignment_word_strings.
14317 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14318 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14319 * config/i386/i386.c (ix86_constant_alignment): Make static.
14320 Use the same interface as the target hook.
14321 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14322 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14323 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14324 constant_alignment_word_strings.
14325 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14326 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14327 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14328 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14329 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14330 constant_alignment_word_strings.
14331 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14332 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14333 constant_alignment_word_strings.
14334 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14335 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14336 constant_alignment_word_strings.
14337 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14338 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14340 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14341 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14342 * config/mips/mips.c (mips_constant_alignment): New function.
14343 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14344 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14345 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14346 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14347 (mmix_constant_alignment): Make static. Use the same interface
14348 as the target hook.
14349 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14350 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14351 constant_alignment_word_strings.
14352 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14353 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14354 constant_alignment_word_strings.
14355 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14356 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14357 constant_alignment_word_strings.
14358 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14359 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14360 (rs6000_constant_alignment): New function.
14361 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14362 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14363 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14364 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14365 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14366 (rs6000_constant_alignment): New function.
14367 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14368 * config/s390/s390.c (s390_constant_alignment): New function.
14369 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14370 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14371 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14372 constant_alignment_word_strings.
14373 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14374 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14375 (sparc_constant_alignment): New function.
14376 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14377 * config/spu/spu.c (spu_constant_alignment): New function.
14378 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14379 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14380 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14381 constant_alignment_word_strings.
14382 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14383 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14384 constant_alignment_word_strings.
14385 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14386 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14387 constant_alignment_word_strings.
14388 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14389 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14390 (visium_constant_alignment): New function.
14391 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14392 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14393 (xtensa_constant_alignment): New function.
14394 * system.h (CONSTANT_ALIGNMENT): Poison.
14396 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14398 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14399 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14400 (rs6000_builtin_valid_without_lhs): New helper function.
14401 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14402 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14404 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14406 * target.h (vec_perm_indices): Use unsigned short rather than
14408 (auto_vec_perm_indices): Likewise.
14409 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14410 Use unsigned int rather than unsigned char.
14411 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14413 2017-09-25 Richard Biener <rguenther@suse.de>
14415 * cfgloop.h (sort_sibling_loops): Declare.
14416 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14417 (sort_sibling_loops): New function sorting the sibling loop list
14419 * graphite.c (graphite_transform_loops): Sort sibling loops.
14421 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14423 * target.def (vec_perm_const_ok): Change sel parameter to
14425 * optabs-query.c (can_vec_perm_p): Update accordingly.
14426 * doc/tm.texi: Regenerate.
14427 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14428 auto_vec_perm_indices and remove separate nelt field.
14429 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14430 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14431 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14432 (aarch64_expand_vec_perm_const): Update accordingly.
14433 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14434 to vec_perm_indices.
14435 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14436 auto_vec_perm_indices and remove separate nelt field.
14437 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14438 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14439 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14441 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14442 to vec_perm_indices.
14443 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14444 sel to vec_perm_indices.
14445 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14446 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14447 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14449 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14452 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14455 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14456 on the FUNCTION_DECL function context if it has a DIE that is a
14459 2017-09-25 Richard Biener <rguenther@suse.de>
14461 PR tree-optimization/82285
14462 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14465 2017-09-25 Tom de Vries <tom@codesourcery.com>
14469 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14472 2017-09-25 Richard Biener <rguenther@suse.de>
14474 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14475 ISL errors other than isl_error_quota happen. Dump if the
14476 schedule is the same.
14477 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14478 errors instead of aborting inside ISL.
14480 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14483 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14484 of libgcc_eh for m64.
14485 * config/i386/darwin64.h: Likewise.
14487 2017-09-25 Richard Biener <rguenther@suse.de>
14489 PR middle-end/82144
14490 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14491 attribute for incomplete types nor twice for complete ones.
14493 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14496 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14497 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14500 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14503 * config/i386/i386.opt (mprefer-avx256): Use
14504 ix86_target_flags variable.
14505 * config/i386/i386.c (ix86_target_string): Move
14506 -mprefer-avx256 to flag2_opts.
14508 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14510 PR middle-end/35691
14511 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14512 and x != -1 | y != -1 into (x & y) != -1.
14514 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14516 * config.gcc: Add new case statement to set
14517 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14518 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14519 s390x-*-linux* case statements. Added aarch64 to the list of
14520 supported architectures.
14522 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14524 PR tree-optimization/82289
14525 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14526 STMT_VINFO_RELEVANT_P.
14528 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14529 Alan Hayward <alan.hayward@arm.com>
14530 David Sherwood <david.sherwood@arm.com>
14532 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14533 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14534 (extract_range_from_binary_expr_1): Don't call
14535 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14537 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14538 Alan Hayward <alan.hayward@arm.com>
14539 David Sherwood <david.sherwood@arm.com>
14541 * target.def (preferred_vector_alignment): New hook.
14542 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14544 * doc/tm.texi: Regenerate.
14545 * targhooks.h (default_preferred_vector_alignment): Declare.
14546 * targhooks.c (default_preferred_vector_alignment): New function.
14547 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14549 (DR_TARGET_ALIGNMENT): New macro.
14550 (aligned_access_p): Update commentary.
14551 (vect_known_alignment_in_bytes): New function.
14552 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14554 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14555 Calculate the misalignment based on the target alignment rather than
14557 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14558 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14559 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14560 the target alignment, rather than masking the element misalignment
14561 with the number of elements in a vector. Also use the target
14562 alignment when calculating the maximum number of peels.
14563 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14564 instead of TYPE_ALIGN_UNIT.
14565 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14566 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14567 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14568 (vect_create_data_ref_ptr): Likewise.
14569 (vect_setup_realignment): Realign by ANDing with
14570 -DR_TARGET_MISALIGNMENT.
14571 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14572 the number of peels based on DR_TARGET_ALIGNMENT.
14573 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14574 with the guaranteed alignment boundary when deciding whether
14576 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14577 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14578 (ensure_base_align): Remove stmt_info parameter. Get the
14579 target base alignment from DR_TARGET_ALIGNMENT.
14580 (vectorizable_store): Update call accordingly. Interpret
14581 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14583 (vectorizable_load): Likewise.
14585 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14586 Alan Hayward <alan.hayward@arm.com>
14587 David Sherwood <david.sherwood@arm.com>
14589 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14590 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14591 (vect_enhance_data_refs_alignment): Likewise.
14593 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14595 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14596 error. Only quit immediately if parsing is complete.
14597 (BEGIN): Initialize fatal_err and parse_done.
14598 (begin fpu, end fpu): Check number of arguments.
14599 (begin arch, end arch): Likewise.
14600 (begin cpu, end cpu): Likewise.
14601 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14602 (optalias): Likewise.
14604 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14606 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14607 * config/arm/arm-isa.h: Delete. Move definitions to ...
14608 * arm-cpus.in: ... here. Use new feature and fgroup values.
14609 * config/arm/arm.c (arm_option_override): Use lower case for feature
14611 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14612 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14613 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14615 (print_isa_bits_for): New function.
14616 (gen_isa): New function.
14617 (gen_comm_data): Use print_isa_bits_for.
14618 (define feature): New keyword.
14619 (define fgroup): New keyword.
14620 * config/arm/t-arm (TM_H): Remove.
14621 (GTM_H): Add arm-isa.h.
14622 (arm-isa.h): Add rule to generate file.
14623 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14624 case for feature bit names.
14626 2017-09-22 Richard Biener <rguenther@suse.de>
14628 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14630 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14631 print of no dependency loops ...
14632 * graphite.c (graphite_transform_loops): ... here.
14633 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14635 (same_close_phi_node, remove_duplicate_close_phi,
14636 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14637 (canonicalize_loop_closed_ssa): ... here and simplify.
14638 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14639 (optimize_isl): Use dump_printf_loc to tell when we stopped
14640 optimizing because of an ISL timeout.
14642 2017-09-22 Richard Biener <rguenther@suse.de>
14644 PR tree-optimization/82291
14645 * tree-if-conv.c (predicate_mem_writes): Make sure to
14646 remove writes in blocks predicated with false.
14648 2017-09-22 Richard Biener <rguenther@suse.de>
14650 * sese.c: Include cfganal.h.
14651 (if_region_set_false_region): Remove.
14652 (create_if_region_on_edge): Likewise.
14653 (move_sese_in_condition): Re-implement without destroying
14656 2017-09-22 Richard Biener <rguenther@suse.de>
14658 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14659 Verify both BBs contain loop PHI nodes before dispatching to
14660 copy_loop_phi_args.
14661 (graphite_regenerate_ast_isl): Do not recompute dominators,
14662 do not verify three times. Restructure for clarity.
14663 * graphite-scop-detection.c (same_close_phi_node,
14664 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14665 defined_in_loop_p, canonicalize_loop_closed_ssa,
14666 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14667 checking and SSA rewrite, move to ...
14668 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14669 (graphite_initialize): Do not pass in ctx, do not reset the
14670 SCEV cache, compute only dominators.
14671 (graphite_transform_loops): Allocate ISL ctx after
14672 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14673 Maintain post-dominators only around build_scops.
14674 * sese.c (if_region_set_false_region): Make static. Free
14675 and recompute dominators.
14676 (move_sese_in_condition): Assert we don't get called with
14677 post-dominators computed.
14678 * sese.h (if_region_set_false_region): Remove.
14680 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14682 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14683 mode attribute for TARGET_AVX512VL.
14685 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14687 * config/i386/i386.opt (mprefer-avx256): New option.
14688 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14690 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14691 for TARGET_PREFER_AVX256.
14692 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14694 2017-09-21 Jeff Law <law@redhat.com>
14696 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14697 Fix dump output if the only stack space is for pushed registers.
14699 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14701 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14704 2017-09-21 Martin Sebor <msebor@redhat.com>
14707 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14708 code (in C++) or code that triggers warnings.
14710 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14712 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14714 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14716 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14717 * sched-rgn.c: Ditto.
14718 * sel-sched-ir.c: Ditto.
14720 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14722 * toplev.h (set_random_seed): Adjust return type.
14723 * toplev.c (init_local_tick): Move eager initialization of random_seed
14724 to get_random_seed. Adjust comment.
14725 (init_random_seed): Inline to get_random_seed, delete.
14726 (get_random_seed): Initialize random_seed lazily.
14727 (set_random_seed): Do not return previous value.
14728 (print_switch_value): Do not call get_random_seed.
14730 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14732 * cgraph.c (delete_function_version): New, broken out from...
14733 (cgraph_node::delete_function_version): ...here. Rename to
14734 cgraph_node::delete_function_version_by_decl. Update all uses.
14735 (cgraph_node::remove): Call delete_function_version.
14737 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14740 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14741 VAR_DECLs to which addressable non-volatile parameters are mapped
14742 and for id->retvar after the return value assignment. Clear
14743 id->retval and id->retbnd after inlining.
14745 2017-09-21 Richard Biener <rguenther@suse.de>
14747 PR tree-optimization/82276
14748 PR tree-optimization/82244
14749 * tree-vrp.c (build_assert_expr_for): Set
14750 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14752 (remove_range_assertions): Revert earlier change.
14754 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14757 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14759 2017-09-21 Richard Biener <rguenther@suse.de>
14761 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14762 Restore valid IL after code generation errors.
14763 * graphite.c (graphite_transform_loops): Diagnose code
14764 generation issues as MSG_MISSED_OPTIMIZATION and continue
14765 with processing SCOPs.
14767 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14768 Alan Hayward <alan.hayward@arm.com>
14769 David Sherwood <david.sherwood@arm.com>
14771 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14772 rather than choosing between plus_constant and gen_rtx_<CODE>.
14773 * expr.c (emit_push_insn): Likewise.
14774 (expand_expr_real_2): Likewise.
14776 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14777 Alan Hayward <alan.hayward@arm.com>
14778 David Sherwood <david.sherwood@arm.com>
14780 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14782 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14783 Alan Hayward <alan.hayward@arm.com>
14784 David Sherwood <david.sherwood@arm.com>
14786 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14787 calling tree_to_uhwi.
14789 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14790 Alan Hayward <alan.hayward@arm.com>
14791 David Sherwood <david.sherwood@arm.com>
14793 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14794 INTEGER_CST rather than a negative test for ADDR_EXPR.
14796 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14797 Alan Hayward <alan.hayward@arm.com>
14798 David Sherwood <david.sherwood@arm.com>
14800 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14801 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14803 2017-09-21 Richard Biener <rguenther@suse.de>
14805 PR tree-optimization/71351
14806 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14807 graphite_create_new_loop_guard): Remove, fold remaining parts
14809 (translate_isl_ast_node_for): ... here and simplify.
14811 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14814 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14815 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14816 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14817 alternative always use QI mode, for -Os imov (=R,R) alternative
14818 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14821 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14822 Jeff Law <law@redhat.com>
14824 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14825 (allocate_stack_space): New function, partially extracted from
14826 s390_emit_prologue.
14827 (s390_emit_prologue): Track offset to most recent stack probe.
14828 Code to allocate space moved into allocate_stack_space.
14829 Dump actions when no stack is allocated.
14830 (s390_prologue_plus_offset): New function.
14831 (s390_emit_stack_probe): Likewise.
14833 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14835 * common.opt (Wa, Wl, Wp, g, gz=): Add
14837 (gno-column-info): Remove.
14838 (gcolumn-info): Drop RejectNegative.
14839 (gno-): New prefix.
14840 (gno-record-gcc-switches): Remove.
14841 (grecord-gcc-switches): Drop RejectNegative.
14842 (gno-split-dwarf): Remove.
14843 (gsplit-dwarf): Drop RejectNegative.
14844 (gno-strict-dwarf): Remove.
14845 (gstrict-dwarf): Drop RejectNegative.
14846 * config/darwin.opt (gfull, gused): Add RejectNegative.
14847 * dwarf2out.c (gen_producer_string): Drop
14848 gno-record-gcc-switches handler.
14849 * optc-gen.awk: Add g to prefixes with negative forms.
14850 * opts-common.c (remapping_prefix_p): New.
14851 (find_opt): Check it.
14852 (generate_canonical_option): Test g prefix.
14853 (option_map): Add -gno- mapping.
14854 (add_misspelling_candidates): Check remapping_prefix_p.
14856 2017-09-20 Jeff Law <law@redhat.com>
14858 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14859 thinko in stack clash protection support.
14861 * explow.c (compute_stack_clash_protection_loop_data): Use
14862 CONST_INT_P instead of explicit test. Verify object is a
14863 CONST_INT_P before looking at INTVAL.
14864 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14865 instead of explicit test.
14867 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14870 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14871 address instead of to r1 and r11.
14873 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14875 * config.gcc: Support "knm".
14876 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14877 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14879 * config/i386/i386.c (m_KNM): Define.
14880 (processor_target_table): Add "knm".
14882 (ix86_option_override_internal): Add "knm".
14883 (ix86_issue_rate): Add PROCESSOR_KNM.
14884 (ix86_adjust_cost): Ditto.
14885 (ia32_multipass_dfa_lookahead): Ditto.
14886 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14887 (fold_builtin_cpu): Add M_INTEL_KNM.
14888 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14889 (processor_type): Add PROCESSOR_KNM.
14890 * config/i386/x86-tune.def: Add m_KNM.
14891 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14893 2017-09-20 Richard Biener <rguenther@suse.de>
14895 PR tree-optimization/80213
14896 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14897 are allowed in empty BBs as well.
14898 (canonicalize_loop_closed_ssa): Also look for other complex
14900 (scop_detection::get_sese): Include the loop-closed PHI block
14902 (scop_detection::merge_sese): Remove code adding extra blocks.
14903 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14904 (build_scops): Assert the final returned scop is invalid.
14906 2017-09-20 Richard Biener <rguenther@suse.de>
14908 PR tree-optimization/82264
14909 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14911 (vn_phi_lookup): Likewise.
14912 (vn_phi_insert): Likewise.
14914 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14916 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14917 that fits into uhwi or shwi, add DW_AT_const_value regardless
14918 of early_dwarf without going through RTL, using add_AT_unsigned
14921 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14922 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14923 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14924 (DEBUG_MACRO_SECTION): Likewise.
14925 (DEBUG_LTO_MACRO_SECTION): Likewise.
14926 (DEBUG_STR_DWO_SECTION): Likewise.
14927 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14928 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14929 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14930 (DEBUG_LTO_LINE_STR_SECTION): Define.
14931 (init_sections_and_labels): Initialize debug_line_str_section
14932 variable. Initialize debug_loc_section for -gdwarf-5 to
14933 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14935 2017-09-20 Richard Biener <rguenther@suse.de>
14937 * graphite-sese-to-poly.c (extract_affine): Properly handle
14938 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14940 2017-09-20 Richard Biener <rguenther@suse.de>
14942 PR tree-optimization/81373
14943 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14944 Force SESE live-out defs to be handled even if they are
14947 2017-09-19 Jeff Law <law@redhat.com>
14949 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14950 nothing for stack adjustments with REG_STACK_CHECK.
14951 * sched-deps.c (parse_add_or_inc): Reject insns with
14952 REG_STACK_CHECK from dependency breaking.
14953 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14954 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14955 * reg-notes.def (STACK_CHECK): New note.
14957 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14958 (ix86_expand_prologue): Dump stack clash info as needed.
14959 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14961 * function.c (dump_stack_clash_frame_info): New function.
14962 * function.h (dump_stack_clash_frame_info): Prototype.
14963 (enum stack_clash_probes): New enum.
14965 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14966 flag_stack_clash_protection.
14967 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14968 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14969 (arm_frame_pointer_required): Likewise.
14970 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14971 (ia64_expand_prologue): Likewise.
14972 * config/mips/mips.c (mips_expand_prologue): Likewise.
14973 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14974 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14975 (sparc_flat_expand_prologue): Likewise.
14976 * config/spu/spu.c (spu_expand_prologue): Likewise.
14978 * explow.c: Include "params.h".
14979 (anti_adjust_stack_and_probe_stack_clash): New function.
14980 (get_stack_check_protect): Likewise.
14981 (compute_stack_clash_protection_loop_data): Likewise.
14982 (emit_stack_clash_protection_loop_start): Likewise.
14983 (emit_stack_clash_protection_loop_end): Likewise.
14984 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14985 Use anti_adjust_stack_and_probe_stack_clash.
14986 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14987 (emit_stack_clash_protection_loop_start): Likewise.
14988 (emit_stack_clash_protection_loop_end): Likewise.
14989 * rtl.h (get_stack_check_protect): Prototype.
14990 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14991 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14992 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14994 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14996 * doc/tm.texi: Rebuilt.
14997 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14998 get_stack_check_protect.
14999 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15000 * config/arm/arm.c (arm_expand_prologue): Likewise.
15001 (arm_frame_pointer_required): Likewise.
15002 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15003 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15004 * config/mips/mips.c (mips_expand_prologue): Likewise.
15005 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15006 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15007 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15008 (sparc_flat_expand_prologue): Likewise.
15010 * common.opt (-fstack-clash-protection): New option.
15011 * flag-types.h (enum stack_check_type): Note difference between
15012 -fstack-check= and -fstack-clash-protection.
15013 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15014 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15015 * toplev.c (process_options): Issue warnings/errors for cases
15016 not handled with -fstack-clash-protection.
15017 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15018 (-fstack-check): Note additional problem with -fstack-check=generic.
15019 Note that -fstack-check is primarily for Ada and refer users
15020 to -fstack-clash-protection for stack-clash-protection.
15021 Document new params for stack clash protection.
15023 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15025 * config/i386/i386.c (ix86_split_long_move): Do not handle
15026 address used for LEA in a special way.
15028 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15030 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15032 2017-09-19 Martin Sebor <msebor@redhat.com>
15035 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15036 of incompatible types.
15038 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15040 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15041 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15042 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15043 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15045 2017-09-19 Richard Biener <rguenther@suse.de>
15047 PR tree-optimization/82244
15048 * tree-vrp.c (remove_range_assertions): Do not propagate
15049 a constant to abnormals but replace the assert with a copy.
15051 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15053 PR rtl-optimization/57878
15054 PR rtl-optimization/68988
15055 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15056 avoidance test involving non_reload_pseudos. Move frequency test
15057 below the general fragmentation avoidance test.
15059 2017-09-19 Richard Biener <rguenther@suse.de>
15061 PR tree-optimization/69728
15062 * graphite-sese-to-poly.c (schedule_error): New global.
15063 (add_loop_schedule): Handle empty domain by failing the
15065 (build_original_schedule): Handle schedule_error.
15067 2017-09-19 Richard Biener <rguenther@suse.de>
15069 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15070 Do not iterate to sibling loops but only to siblings of inner
15073 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15076 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15077 that the registers are different.
15079 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15081 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15082 to processor_model and "amdfam17h" to arch_names_table.
15083 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15085 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15088 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15090 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15091 Alan Hayward <alan.hayward@arm.com>
15092 David Sherwood <david.sherwood@arm.com>
15094 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15096 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15098 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15099 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15100 vect_schedule_slp_instance.
15101 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15102 Update call to vect_slp_analyze_node_operations. Simplify return
15104 (vect_slp_analyze_bb_1): Update call accordingly.
15105 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15106 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15107 (vect_schedule_slp): Update call accordingly.
15109 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15110 Alan Hayward <alan.hayward@arm.com>
15111 David Sherwood <david.sherwood@arm.com>
15113 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15114 with types that aren't in fact scalar.
15116 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15118 * tree-vect-slp.c (vect_record_max_nunits): New function,
15120 (vect_build_slp_tree_1): ...here.
15121 (vect_build_slp_tree_2): Call it for phis too.
15123 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15125 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15126 to vect_get_vec_def_for_operand when getting the mask operand.
15128 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15129 Alan Hayward <alan.hayward@arm.com>
15130 David Sherwood <david.sherwood@arm.com>
15132 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15135 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15136 Alan Hayward <alan.hayward@arm.com>
15137 David Sherwood <david.sherwood@arm.com>
15139 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15140 calculation for vector booleans.
15142 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15143 Alan Hayward <alan.hayward@arm.com>
15144 David Sherwood <david.sherwood@arm.com>
15146 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15148 (vect_transform_stmt): ...here.
15149 (vect_analyze_stmt): Use it instead of calling
15150 vectorizable_live_operation directly.
15152 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15154 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15155 non-SIMT targets in acc vector loops.
15157 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15159 * configure.ac: Add arc and check if assembler supports gdwarf2.
15160 * configure: Regenerate.
15162 2017-09-18 Richard Biener <rguenther@suse.de>
15164 PR tree-optimization/82220
15165 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15166 epilogue niters from the min_profitable_iters compute.
15168 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15171 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15172 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15173 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15175 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15178 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15179 switching to a new text section.
15181 2017-09-18 Richard Biener <rguenther@suse.de>
15183 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15185 (build_alias_set): Reject aliases with no access function.
15187 2017-09-18 Richard Biener <rguenther@suse.de>
15189 PR tree-optimization/79622
15190 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15192 (build_cross_bb_scalars_use): Likewise.
15194 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15196 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15198 2017-09-18 Alan Modra <amodra@gmail.com>
15201 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15202 stack_pointer_rtx for count 0. Update comments. Break up
15203 large rtl expression.
15205 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15207 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15208 Increase to 20 bytes.
15209 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15210 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15211 or avx version of the stub.
15213 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15216 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15217 compute the minimum stack alignment. Also update preferred stack
15218 boundary for leaf functions.
15220 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15222 PR tree-optimization/82228
15223 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15226 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15228 * common/config/nds32/nds32-common.c
15229 (nds32_option_optimization_table): Refine formatting.
15230 (nds32_option_optimization_table): Use -fsched-pressure and
15231 -fomit-frame-pointer for specific optimization level.
15233 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15235 * config/nds32/nds32.c: Refine formatting and comments.
15236 * config/nds32/nds32.h: Likewise.
15237 * config/nds32/nds32.md: Likewise.
15238 * config/nds32/nds32-cost.c: Likewise.
15239 * config/nds32/nds32-isr.c: Likewise.
15240 * config/nds32/nds32-md-auxiliary.c: Likewise.
15241 * config/nds32/nds32-multiple.md: Likewise.
15242 * config/nds32/nds32-predicates.c: Likewise.
15244 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15245 Jakub Jelinek <jakub@redhat.com>
15247 Add support for -std=c++2a.
15248 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15250 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15252 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15255 * doc/extend.texi (Common Function Attributes): Add
15256 references to ARM, AArch64, and S/390 specific attributes.
15257 (Function Specific Option Pragmas): Add AArch64 and S/390
15258 to list of back ends that support the target pragma.
15260 2017-09-15 Nathan Sidwell <nathan@acm.org>
15262 * doc/standards.texi: Fix C++17 description. Update URLs for
15265 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15267 * common.opt (Wcast-align=strict): New warning option.
15268 * doc/invoke.texi: Document -Wcast-align=strict.
15270 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15272 * cgraph.h (cgraph_thunk_info): Add comments.
15273 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15274 assert for VIRTUAL_* arguments stricter.
15276 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15278 PR tree-optimization/71026
15279 * match.pd: Move RDIV patterns from fold-const.c
15280 * fold-const.c (distribute_real_division): Removed.
15281 (fold_binary_loc): Remove calls to distribute_real_divison.
15283 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15285 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15286 c++1z and gnu++1z as deprecated. Change other references to
15287 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15288 Change -Wc++1z-compat to -Wc++17-compat.
15289 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15290 * dwarf2out.c (highest_c_language): Handle C++17.
15291 (gen_compile_unit_die): Likewise.
15293 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15295 PR rtl-optimization/82192
15296 * combine.c (make_extraction): Don't look through non-paradoxical
15297 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15300 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15301 Alan Hayward <alan.hayward@arm.com>
15302 David Sherwood <david.sherwood@arm.com>
15304 * target.def (function_arg_offset): New hook.
15305 * targhooks.h (default_function_arg_offset): Declare.
15306 * targhooks.c (default_function_arg_offset): New function.
15307 * function.c (locate_and_pad_parm): Use
15308 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15309 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15310 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15311 * doc/tm.texi: Regenerate.
15312 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15313 * config/spu/spu.c (spu_function_arg_offset): New function.
15314 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15315 * system.h (FUNCTION_ARG_OFFSET): Poison.
15317 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15318 Alan Hayard <alan.hayward@arm.com>
15319 David Sherwood <david.sherwood@arm.com>
15321 * target.def (truly_noop_truncation): New hook.
15322 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15323 than TRULY_NOOP_TRUNCATION.
15324 * hooks.h (hook_bool_uint_uint_true): Declare.
15325 * hooks.c (hook_bool_uint_uint_true): New function.
15326 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15327 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15328 * doc/tm.texi: Regenerate.
15329 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15330 rather than TRULY_NOOP_TRUNCATION in comments.
15331 (simplify_comparison): Likewise.
15332 (record_truncated_value): Likewise.
15333 * expmed.c (extract_bit_field_1): Likewise.
15334 (extract_split_bit_field): Likewise.
15335 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15336 instead of TRULY_NOOP_TRUNCATION.
15337 * function.c (assign_parm_setup_block): Likewise.
15338 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15339 * rtlhooks.c: Include target.h.
15340 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15341 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15342 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15343 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15344 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15345 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15346 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15347 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15348 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15349 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15350 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15351 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15352 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15353 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15354 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15355 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15356 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15357 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15358 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15359 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15360 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15361 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15362 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15363 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15364 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15365 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15366 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15367 rather than TRULY_NOOP_TRUNCATION in comments.
15368 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15369 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15370 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15371 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15372 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15373 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15374 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15375 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15376 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15377 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15378 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15379 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15380 rather than TRULY_NOOP_TRUNCATION in comments.
15381 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15382 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15383 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15384 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15385 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15386 TRULY_NOOP_TRUNCATION condition.
15387 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15388 (TRULY_NOOP_TRUNCATION): Delete.
15389 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15390 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15391 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15392 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15393 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15394 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15395 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15396 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15397 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15398 rather than TRULY_NOOP_TRUNCATION in comments.
15399 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15400 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15401 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15402 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15403 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15404 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15406 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15409 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15410 (*cmp_ior): Likewise.
15411 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15412 (*ior_scc_scc_cmp): Likewise.
15413 (*and_scc_scc): Likewise.
15414 (*and_scc_scc_cmp): Likewise.
15416 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15417 Alan Hayard <alan.hayward@arm.com>
15418 David Sherwood <david.sherwood@arm.com>
15420 * target.def (can_change_mode_class): New hook.
15421 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15422 (hard_regno_nregs): Likewise.
15423 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15424 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15425 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15426 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15427 (LOAD_EXTEND_OP): Update accordingly.
15428 * doc/tm.texi: Regenerate.
15429 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15430 CANNOT_CHANGE_MODE_CLASS.
15431 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15432 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15433 * combine.c (simplify_set): Update accordingly.
15434 * emit-rtl.c (validate_subreg): Likewise.
15435 * recog.c (general_operand): Likewise.
15436 * regcprop.c (mode_change_ok): Likewise.
15437 * reload1.c (choose_reload_regs): Likewise.
15438 (inherit_piecemeal_p): Likewise.
15439 * rtlanal.c (simplify_subreg_regno): Likewise.
15440 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15441 instead of CANNOT_CHANGE_MODE_CLASS.
15442 (reload_cse_simplify_operands): Likewise.
15443 * reload.c (push_reload): Use targetm.can_change_mode_class
15444 instead of CANNOT_CHANGE_MODE_CLASS.
15445 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15446 REG_CANNOT_CHANGE_MODE_P.
15447 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15448 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15449 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15450 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15451 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15452 (arm_can_change_mode_class): New function.
15453 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15454 than CANNOT_CHANGE_MODE_CLASS in comments.
15455 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15456 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15457 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15458 (ix86_can_change_mode_class): ...this new function, inverting the
15459 sense of the return value.
15460 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15461 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15462 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15463 (ia64_can_change_mode_class): New function.
15464 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15465 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15466 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15467 (m32c_can_change_mode_class): ...this new function, inverting the
15468 sense of the return value.
15469 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15470 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15471 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15472 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15473 (mips_can_change_mode_class): ...this new function, inverting the
15474 sense of the return value.
15475 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15476 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15477 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15478 (msp430_can_change_mode_class): New function.
15479 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15480 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15481 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15482 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15483 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15484 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15485 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15486 (pa_cannot_change_mode_class): Replace with...
15487 (pa_can_change_mode_class): ...this new function, inverting the
15488 sense of the return value.
15489 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15490 than CANNOT_CHANGE_MODE_CLASS in comments.
15491 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15492 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15493 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15494 (pdp11_cannot_change_mode_class): Replace with...
15495 (pdp11_can_change_mode_class): ...this new function, inverting the
15496 sense of the return value.
15497 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15498 * config/powerpcspe/powerpcspe-protos.h
15499 (rs6000_cannot_change_mode_class_ptr): Delete.
15500 * config/powerpcspe/powerpcspe.c
15501 (rs6000_cannot_change_mode_class_ptr): Delete.
15502 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15503 (rs6000_option_override_internal): Assign to
15504 targetm.can_change_mode_class instead of
15505 rs6000_cannot_change_mode_class_ptr.
15506 (rs6000_cannot_change_mode_class): Replace with...
15507 (rs6000_can_change_mode_class): ...this new function, inverting the
15508 sense of the return value.
15509 (rs6000_debug_cannot_change_mode_class): Replace with...
15510 (rs6000_debug_can_change_mode_class): ...this new function.
15511 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15512 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15513 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15514 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15515 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15517 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15518 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15519 (rs6000_option_override_internal): Assign to
15520 targetm.can_change_mode_class instead of
15521 rs6000_cannot_change_mode_class_ptr.
15522 (rs6000_cannot_change_mode_class): Replace with...
15523 (rs6000_can_change_mode_class): ...this new function, inverting the
15524 sense of the return value.
15525 (rs6000_debug_cannot_change_mode_class): Replace with...
15526 (rs6000_debug_can_change_mode_class): ...this new function.
15527 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15528 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15529 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15530 (s390_can_change_mode_class): ...this new function, inverting the
15531 sense of the return value.
15532 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15533 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15534 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15535 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15536 (sh_cannot_change_mode_class): Replace with...
15537 (sh_can_change_mode_class): ...this new function, inverting the
15538 sense of the return value.
15539 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15540 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15541 (sparc_can_change_mode_class): New function.
15542 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15543 * config/spu/spu.c (spu_can_change_mode_class): New function.
15544 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15545 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15546 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15547 (visium_can_change_mode_class): New function.
15548 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15550 2017-09-15 Richard Biener <rguenther@suse.de>
15552 PR tree-optimization/82217
15553 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15554 but not undefined case.
15556 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15559 * postreload.c (reload_cse_simplify_operands): Skip
15560 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15562 2017-09-15 Richard Biener <rguenther@suse.de>
15564 PR tree-optimization/68823
15565 * graphite-scop-detection.c (build_alias_set): If we have a
15566 possible dependence check whether we can handle them by just
15567 looking at the DRs DR_ACCESS_FNs.
15568 (build_scops): If build_alias_set fails, fail the SCOP.
15570 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15572 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15573 to support float128 built-in functions that require the ISA 3.0
15575 (BU_FLOAT128_3_HW): Likewise.
15576 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15577 built-in functions.
15578 (FMAF128): Likewise.
15580 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15581 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15582 floating point instructions.
15583 (rs6000_invalid_builtin): Likewise.
15584 (rs6000_builtin_mask_names): Likewise.
15585 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15586 (RS6000_BTM_FLOAT128_HW): Likewise.
15587 (RS6000_BTM_COMMON): Likewise.
15588 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15590 * doc/extend.texi (RS/6000 built-in functions): Document the
15591 IEEE 128-bit floating point square root and fused multiply-add
15592 built-in functions.
15594 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15596 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15597 reg (r2) isn't in the set of registers defined in the prologue.
15599 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15600 Alan Hayward <alan.hayward@arm.com>
15601 David Sherwood <david.sherwood@arm.com>
15603 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15604 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15605 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15606 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15607 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15609 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15610 max_vectorization_factor.
15611 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15613 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15614 Alan Hayward <alan.hayward@arm.com>
15615 David Sherwood <david.sherwood@arm.com>
15617 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15618 (vect_worthwhile_without_simd_p): Declare.
15619 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15620 (vectorizable_reduction): Use it.
15621 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15622 (vectorizable_operation): Likewise.
15624 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15625 Alan Hayward <alan.hayward@arm.com>
15626 David Sherwood <david.sherwood@arm.com>
15628 * tree-vectorizer.h (vect_get_num_copies): New function.
15629 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15630 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15631 (vectorizable_induction): Likewise.
15632 (vectorizable_live_operation): Likewise.
15633 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15634 (vectorizable_bswap): Likewise.
15635 (vectorizable_call): Likewise.
15636 (vectorizable_conversion): Likewise.
15637 (vectorizable_assignment): Likewise.
15638 (vectorizable_shift): Likewise.
15639 (vectorizable_operation): Likewise.
15640 (vectorizable_store): Likewise.
15641 (vectorizable_load): Likewise.
15642 (vectorizable_condition): Likewise.
15643 (vectorizable_comparison): Likewise.
15644 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15646 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15647 Alan Hayward <alan.hayward@arm.com>
15648 David Sherwood <david.sherwood@arm.com>
15650 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15651 of vect_init_vector.
15653 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15654 Alan Hayward <alan.hayward@arm.com>
15655 David Sherwood <david.sherwood@arm.com>
15657 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15658 an inline wrapper that provides a location.
15659 (gimple_build_vector): Likewise.
15660 * gimple-fold.c (gimple_build_vector_from_val): New function.
15661 (gimple_build_vector): Likewise.
15662 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15663 functions to build the initial value. Always return a gimple value.
15664 (get_initial_defs_for_reduction): Likewise. Only compute
15666 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15667 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15668 (vectorizable_induction): Use gimple_build_vector rather than
15671 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15672 Alan Hayward <alan.hayward@arm.com>
15673 David Sherwood <david.sherwood@arm.com>
15675 * target.h (vec_perm_indices): New typedef.
15676 (auto_vec_perm_indices): Likewise.
15677 * optabs-query.h: Include target.h
15678 (can_vec_perm_p): Take a vec_perm_indices *.
15679 * optabs-query.c (can_vec_perm_p): Likewise.
15680 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15681 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15682 * tree-vect-generic.c (lower_vec_perm): Likewise.
15683 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15684 (vect_grouped_load_supported): Likewise.
15685 (vect_shift_permute_load_chain): Likewise.
15686 (vect_permute_store_chain): Use auto_vec_perm_indices.
15687 (vect_permute_load_chain): Likewise.
15688 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15689 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15690 Update uses of can_vec_perm_p.
15691 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15692 mode with a number of elements. Take a vec_perm_indices *.
15693 (vect_create_epilog_for_reduction): Update accordingly.
15694 Use auto_vec_perm_indices.
15695 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15696 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15697 (vect_transform_slp_perm_load): Likewise.
15698 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15699 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15700 (vect_gen_perm_mask_checked): Likewise.
15701 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15702 (vect_gen_perm_mask_checked): Likewise.
15703 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15704 (vectorizable_store): Likewise.
15705 (vectorizable_load): Likewise.
15706 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15707 (vectorizable_bswap): Likewise.
15709 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15710 Alan Hayward <alan.hayward@arm.com>
15711 David Sherwood <david.sherwood@arm.com>
15713 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15714 * tree.c (build_vector): Likewise.
15715 (build_vector_from_ctor): Update accordingly.
15716 (build_vector_from_val): Likewise.
15717 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15718 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15719 * tree-vect-generic.c (add_rshift): Likewise.
15720 (expand_vector_divmod): Likewise.
15721 (optimize_vector_constructor): Likewise.
15722 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15723 (vect_transform_slp_perm_load): Likewise.
15724 (vect_schedule_slp_instance): Likewise.
15725 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15726 (vectorizable_call): Likewise.
15727 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15728 * expmed.c (make_tree): Likewise.
15729 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15730 a vector passed to build_vector.
15731 (fold_convert_const): Likewise.
15732 (exact_inverse): Likewise.
15733 (fold_ternary_loc): Likewise.
15734 (fold_relational_const): Likewise.
15735 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15736 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15737 (const_unop): Likewise. Store the reduction accumulator in a
15738 variable rather than an array.
15739 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15740 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15741 the new vector, rather than constructing it after the input arrays.
15742 (native_interpret_vector): Use auto_vec<tree> when building
15743 a vector passed to build_vector. Add elements in order.
15744 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15745 auto_vec<tree> when building a vector passed to build_vector.
15746 (vect_create_epilog_for_reduction): Likewise.
15747 (vectorizable_induction): Likewise.
15748 (get_initial_def_for_reduction): Likewise. Fix indentation of
15750 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15752 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15753 passed to build_vector.
15755 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15756 Alan Hayward <alan.hayward@arm.com>
15757 David Sherwood <david.sherwood@arm.com>
15759 * tree-core.h (tree_base::u): Add an "nelts" field.
15760 (tree_vector): Use VECTOR_CST_NELTS as the length.
15761 * tree.c (tree_size): Likewise.
15762 (make_vector): Initialize VECTOR_CST_NELTS.
15763 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15764 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15765 TYPE_VECTOR_SUBPARTS.
15766 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15767 as the number of units, setting it from VECTOR_CST_NELTS.
15768 (const_vector_from_tree): Likewise.
15769 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15770 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15771 (fold_negate_expr_1): Likewise.
15772 (fold_convert_const): Likewise.
15773 (const_binop): Likewise. Differentiate the number of output and
15775 (const_unop): Likewise.
15776 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15777 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15778 in cases that did the opposite.
15780 2017-09-14 Richard Biener <rguenther@suse.de>
15782 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15785 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15787 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15789 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15792 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15793 if and where to split a bb, except for splitting before debug insn
15794 sequences followed by non-label real insn. Delete debug insns
15795 in between basic blocks.
15797 * combine.c (make_compound_operation_int): Formatting fixes.
15799 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15800 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15801 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15802 * config/sol2.h (LINK_EH_SPEC): Likewise.
15803 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15804 * config/s390/linux.h (LINK_SPEC): Likewise.
15805 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15806 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15807 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15808 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15809 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15810 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15811 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15812 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15814 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15816 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15818 (ENDFILE_LINUX_SPEC): Likewise.
15819 (LINK_EH_SPEC): Likewise.
15820 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15821 (LINK_OS_LINUX_SPEC32): Likewise.
15822 (LINK_OS_LINUX_SPEC64): Likewise.
15823 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15824 (LINK_OS_LINUX_SPEC): Likewise.
15826 2017-09-13 Martin Liska <mliska@suse.cz>
15828 PR middle-end/82154
15829 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15830 CASE_HIGH is NULL_TREE.
15832 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15833 Alan Hayward <alan.hayward@arm.com>
15834 David Sherwood <david.sherwood@arm.com>
15836 * target.def (secondary_memory_needed): New hook.
15837 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15838 instead of SECONDARY_MEMORY_NEEDED.
15839 (secondary_memory_needed_mode): Likewise.
15840 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15841 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15842 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15843 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15844 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15845 * doc/tm.texi: Regenerate.
15846 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15847 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15848 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15849 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15850 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15851 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15852 mode argument first and change the reg_class arguments to reg_class_t.
15853 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15854 Make static. Update the call to inline_secondary_memory_needed.
15855 (ix86_register_move_cost): Update the call to
15856 inline_secondary_memory_needed.
15857 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15858 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15860 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15861 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15863 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15864 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15865 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15866 and match hook interface. Add comment from mips.h.
15867 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15868 * config/mmix/mmix.md (truncdfsf2): Refer to
15869 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15871 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15872 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15873 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15874 (pa_secondary_memory_needed): New function.
15875 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15876 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15877 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15878 (pdp11_secondary_memory_needed): Make static and match hook interface.
15879 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15880 * config/powerpcspe/powerpcspe-protos.h
15881 (rs6000_secondary_memory_needed_ptr): Delete.
15882 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15884 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15885 (rs6000_option_override_internal): Assign to
15886 targetm.secondary_memory_needed rather than
15887 rs6000_secondary_memory_needed_ptr.
15888 (rs6000_secondary_memory_needed): Match hook interface.
15889 (rs6000_debug_secondary_memory_needed): Likewise.
15890 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15891 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15892 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15893 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15894 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15895 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15897 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15898 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15899 (rs6000_option_override_internal): Assign to
15900 targetm.secondary_memory_needed rather than
15901 rs6000_secondary_memory_needed_ptr.
15902 (rs6000_secondary_memory_needed): Match hook interface.
15903 (rs6000_debug_secondary_memory_needed): Likewise.
15904 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15905 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15906 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15907 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15908 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15909 (sparc_secondary_memory_needed): New function.
15910 * lra-constraints.c (check_and_process_move): Refer to
15911 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15913 (curr_insn_transform): Likewise.
15914 (process_alt_operands): Use targetm.secondary_memory_needed
15915 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15916 (check_secondary_memory_needed_p): Likewise.
15917 (choose_split_class): Likewise.
15918 * reload.c: Unconditionally include code that was previously
15919 conditional on SECONDARY_MEMORY_NEEDED.
15920 (push_secondary_reload): Use targetm.secondary_memory_needed
15921 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15922 (push_reload): Likewise.
15923 * reload1.c: Unconditionally include code that was previously
15924 conditional on SECONDARY_MEMORY_NEEDED.
15925 (choose_reload_regs): Use targetm.secondary_memory_needed
15926 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15927 (gen_reload): Likewise.
15928 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15930 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15931 Alan Hayward <alan.hayward@arm.com>
15932 David Sherwood <david.sherwood@arm.com>
15934 * target.def (secondary_memory_needed_mode): New hook:
15935 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15936 * targhooks.h (default_secondary_memory_needed_mode): New function.
15937 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15938 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15939 * doc/tm.texi: Regenerate.
15940 * lra-constraints.c (check_and_process_move): Use
15941 targetm.secondary_memory_needed_mode instead of
15942 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15943 (curr_insn_transform): Likewise.
15944 * reload.c (get_secondary_mem): Likewise.
15945 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15946 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15948 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15949 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15950 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15951 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15952 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15954 * config/powerpcspe/powerpcspe-protos.h
15955 (rs6000_secondary_memory_needed_mode): Delete.
15956 * config/powerpcspe/powerpcspe.c
15957 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15958 (rs6000_secondary_memory_needed_mode): Make static.
15959 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15960 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15962 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15964 (rs6000_secondary_memory_needed_mode): Make static.
15965 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15966 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15967 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15968 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15969 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15971 (sparc_secondary_memory_needed_mode): New function.
15972 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15974 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15976 * config/aarch64/constraints.md (Umq): New constraint.
15977 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15979 (mov<mode>): Update condition.
15981 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15983 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15984 when bitposition is the same.
15986 2017-09-13 Richard Biener <rguenther@suse.de>
15988 * dwarf2out.c (output_die_symbol): Remove.
15989 (output_die): Do not output a DIEs symbol.
15991 2017-09-13 Richard Biener <rguenther@suse.de>
15993 PR middle-end/82128
15994 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15995 default-def to avoid breaking iterator update with the weird
15996 interaction with cgraph_update_edges_for_call_stmt_node.
15998 2017-09-13 Richard Biener <rguenther@suse.de>
16000 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16001 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16002 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16003 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16005 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16007 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16008 Disable pc relative literal load irrespective of
16009 TARGET_FIX_ERR_A53_84341 for default.
16011 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16013 * config/sparc/sparc.c (output_return): Output the source location of
16014 the insn in the delay slot, if any.
16015 (output_sibcall): Likewise.
16017 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16020 * common.opt (-static-pie): New alias.
16021 (shared): Negate static-pie.
16022 (-no-pie): Update help text.
16024 (static-pie): New option.
16025 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16026 -static-pie support.
16027 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16028 (LINK_EH_SPEC): Likewise.
16029 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16030 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16031 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16032 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16033 (init_gcc_specs): Likewise.
16034 (init_spec): Likewise.
16035 (display_help): Update help message for -pie.
16036 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16039 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16041 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16042 (movdi_aarch64): Likewise.
16043 (movti_aarch64): Likewise.
16045 2017-09-12 Simon Wright <simon@pushface.org>
16048 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16049 calculation of the minor version, always output as 0.
16051 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16054 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16055 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16056 on it early, rather than manual conversion late. For
16057 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16058 instead of performing manual conversion.
16060 2017-09-12 Carl Love <cel@us.ibm.com>
16062 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16063 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16065 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16066 VMULOSW): Add definitions.
16067 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16068 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16069 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16070 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16071 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16073 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16075 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16077 (movti_aarch64): Likewise.
16078 (movdf_aarch64): Likewise.
16079 (movtf_aarch64): Likewise.
16080 (load_pairdi): Likewise.
16081 (store_pairdi): Likewise.
16082 (load_pairdf): Likewise.
16083 (store_pairdf): Likewise.
16084 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16085 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16086 (ldr_got_small_<mode>): Likewise.
16087 (ldr_got_small_28k_<mode>): Likewise.
16088 (ldr_got_tiny): Likewise.
16089 * config/aarch64/iterators.md (ldst_sz): New.
16090 (ldpstp_sz): Likewise.
16091 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16093 (thunderx_load): Split load_8 to load_16.
16094 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16096 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16097 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16098 (xgene1_store_pair): Split store_8 to store_16.
16099 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16100 (falkor_st_0_st_sd): Split store_8 to store_16.
16102 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16104 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16105 and store1/2/3/4 to store_4/8/12/16.
16106 * config/aarch64/aarch64.md: Update for rename.
16107 * config/arm/arm.md: Likewise.: Likewise.
16108 * config/arm/arm.c: Likewise.
16109 * config/arm/thumb1.md: Likewise.
16110 * config/arm/thumb2.md: Likewise.
16111 * config/arm/vfp.md: Likewise.
16112 * config/arm/arm-generic.md: Likewise.
16113 * config/arm/arm1020e.md: Likewise.
16114 * config/arm/arm1026ejs.md: Likewise.
16115 * config/arm/arm1136jfs.md: Likewise.
16116 * config/arm/arm926ejs.md: Likewise.
16117 * config/arm/cortex-a15.md: Likewise.
16118 * config/arm/cortex-a17.md: Likewise.
16119 * config/arm/cortex-a5.md: Likewise.
16120 * config/arm/cortex-a53.md: Likewise.
16121 * config/arm/cortex-a57.md: Likewise.
16122 * config/arm/cortex-a7.md: Likewise.
16123 * config/arm/cortex-a8.md: Likewise.
16124 * config/arm/cortex-a9.md: Likewise.
16125 * config/arm/cortex-m4.md: Likewise.
16126 * config/arm/cortex-m7.md: Likewise.
16127 * config/arm/cortex-r4.md: Likewise.
16128 * config/arm/exynos-m1.md: Likewise.
16129 * config/arm/fa526.md: Likewise.
16130 * config/arm/fa606te.md: Likewise.
16131 * config/arm/fa626te.md: Likewise.
16132 * config/arm/fa726te.md: Likewise.
16133 * config/arm/fmp626.md: Likewise.
16134 * config/arm/iwmmxt.md: Likewise.
16135 * config/arm/ldmstm.md: Likewise.
16136 * config/arm/marvell-pj4.md: Likewise.
16137 * config/arm/xgene1.md: Likewise.
16138 * config/aarch64/thunderx.md: Likewise.
16139 * config/aarch64/thunderx2t99.md: Likewise.
16140 * config/aarch64/falkor.md: Likewise.
16142 2017-09-12 Martin Liska <mliska@suse.cz>
16144 * attribs.c (private_lookup_attribute): New function.
16145 * attribs.h (private_lookup_attribute): Declared here.
16146 (lookup_attribute): Called from this place.
16148 2017-09-12 Richard Biener <rguenther@suse.de>
16150 PR tree-optimization/82157
16151 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16152 stmts with side-effects.
16154 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16155 Alan Hayward <alan.hayward@arm.com>
16156 David Sherwood <david.sherwood@arm.com>
16158 * target.def (hard_regno_nregs): New hook.
16159 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16160 * targhooks.h (default_hard_regno_nregs): Declare.
16161 * targhooks.c (default_hard_regno_nregs): New function.
16162 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16163 (TARGET_HARD_REGNO_NREGS): ...this hook.
16164 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16165 (CLASS_MAX_NREGS): Likewise.
16166 * doc/tm.texi: Regenerate.
16167 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16168 instead of HARD_REGNO_NREGS.
16169 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16170 HARD_REGNO_NREGS in the comment.
16171 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16172 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16173 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16174 Return an unsigned int.
16175 (TARGET_HARD_REGNO_NREGS): Redefine.
16176 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16177 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16178 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16179 (arc_hard_regno_nregs): New function.
16180 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16181 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16182 (arm_hard_regno_nregs): New function.
16183 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16184 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16185 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16186 (TARGET_HARD_REGNO_NREGS): Redefine.
16187 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16188 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16189 (HARD_REGNO_NREGS): Delete.
16190 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16191 (cr16_hard_regno_nregs): New function.
16192 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16193 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16194 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16195 (cris_hard_regno_nregs): New function.
16196 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16197 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16198 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16199 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16200 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16201 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16202 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16203 (frv_hard_regno_nregs): Make static. Take and return an
16205 (frv_class_max_nregs): Remove outdated copy of documentation.
16206 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16207 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16208 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16209 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16210 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16211 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16212 (TARGET_HARD_REGNO_NREGS): Redefine.
16213 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16214 (CLASS_MAX_NREGS): Update comment.
16215 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16216 (ia64_hard_regno_nregs): New function.
16217 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16218 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16219 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16220 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16221 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16223 (m32c_hard_regno_nregs): Likewise. Make static.
16224 (TARGET_HARD_REGNO_NREGS): Redefine.
16225 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16226 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16227 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16228 (m68k_hard_regno_nregs): New function.
16229 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16230 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16231 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16232 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16233 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16234 Take and return an unsigned int.
16235 (TARGET_HARD_REGNO_NREGS): Redefine.
16236 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16237 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16238 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16239 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16240 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16241 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16242 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16243 (msp430_hard_regno_nregs): Make static. Take and return an
16245 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16246 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16247 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16248 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16249 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16250 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16251 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16252 (TARGET_HARD_REGNO_NREGS): Redefine.
16253 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16254 (PA_HARD_REGNO_NREGS): ...this.
16255 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16256 (PA_HARD_REGNO_NREGS): ...this.
16257 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16258 (pa_hard_regno_nregs): New function.
16259 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16260 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16261 (pdp11_hard_regno_nregs): New function.
16262 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16263 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16264 (rs6000_hard_regno_nregs_hook): New function.
16265 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16266 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16267 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16268 Take and return an unsigned int. Move earlier in file.
16269 (TARGET_HARD_REGNO_NREGS): Redefine.
16270 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16271 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16272 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16273 (rl78_hard_regno_nregs): Make static. Take and return an
16275 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16276 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16277 (rs6000_hard_regno_nregs_hook): New function.
16278 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16279 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16280 (TARGET_HARD_REGNO_NREGS): Redefine.
16281 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16282 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16283 instead of HARD_REGNO_NREGS.
16284 (s390_hard_regno_nregs): New function.
16285 (s390_hard_regno_mode_ok): Add comment from s390.h.
16286 (TARGET_HARD_REGNO_NREGS): Redefine.
16287 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16288 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16289 (sh_hard_regno_nregs): New function.
16290 (sh_pass_in_reg_p): Use it.
16291 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16292 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16293 (sparc_hard_regno_nregs): New function.
16294 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16295 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16296 (spu_function_arg_advance): Use it, supplying a valid register number.
16297 (TARGET_HARD_REGNO_NREGS): Redefine.
16298 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16299 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16300 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16301 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16302 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16303 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16304 (CLASS_MAX_NREGS): Remove copy of old documentation.
16305 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16306 (visium_hard_regno_nregs): New function.
16307 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16308 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16309 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16310 xtensa_hard_regno_nregs): New function.
16311 * system.h (HARD_REGNO_NREGS): Poison.
16313 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16315 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16316 hard_regno_nregs instead of HARD_REGNO_NREGS.
16317 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16318 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16319 (c6x_expand_epilogue): Likewise.
16320 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16321 (frv_read_iacc_argument): Likewise.
16322 * config/sh/sh.c: Include regs.h.
16323 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16324 (regs_used): Likewise.
16325 (output_stack_adjust): Likewise.
16326 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16327 * expmed.c: Include regs.h.
16328 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16329 * ree.c: Include regs.h.
16330 (combine_reaching_defs): Use hard_regno_nregs instead of
16332 (add_removable_extension): Likewise.
16334 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16336 * regs.h (hard_regno_nregs): Turn into a function.
16337 (end_hard_regno): Update accordingly.
16338 * caller-save.c (setup_save_areas): Likewise.
16339 (save_call_clobbered_regs): Likewise.
16340 (replace_reg_with_saved_mem): Likewise.
16341 (insert_restore): Likewise.
16342 (insert_save): Likewise.
16343 * combine.c (can_change_dest_mode): Likewise.
16344 (move_deaths): Likewise.
16345 (distribute_notes): Likewise.
16346 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16347 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16348 (rs6000_split_multireg_move): Likewise.
16349 (rs6000_register_move_cost): Likewise.
16350 (rs6000_memory_move_cost): Likewise.
16351 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16352 (rs6000_split_multireg_move): Likewise.
16353 (rs6000_register_move_cost): Likewise.
16354 (rs6000_memory_move_cost): Likewise.
16355 * cselib.c (cselib_reset_table): Likewise.
16356 (cselib_lookup_1): Likewise.
16357 * emit-rtl.c (set_mode_and_regno): Likewise.
16358 * function.c (aggregate_value_p): Likewise.
16359 * ira-color.c (setup_profitable_hard_regs): Likewise.
16360 (check_hard_reg_p): Likewise.
16361 (calculate_saved_nregs): Likewise.
16362 (assign_hard_reg): Likewise.
16363 (improve_allocation): Likewise.
16364 (calculate_spill_cost): Likewise.
16365 * ira-emit.c (modify_move_list): Likewise.
16366 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16367 (ira_hard_reg_in_set_p): Likewise.
16368 * ira.c (setup_reg_mode_hard_regset): Likewise.
16369 (clarify_prohibited_class_mode_regs): Likewise.
16370 (check_allocation): Likewise.
16371 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16372 (lra_setup_reg_renumber): Likewise.
16373 (setup_try_hard_regno_pseudos): Likewise.
16374 (spill_for): Likewise.
16375 (assign_hard_regno): Likewise.
16376 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16377 * lra-constraints.c (in_class_p): Likewise.
16378 (lra_constraint_offset): Likewise.
16379 (simplify_operand_subreg): Likewise.
16380 (lra_constraints): Likewise.
16381 (split_reg): Likewise.
16382 (split_if_necessary): Likewise.
16383 (invariant_p): Likewise.
16384 (inherit_in_ebb): Likewise.
16385 * lra-lives.c (process_bb_lives): Likewise.
16386 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16387 (get_hard_regs): Likewise.
16388 (do_remat): Likewise.
16389 * lra-spills.c (assign_spill_hard_regs): Likewise.
16390 * mode-switching.c (create_pre_exit): Likewise.
16391 * postreload.c (reload_combine_recognize_pattern): Likewise.
16392 * recog.c (peep2_find_free_register): Likewise.
16393 * regcprop.c (kill_value_regno): Likewise.
16394 (set_value_regno): Likewise.
16395 (copy_value): Likewise.
16396 (maybe_mode_change): Likewise.
16397 (find_oldest_value_reg): Likewise.
16398 (copyprop_hardreg_forward_1): Likewise.
16399 * regrename.c (check_new_reg_p): Likewise.
16400 (regrename_do_replace): Likewise.
16401 * reload.c (push_reload): Likewise.
16402 (combine_reloads): Likewise.
16403 (find_dummy_reload): Likewise.
16404 (operands_match_p): Likewise.
16405 (find_reloads): Likewise.
16406 (find_equiv_reg): Likewise.
16407 (reload_adjust_reg_for_mode): Likewise.
16408 * reload1.c (count_pseudo): Likewise.
16409 (count_spilled_pseudo): Likewise.
16410 (find_reg): Likewise.
16411 (clear_reload_reg_in_use): Likewise.
16412 (free_for_value_p): Likewise.
16413 (allocate_reload_reg): Likewise.
16414 (choose_reload_regs): Likewise.
16415 (reload_adjust_reg_for_temp): Likewise.
16416 (emit_reload_insns): Likewise.
16417 (delete_output_reload): Likewise.
16418 * rtlanal.c (subreg_get_info): Likewise.
16419 * sched-deps.c (sched_analyze_reg): Likewise.
16420 * sel-sched.c (init_regs_for_mode): Likewise.
16421 (mark_unavailable_hard_regs): Likewise.
16422 (choose_best_reg_1): Likewise.
16423 (verify_target_availability): Likewise.
16424 * valtrack.c (dead_debug_insert_temp): Likewise.
16425 * var-tracking.c (track_loc_p): Likewise.
16426 (emit_note_insn_var_location): Likewise.
16427 * varasm.c (make_decl_rtl): Likewise.
16428 * reginfo.c (choose_hard_reg_mode): Likewise.
16429 (init_reg_modes_target): Refer directly to
16430 this_target_regs->x_hard_regno_nregs.
16432 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16434 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16435 instead of hard_regno_nregs.
16437 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16439 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16440 end_hard_regno instead of hard_regno_nregs.
16441 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16442 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16443 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16444 * ira-color.c (improve_allocation): Likewise.
16445 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16446 * lra-lives.c (mark_regno_live): Likewise.
16447 (mark_regno_dead): Likewise.
16448 * lra-remat.c (operand_to_remat): Likewise.
16449 * lra.c (collect_non_operand_hard_regs): Likewise.
16450 * postreload.c (reload_combine_note_store): Likewise.
16451 (move2add_valid_value_p): Likewise.
16452 * reload.c (regno_clobbered_p): Likewise.
16454 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16456 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16458 * config/v850/v850.c (v850_reorg): Likewise.
16459 * reload.c (refers_to_regno_for_reload_p): Likewise.
16460 (find_equiv_reg): Likewise.
16461 * reload1.c (reload_reg_reaches_end_p): Likewise.
16463 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16465 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16467 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16468 * config/arm/arm.c (output_move_neon): Likewise.
16469 (arm_attr_length_move_neon): Likewise.
16470 (neon_split_vcombine): Likewise.
16471 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16472 (c6x_mark_reg_written): Likewise.
16473 (c6x_dwarf_register_span): Likewise.
16474 * config/i386/i386.c (ix86_save_reg): Likewise.
16475 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16476 (rws_access_reg): Likewise.
16477 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16478 * mode-switching.c (create_pre_exit): Likewise.
16479 * ree.c (combine_reaching_defs): Likewise.
16480 (add_removable_extension): Likewise.
16481 * regcprop.c (find_oldest_value_reg): Likewise.
16482 (copyprop_hardreg_forward_1): Likewise.
16483 * reload.c (reload_inner_reg_of_subreg): Likewise.
16484 (push_reload): Likewise.
16485 (combine_reloads): Likewise.
16486 (find_dummy_reload): Likewise.
16487 (reload_adjust_reg_for_mode): Likewise.
16488 * reload1.c (find_reload_regs): Likewise.
16489 (forget_old_reloads_1): Likewise.
16490 (reload_reg_free_for_value_p): Likewise.
16491 (reload_adjust_reg_for_temp): Likewise.
16492 (emit_reload_insns): Likewise.
16493 (delete_output_reload): Likewise.
16494 * sel-sched.c (choose_best_reg_1): Likewise.
16495 (choose_best_pseudo_reg): Likewise.
16497 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16498 Alan Hayward <alan.hayward@arm.com>
16499 David Sherwood <david.sherwood@arm.com>
16501 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16502 * target.def (slow_unaligned_access): New hook.
16503 * targhooks.h (default_slow_unaligned_access): Declare.
16504 * targhooks.c (default_slow_unaligned_access): New function.
16505 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16506 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16507 * doc/tm.texi: Regenerate.
16508 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16509 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16510 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16512 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16513 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16515 (rs6000_slow_unaligned_access): New function.
16516 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16517 (expand_block_compare): Likewise.
16518 (expand_strn_compare): Likewise.
16519 (rs6000_rtx_costs): Likewise.
16520 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16521 (riscv_slow_unaligned_access): Likewise.
16522 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16523 (riscv_slow_unaligned_access_p): ...this and make static.
16524 (riscv_option_override): Update accordingly.
16525 (riscv_slow_unaligned_access): New function.
16526 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16527 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16528 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16529 (rs6000_slow_unaligned_access): New function.
16530 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16531 (rs6000_rtx_costs): Likewise.
16532 * config/rs6000/rs6000-string.c (expand_block_compare)
16533 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16534 of SLOW_UNALIGNED_ACCESS.
16535 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16536 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16537 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16538 of SLOW_UNALIGNED_ACCESS.
16539 * expmed.c (simple_mem_bitfield_p): Likewise.
16540 * expr.c (alignment_for_piecewise_move): Likewise.
16541 (emit_group_load_1): Likewise.
16542 (emit_group_store): Likewise.
16543 (copy_blkmode_from_reg): Likewise.
16544 (emit_push_insn): Likewise.
16545 (expand_assignment): Likewise.
16546 (store_field): Likewise.
16547 (expand_expr_real_1): Likewise.
16548 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16549 * lra-constraints.c (simplify_operand_subreg): Likewise.
16550 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16551 * gimple-ssa-store-merging.c: Likewise in block comment at start
16553 * tree-ssa-strlen.c: Include target.h.
16554 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16555 of SLOW_UNALIGNED_ACCESS.
16556 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16558 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16560 PR rtl-optimization/82185
16561 * expmed.c (emit_store_flag_int): Only test tem if it has been
16564 2017-09-12 Richard Biener <rguenther@suse.de>
16566 PR middle-end/82149
16567 * match.pd ((FTYPE) N CMP CST): Fix typo.
16569 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16571 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16573 (mips_near_type_p): Add 'short_call' attribute as a synonym
16575 * doc/extend.texi (short_call): Document new function attribute.
16577 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16580 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16581 assertion check that in the condition.
16582 (get_atomic_generic_size): Likewise. Before testing if parameter
16583 has pointer type, if it has array type, call for C++
16584 default_conversion to perform array-to-pointer conversion.
16586 2017-09-12 Richard Biener <rguenther@suse.de>
16588 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16589 for operations we cannot scalarize.
16591 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16593 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16594 vectors heap vectors. Clean up comments.
16595 Make visited_bbs a reference.
16596 (profitable_jump_thread_path): Make GC
16597 vectors heap vectors. Clean up comments.
16599 (convert_and_register_jump_thread_path): Make GC vectors heap
16601 (check_subpath_and_update_thread_path): Same. Clean up comments.
16602 Make visited_bbs a reference.
16603 (handle_phi): Abstract common code to to
16604 register_jump_thread_path_if_profitable.
16605 Rename VAR_BB to DEF_BB.
16607 Make GC vectors heap vectors.
16608 Make visited_bbs a reference.
16609 (handle_assignment): Same.
16610 (register_jump_thread_path_if_profitable): New.
16611 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16613 Make GC vectors heap vectors. Clean up comments.
16614 Make visited_bbs a reference.
16615 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16616 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16619 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16622 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16623 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16625 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16627 Revert r251800 and r251799.
16629 2017-09-11 Martin Jambor <mjambor@suse.cz>
16632 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16633 arguments in advance.
16634 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16635 use it to find predecessor edges.
16636 (naive_outof_ssa): Collect vector of predecessors.
16638 2017-09-08 Jason Merrill <jason@redhat.com>
16640 PR c++/70029 - ICE with ref-qualifier and -flto
16641 * langhooks.h (struct lang_hooks_for_types): Add
16642 copy_lang_qualifiers.
16643 * attribs.c (build_type_attribute_qual_variant): Use it.
16644 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16646 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16647 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16649 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16652 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16653 (*mulsi3_sp64): New instruction.
16654 (mulsi3): New expander.
16656 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16658 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16660 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16662 * sancov.c: Include memmodel.h.
16664 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16667 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16670 2017-09-07 Carl Love <cel@us.ibm.com>
16672 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16673 the sldi instruction.
16675 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16677 * sancov.c: Include tm_p.h.
16679 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16682 * output.h (switch_to_other_text_partition): New declaration.
16683 * varasm.c (switch_to_other_text_partition): New function.
16684 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16685 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16686 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16687 to the other text partition before emitting LCL label and switch back
16688 after emitting the word after it.
16690 2017-09-07 Richard Biener <rguenther@suse.de>
16692 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16693 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16694 critical edges here, after loop init.
16695 (pass_data_pre): Remove PROP_no_crit_edges flags.
16696 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16697 for valueization of call args to avoid leaking VN_TOP.
16698 (visit_use): Assert we do not visit default defs.
16699 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16700 Use error_mark_node to more easily detect leaking VN_TOP.
16701 All default-defs are varying, not VN_TOP. Mark them visited.
16702 (run_scc_vn): Make code match comment.
16704 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16706 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16707 OPTION_MASK_FLOAT128_KEYWORD.
16708 (POWERPC_MASKS): Likewise.
16709 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16710 support for the -mfloat128-type option, and make -mfloat128
16711 default on PowerPC Linux systems. Define or undefine
16712 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16713 Define __float128 to be __ieee128 if IEEE 128-bit support is
16714 enabled, or undefine it.
16715 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16716 Delete defining __FLOAT128_TYPE__.
16717 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16718 -mfloat128-type option and make -mfloat128 default on PowerPC
16720 (TARGET_FLOAT128_TYPE): Likewise.
16721 (-mfloat128-type): Likewise.
16722 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16723 Delete the -mfloat128-type option and make -mfloat128 default on
16724 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16725 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16726 128-bit floating point is enabled. Change tests from using
16727 -mfloat128-type to -mfloat128.
16728 (rs6000_mangle_type): Use the correct mangling for the __float128
16729 type even if normal long double is restricted to 64-bits.
16730 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16732 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16733 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16734 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16736 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16737 documentation for -mfloat128.
16739 2017-09-06 Olivier Hainque <hainque@adacore.com>
16741 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16743 2017-09-06 Wish Wu <wishwu007@gmail.com>
16744 Jakub Jelinek <jakub@redhat.com>
16746 * asan.c (initialize_sanitizer_builtins): Add
16747 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16748 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16749 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16750 BT_FN_VOID_UINT64_PTR variables.
16751 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16752 (BT_FN_VOID_UINT16_UINT16): Likewise.
16753 (BT_FN_VOID_UINT32_UINT32): Likewise.
16754 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16755 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16756 (BT_FN_VOID_UINT64_PTR): Likewise.
16757 * common.opt (flag_sanitize_coverage): New variable.
16758 (fsanitize-coverage=trace-pc): Remove.
16759 (fsanitize-coverage=): Add.
16760 * flag-types.h (enum sanitize_coverage_code): New enum.
16761 * fold-const.c (fold_range_test): Disable non-short-circuit
16762 optimization if flag_sanitize_coverage.
16763 (fold_truth_andor): Likewise.
16764 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16765 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16766 (coverage_sanitizer_opts): New array.
16767 (get_closest_sanitizer_option): Add OPTS argument, handle also
16768 OPT_fsanitize_coverage_.
16769 (parse_sanitizer_options): Adjusted to also handle
16770 OPT_fsanitize_coverage_.
16771 (common_handle_option): Add OPT_fsanitize_coverage_.
16772 * sancov.c (instrument_comparison, instrument_switch): New function.
16773 (sancov_pass): Add trace-cmp support.
16774 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16775 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16776 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16777 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16778 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16779 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16780 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16781 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16782 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16783 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16785 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16787 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16788 error. Only quit immediately if parsing is complete.
16789 (BEGIN): Initialize fatal_err and parse_done.
16790 (begin fpu, end fpu): Check number of arguments.
16791 (begin arch, end arch): Likewise.
16792 (begin cpu, end cpu): Likewise.
16793 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16794 (optalias): Likewise.
16796 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16798 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16799 * config/arm/arm-isa.h: Delete. Move definitions to ...
16800 * arm-cpus.in: ... here. Use new feature and fgroup values.
16801 * config/arm/arm.c (arm_option_override): Use lower case for feature
16803 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16804 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16805 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16807 (print_isa_bits_for): New function.
16808 (gen_isa): New function.
16809 (gen_comm_data): Use print_isa_bits_for.
16810 (define feature): New keyword.
16811 (define fgroup): New keyword.
16812 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16813 (arm-isa.h): Add rule to generate file.
16814 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16815 case for feature bit names.
16817 2017-09-06 Richard Biener <rguenther@suse.de>
16819 * tree-ssa-pre.c (NECESSARY): Remove.
16820 (create_expression_by_pieces): Do not touch pass-local flags.
16821 (insert_into_preds_of_block): Likewise.
16822 (do_pre_regular_insertion): Likewise.
16823 (eliminate_insert): Likewise.
16824 (eliminate_dom_walker::before_dom_children): Likewise.
16825 (fini_eliminate): Do not look at inserted_exprs.
16826 (mark_operand_necessary): Remove.
16827 (remove_dead_inserted_code): Replace with simple work-list
16828 algorithm based on inserted_exprs and SSA uses.
16829 (pass_pre::execute): Re-order fini_eliminate and
16830 remove_dead_inserted_code.
16832 2017-09-06 Olivier Hainque <hainque@adacore.com>
16834 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16835 for VxWorks 7. Adjust surrounding comments.
16837 2017-09-06 Richard Biener <rguenther@suse.de>
16839 * gimple-ssa-strength-reduction.c
16840 (find_candidates_dom_walker::before_dom_children): Also allow
16843 2017-09-06 Richard Biener <rguenther@suse.de>
16845 PR tree-optimization/82108
16846 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16847 for gap in the non-permutation SLP case.
16849 2017-09-06 Martin Jambor <mjambor@suse.cz>
16851 PR tree-optimization/82078
16852 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16853 add_access_to_work_queue...
16854 (build_accesses_from_assign): ...here.
16855 (propagate_all_subaccesses): Make sure racc is the group
16856 representative, if there is one.
16858 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16860 PR middle-end/82095
16861 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16864 2017-09-06 Richard Biener <rguenther@suse.de>
16866 * gimple-ssa-strength-reduction.c
16867 (find_candidates_dom_walker::before_doom_children): Use a
16868 type and not a mode check.
16870 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16873 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16875 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16877 2017-09-05 Jeff Law <law@redhat.com>
16879 PR tree-optimization/64910
16880 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16881 cases where we have 3 or more operands.
16883 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16885 PR middle-end/81768
16886 * omp-low.c (lower_omp_for): Recompute tree invariant if
16887 gimple_omp_for_initial/final is ADDR_EXPR.
16889 PR middle-end/81768
16890 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16891 into gimple val before gimplification fo the COND_EXPR.
16893 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16895 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16896 REGION_COPY argument.
16897 (thread_through_all_blocks): Remove unused argument to
16898 duplicate_thread_path.
16900 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16901 Alan Hayward <alan.hayward@arm.com>
16902 David Sherwood <david.sherwood@arm.com>
16904 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16905 Take a scalar_mode rather than a machine_mode.
16906 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16907 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16908 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16909 (aarch64_gen_adjusted_ldpstp): Likewise.
16910 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16912 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16913 Alan Hayward <alan.hayward@arm.com>
16914 David Sherwood <david.sherwood@arm.com>
16916 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16917 Take a scalar_int_mode instead of a machine_mode.
16918 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16919 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16920 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16921 (aarch64_simd_attr_length_rglist): Delete.
16922 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16923 a scalar_int_mode instead of a machine_mode.
16924 (aarch64_add_offset): Likewise.
16925 (aarch64_internal_mov_immediate): Likewise
16926 (aarch64_add_constant_internal): Likewise.
16927 (aarch64_add_constant): Likewise.
16928 (aarch64_movw_imm): Likewise.
16929 (aarch64_rtx_arith_op_extract_p): Likewise.
16930 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16931 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16932 Remove assert that the mode isn't a vector.
16933 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16934 (aarch64_expand_mov_immediate): Update calls after above changes.
16935 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16936 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16937 (aarch64_move_imm): Likewise.
16938 (aarch64_can_const_movi_rtx_p): Likewise.
16939 (aarch64_strip_extend): Likewise.
16940 (aarch64_extr_rtx_p): Likewise.
16941 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16942 a CONST_INT when the mode parameter is VOIDmode.
16943 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16945 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16947 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16948 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16949 (bitwise_type_for_mode): Update accordingly.
16951 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16953 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16954 * stor-layout.c (mode_for_size_tree): Likewise.
16955 (mode_for_array): Update accordingly.
16956 (layout_decl): Likewise.
16957 (compute_record_mode): Likewise. Only set the mode once.
16959 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16961 * target.def (get_mask_mode): Change return type to opt_mode.
16963 * doc/tm.texi: Regenerate.
16964 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16965 * targhooks.c (default_get_mask_mode): Likewise.
16966 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16967 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16968 targetm.get_mask_mode.
16969 * tree.c (build_truth_vector_type): Likewise.
16971 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16973 * machmode.h (mode_for_vector): Return an opt_mode.
16974 * stor-layout.c (mode_for_vector): Likewise.
16975 (mode_for_int_vector): Update accordingly.
16976 (layout_type): Likewise.
16977 * config/i386/i386.c (emit_memmov): Likewise.
16978 (ix86_expand_set_or_movmem): Likewise.
16979 (ix86_expand_vector_init): Likewise.
16980 (ix86_get_mask_mode): Likewise.
16981 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16983 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16984 * expmed.c (extract_bit_field_1): Likewise.
16985 * expr.c (expand_expr_real_2): Likewise.
16986 * optabs-query.c (can_vec_perm_p): Likewise.
16987 (can_vec_mask_load_store_p): Likewise.
16988 * optabs.c (expand_vec_perm): Likewise.
16989 * targhooks.c (default_get_mask_mode): Likewise.
16990 * tree-vect-stmts.c (vectorizable_store): Likewise.
16991 (vectorizable_load): Likewise.
16992 (get_vectype_for_scalar_type_and_size): Likewise.
16994 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16996 * machmode.h (mode_for_int_vector): New function.
16997 * stor-layout.c (mode_for_int_vector): Likewise.
16998 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16999 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17000 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17001 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17002 (s390_expand_vcond): Likewise.
17004 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17006 * machmode.h (opt_machine_mode): New type.
17007 (opt_mode<T>): Allow construction from anything that can be
17009 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17010 (mode_for_size): Return an opt_machine_mode.
17011 * stor-layout.c (mode_for_size): Likewise.
17012 (mode_for_size_tree): Update call accordingly.
17013 (bitwise_mode_for_mode): Likewise.
17014 (make_fract_type): Likewise.
17015 (make_accum_type): Likewise.
17016 * caller-save.c (replace_reg_with_saved_mem): Update call
17018 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17019 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17020 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17021 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17022 * expmed.c (extract_bit_field_1): Likewise.
17023 * reload.c (get_secondary_mem): Likewise.
17024 * varasm.c (assemble_integer): Likewise.
17025 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17028 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17030 * machmode.h (decimal_float_mode_for_size): New function.
17031 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17032 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17033 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17034 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17035 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17036 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17038 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17040 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17041 (get_builtin_sync_mode): Likewise.
17042 (expand_ifn_atomic_compare_exchange): Likewise.
17043 (expand_builtin_atomic_clear): Likewise.
17044 (expand_builtin_atomic_test_and_set): Likewise.
17045 (fold_builtin_atomic_always_lock_free): Likewise.
17046 * calls.c (compute_argument_addresses): Likewise.
17047 (emit_library_call_value_1): Likewise.
17048 (store_one_arg): Likewise.
17049 * combine.c (combine_instructions): Likewise.
17050 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17051 * config/arm/arm.c (arm_function_value): Likewise.
17052 (aapcs_allocate_return_reg): Likewise.
17053 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17054 * config/i386/i386.c (construct_container): Likewise.
17055 (ix86_gimplify_va_arg): Likewise.
17056 (ix86_expand_sse_cmp): Likewise.
17057 (emit_memmov): Likewise.
17058 (emit_memset): Likewise.
17059 (expand_small_movmem_or_setmem): Likewise.
17060 (ix86_expand_pextr): Likewise.
17061 (ix86_expand_pinsr): Likewise.
17062 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17063 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17065 * config/mips/mips.c (mips_function_value_1) Likewise.
17066 (mips_block_move_straight): Likewise.
17067 (mips_expand_ins_as_unaligned_store): Likewise.
17068 * config/powerpcspe/powerpcspe.c
17069 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17070 (rs6000_darwin64_record_arg_flush): Likewise.
17071 * config/rs6000/rs6000.c
17072 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17073 (rs6000_darwin64_record_arg_flush): Likewise.
17074 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17075 (sparc_function_value_1): Likewise.
17076 * config/spu/spu.c (adjust_operand): Likewise.
17077 (spu_emit_branch_or_set): Likewise.
17078 (arith_immediate_p): Likewise.
17079 * emit-rtl.c (gen_lowpart_common): Likewise.
17080 * expr.c (expand_expr_real_1): Likewise.
17081 * function.c (assign_parm_setup_block): Likewise.
17082 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17083 * reload1.c (alter_reg): Likewise.
17084 * stor-layout.c (mode_for_vector): Likewise.
17085 (layout_type): Likewise.
17087 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17089 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17090 (spu_convert_move): Likewise.
17091 * lower-subreg.c (resolve_simple_move): Likewise.
17093 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17096 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17097 define_insn to a define_expand.
17098 (altivec_vsum2sws_direct): New define_insn.
17099 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17101 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17103 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17106 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17111 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17112 realign stack if stack alignment needed is less than incoming
17115 2017-09-05 Marek Polacek <polacek@redhat.com>
17118 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17121 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17123 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17125 2017-09-05 Richard Biener <rguenther@suse.de>
17127 PR tree-optimization/82084
17128 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17130 2017-09-05 Richard Biener <rguenther@suse.de>
17132 PR tree-optimization/82102
17133 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17135 2017-09-05 Martin Liska <mliska@suse.cz>
17137 PR tree-optimization/82032
17138 * tree-cfg.c (generate_range_test): New function.
17139 * tree-cfg.h (generate_range_test): Declared here.
17140 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17141 (cleanup_control_expr_graph): Use it.
17142 * tree-switch-conversion.c (try_switch_expansion): Remove
17144 (emit_case_nodes): Use generate_range_test.
17146 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17149 * config/i386/i386.md (*<btsc><mode>_mask): Add
17150 TARGET_USE_BT to insn constraint.
17151 (*btr<mode>_mask): Ditto.
17153 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17155 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17156 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17158 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17161 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17162 TARGET_NEON and TARGET_IWMMXT.
17163 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17164 TARGET_NEON and TARGET_IWMMXT.
17165 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17167 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17169 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17170 (ix86_rewrite_tls_address): Ditto.
17171 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17172 (ix86_rewrite_tls_address_1): Ditto.
17173 (ix86_rewrite_tls_address): Ditto.
17174 * config/i386/predicates.md (tls_address_pattern): New predicate.
17175 * config/i386/i386.md (TLS address splitter): New splitter.
17177 2017-09-04 Richard Biener <rguenther@suse.de>
17179 PR tree-optimization/82084
17180 * fold-const.h (can_native_encode_string_p): Declare.
17181 * fold-const.c (can_native_encode_string_p): Factor out from ...
17182 (native_encode_string): ... here.
17183 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17184 vectorizing stores from constants we later cannot handle.
17186 2017-09-04 Marek Polacek <polacek@redhat.com>
17189 * doc/invoke.texi: Update -Wtautological-compare documentation.
17191 2017-09-04 Jeff Law <law@redhat.com>
17193 PR tree-optimization/64910
17194 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17195 swap the first and last operand if the last is a constant.
17197 2017-09-04 Marek Polacek <polacek@redhat.com>
17200 * convert.c (do_narrow): When sanitizing signed integer overflows,
17201 bail out for signed types.
17202 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17204 2017-09-04 Richard Biener <rguenther@suse.de>
17206 PR tree-optimization/82060
17207 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17208 Move devirtualization after stmt folding and before EH/AB/noreturn
17209 cleanup to get the stmt refs canonicalized. Use a bool instead
17210 of gimple_modified_p since that doesn't work for NOPs. Schedule
17211 NOPs generated by folding for removal.
17213 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17214 Alan Hayward <alan.hayward@arm.com>
17215 David Sherwood <david.sherwood@arm.com>
17217 * coretypes.h (pad_direction): New enum.
17218 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17219 (FUNCTION_ARG_PADDING): Likewise.
17220 * target.def (function_arg_padding): New hook.
17221 * targhooks.h (default_function_arg_padding): Declare.
17222 * targhooks.c (default_function_arg_padding): New function.
17223 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17224 (TARGET_FUNCTION_ARG_PADDING): ...this.
17225 * doc/tm.texi: Regenerate.
17226 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17227 instead of direction.
17228 (compute_argument_addresses): Likewise.
17229 (load_register_parameters): Likewise.
17230 (emit_library_call_value_1): Likewise.
17231 (store_one_arg): Use targetm.calls.function_arg_padding instead
17232 of FUNCTION_ARG_PADDING.
17233 (must_pass_in_stack_var_size_or_pad): Likewise.
17234 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17235 (emit_group_store): Likewise.
17236 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17237 instead of FUNCTION_ARG_PADDING.
17238 (emit_push_insn): Likewise, and propagate enum change throughout
17240 * function.h (direction): Delete.
17241 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17243 * function.c (assign_parm_find_stack_rtl): Likewise.
17244 (assign_parm_setup_block_p): Likewise.
17245 (assign_parm_setup_block): Likewise.
17246 (gimplify_parameters): Likewise.
17247 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17248 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17250 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17251 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17252 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17253 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17254 (aarch64_function_arg_padding): ...this new function.
17255 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17256 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17257 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17258 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17259 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17260 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17261 (arm_pad_arg_upward): Replace with...
17262 (arm_function_arg_padding): ...this new function.
17263 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17265 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17266 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17267 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17268 (ia64_hpux_function_arg_padding): Replace with...
17269 (ia64_function_arg_padding): ...this new function. Use pad_direction
17270 instead of direction. Check for TARGET_HPUX.
17271 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17272 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17273 (iq2000_function_arg_padding): New function.
17274 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17275 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17276 (mips_function_arg_padding): ...this new function.
17277 (mips_pad_reg_upward): Update accordingly.
17278 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17279 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17280 targetm.calls.function_arg_padding.
17281 (FUNCTION_ARG_PADDING): Delete.
17282 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17283 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17284 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17285 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17286 (nios2_block_reg_padding): Return pad_direction instead of direction.
17287 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17288 instead of direction.
17289 (nios2_function_arg_padding): Likewise. Make static.
17290 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17291 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17292 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17293 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17294 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17295 (pa_function_arg_padding): Make static. Return pad_direction instead
17297 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17298 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17299 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17300 instead of direction. Use targetm.calls.function_arg_padding.
17301 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17302 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17303 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17304 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17305 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17307 (function_arg_padding): Rename to...
17308 (rs6000_function_arg_padding): ...this. Make static. Return
17309 pad_direction instead of direction.
17310 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17311 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17312 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17313 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17314 instead of direction. Use targetm.calls.function_arg_padding.
17315 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17316 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17317 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17318 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17319 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17320 (function_arg_padding): Rename to...
17321 (rs6000_function_arg_padding): ...this. Make static. Return
17322 pad_direction instead of direction.
17323 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17324 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17325 * config/s390/s390.c (s390_function_arg_padding): New function.
17326 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17327 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17328 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17329 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17330 (function_arg_padding): Rename to...
17331 (sparc_function_arg_padding): ...this. Make static. Return
17332 pad_direction instead of direction.
17333 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17334 * config/spu/spu.c (spu_function_arg_padding): New function.
17335 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17336 * system.h (FUNCTION_ARG_PADDING): Poison.
17338 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17339 Alan Hayward <alan.hayward@arm.com>
17340 David Sherwood <david.sherwood@arm.com>
17342 * target.def (modes_tieable_p): New hook.
17343 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17344 (TARGET_MODES_TIEABLE_P): ...this.
17345 * doc/tm.texi.in: Regenerate.
17346 * hooks.h (hook_bool_mode_mode_true): Declare.
17347 * hooks.c (hook_bool_mode_mode_true): New function.
17348 * combine.c (subst): Use targetm.modes_tieable_p instead of
17350 * dse.c (find_shift_sequence): Likewise.
17351 * expmed.c (extract_low_bits): Likewise.
17352 * lower-subreg.c: Include target.h.
17353 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17355 * rtlanal.c (rtx_cost): Likewise.
17356 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17357 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17358 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17359 (TARGET_MODES_TIEABLE_P): Redefine.
17360 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17361 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17362 (TARGET_MODES_TIEABLE_P): Redefine.
17363 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17364 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17365 (arc_modes_tieable_p): New function.
17366 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17367 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17368 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17369 (arm_modes_tieable_p): Make static.
17370 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17371 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17372 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17373 (TARGET_MODES_TIEABLE_P): Redefine.
17374 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17375 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17376 (TARGET_MODES_TIEABLE_P): Redefine.
17377 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17378 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17379 (cr16_modes_tieable_p): New function.
17380 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17381 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17382 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17383 (TRULY_NOOP_TRUNCATION): Update comment.
17384 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17385 (TRULY_NOOP_TRUNCATION): Update comment.
17386 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17387 (frv_modes_tieable_p): New function.
17388 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17389 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17390 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17391 (TARGET_MODES_TIEABLE_P): Redefine.
17392 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17393 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17394 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17395 (TARGET_MODES_TIEABLE_P): Redefine.
17396 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17397 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17398 (ia64_modes_tieable_p): New function.
17399 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17400 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17401 (iq2000_modes_tieable_p): New function.
17402 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17403 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17404 (lm32_modes_tieable_p): New function.
17405 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17406 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17407 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17408 (TARGET_MODES_TIEABLE_P): Redefine.
17409 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17410 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17411 (m32r_modes_tieable_p): New function.
17412 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17413 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17414 (m68k_modes_tieable_p): New function.
17415 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17416 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17417 (mcore_modes_tieable_p): New function.
17418 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17419 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17421 (TARGET_MODES_TIEABLE_P): Redefine.
17422 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17423 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17424 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17425 (TARGET_MODES_TIEABLE_P): Redefine.
17426 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17427 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17428 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17429 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17430 (mn10300_modes_tieable_p): ...this and make static.
17431 (TARGET_MODES_TIEABLE_P): Redefine.
17432 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17433 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17434 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17435 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17436 (msp430_modes_tieable_p): Make static.
17437 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17438 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17439 (TARGET_MODES_TIEABLE_P): Redefine.
17440 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17441 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17442 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17443 (TARGET_MODES_TIEABLE_P): Redefine.
17444 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17445 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17446 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17447 (TARGET_MODES_TIEABLE_P): Redefine.
17448 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17449 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17450 (pdp11_modes_tieable_p): New function.
17451 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17452 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17453 (rs6000_modes_tieable_p): New function.
17454 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17455 * config/powerpcspe/powerpcspe.md: Update comment.
17456 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17457 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17458 (TARGET_MODES_TIEABLE_P): Redefine.
17459 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17460 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17461 (rl78_modes_tieable_p): New function.
17462 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17463 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17464 (rs6000_modes_tieable_p): New function.
17465 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17466 * config/rs6000/rs6000.md: Update comment.
17467 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17468 * config/rx/rx.c (rx_modes_tieable_p): New function.
17469 (TARGET_MODES_TIEABLE_P): Redefine.
17470 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17471 * config/s390/s390.c (s390_modes_tieable_p): New function.
17472 (TARGET_MODES_TIEABLE_P): Redefine.
17473 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17474 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17475 (sh_modes_tieable_p): New function.
17476 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17477 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17478 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17479 (sparc_modes_tieable_p): Make static.
17480 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17481 * config/spu/spu.c (spu_modes_tieable_p): New function.
17482 (TARGET_MODES_TIEABLE_P): Redefine.
17483 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17484 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17485 (TARGET_MODES_TIEABLE_P): Redefine.
17486 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17487 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17488 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17489 * config/v850/v850.c (v850_modes_tieable_p): New function.
17490 (TARGET_MODES_TIEABLE_P): Redefine.
17491 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17492 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17493 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17494 (visium_modes_tieable_p): New function.
17495 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17496 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17497 (xtensa_modes_tieable_p): New function.
17498 * system.h (MODES_TIEABLE_P): Poison.
17500 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17501 Alan Hayward <alan.hayward@arm.com>
17502 David Sherwood <david.sherwood@arm.com>
17504 * target.def (hard_regno_mode_ok): New hook.
17505 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17506 (TARGET_HARD_REGNO_MODE_OK): ...this.
17507 * doc/tm.texi.in: Regenerate.
17508 * hooks.h (hook_bool_uint_mode_true): Declare.
17509 * hooks.c (hook_bool_uint_mode_true): New function.
17510 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17511 HARD_REGNO_MODE_OK.
17512 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17513 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17514 instead of HARD_REGNO_MODE_OK.
17515 * caller-save.c: Include target.h.
17516 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17517 HARD_REGNO_MODE_OK.
17518 * combine.c (can_combine_p): Likewise.
17519 (combinable_i3pat): Likewise.
17520 (can_change_dest_mode): Likewise.
17521 * expr.c (init_expr_target): Likewise.
17522 (convert_move): Likewise.
17523 (convert_modes): Likewise.
17524 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17525 (setup_prohibited_mode_move_regs): Likewise.
17526 * ira.h (target_ira): Likewise.
17527 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17528 * lra-constraints.c (process_alt_operands): Likewise.
17529 (split_reg): Likewise.
17530 * recog.c (peep2_find_free_register): Likewise.
17531 * ree.c (combine_reaching_defs): Likewise.
17532 * regcprop.c (maybe_mode_change): Likewise.
17533 * reginfo.c (init_reg_sets_1): Likewise.
17534 (choose_hard_reg_mode): Likewise.
17535 (simplifiable_subregs): Likewise.
17536 * regrename.c (check_new_reg_p): Likewise.
17537 * reload.c (find_valid_class): Likewise.
17538 (find_valid_class_1): Likewise.
17539 (reload_inner_reg_of_subreg): Likewise.
17540 (push_reload): Likewise.
17541 (combine_reloads): Likewise.
17542 (find_dummy_reload): Likewise.
17543 (find_reloads): Likewise.
17544 * reload1.c (find_reg): Likewise.
17545 (set_reload_reg): Likewise.
17546 (allocate_reload_reg): Likewise.
17547 (choose_reload_regs): Likewise.
17548 (reload_adjust_reg_for_temp): Likewise.
17549 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17550 (simplify_subreg_regno): Likewise.
17551 * sel-sched.c (init_regs_for_mode): Likewise.
17552 * varasm.c (make_decl_rtl): Likewise.
17553 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17554 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17555 HARD_REGNO_MODE_OK.
17556 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17557 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17558 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17559 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17560 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17561 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17562 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17563 (arc_mode_class): Delete.
17564 (HARD_REGNO_MODE_OK): Delete.
17565 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17566 (arc_hard_regno_mode_ok): Rename old array to...
17567 (arc_hard_regno_mode_ok_modes): ...this.
17568 (arc_conditional_register_usage): Update accordingly.
17569 (arc_mode_class): Make static.
17570 (arc_hard_regno_mode_ok): New function.
17571 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17572 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17573 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17574 (arm_hard_regno_mode_ok): Make static.
17575 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17576 HARD_REGNO_MODE_OK.
17577 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17578 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17579 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17581 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17582 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17583 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17584 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17585 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17586 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17587 * config/bfin/predicates.md (valid_reg_operand): Use
17588 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17589 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17590 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17591 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17592 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17593 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17594 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17595 (cr16_hard_regno_mode_ok): Make static and return a bool.
17596 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17597 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17598 (cris_hard_regno_mode_ok): New function.
17599 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17600 (epiphany_mode_class): Delete.
17601 (HARD_REGNO_MODE_OK): Delete.
17602 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17603 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17604 (hard_regno_mode_ok): Rename to...
17605 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17606 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17607 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17608 HARD_REGNO_MODE_OK.
17609 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17610 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17611 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17612 (frv_hard_regno_mode_ok): Make static and return a bool.
17613 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17614 HARD_REGNO_MODE_OK.
17615 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17616 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17617 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17618 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17620 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17621 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17622 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17623 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17625 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17626 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17627 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17628 (ia64_hard_regno_mode_ok): New function.
17629 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17630 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17631 (iq2000_hard_regno_mode_ok): New function.
17632 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17633 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17634 (lm32_hard_regno_mode_ok): New function.
17635 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17636 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17637 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17638 instead of HARD_REGNO_MODE_OK.
17639 (m32c_hard_regno_ok): Rename to...
17640 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17641 (m32c_cannot_change_mode_class): Update accordingly.
17642 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17643 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17644 (m32r_mode_class): Delete.
17645 (HARD_REGNO_MODE_OK): Delete.
17646 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17647 (m32r_hard_regno_mode_ok): Rename to...
17648 (m32r_hard_regno_modes): ...this.
17649 (m32r_mode_class): Make static.
17650 (m32r_hard_regno_mode_ok): New function.
17651 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17652 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17653 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17654 (m68k_hard_regno_mode_ok): Make static.
17655 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17656 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17657 (mcore_hard_regno_mode_ok): New function.
17658 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17659 (HARD_REGNO_MODE_OK): Delete.
17660 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17662 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17663 (microblaze_hard_regno_mode_ok): New function.
17664 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17665 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17666 (mips_hard_regno_mode_ok): Delete.
17667 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17668 (mips_hard_regno_mode_ok_p): ...this and make static.
17669 (mips_hard_regno_mode_ok_p): Rename to...
17670 (mips_hard_regno_mode_ok_uncached): ...this.
17671 (mips_hard_regno_mode_ok): New function.
17672 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17673 of HARD_REGNO_MODE_OK.
17674 (mips_option_override): Update after above name changes.
17675 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17676 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17677 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17678 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17679 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17680 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17681 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17682 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17683 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17684 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17685 (msp430_hard_regno_mode_ok): Make static and return a bool.
17686 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17687 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17688 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17690 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17691 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17692 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17693 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17694 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17695 (PA_HARD_REGNO_MODE_OK): ...this
17696 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17697 (PA_HARD_REGNO_MODE_OK): ...this.
17698 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17699 (pa_hard_regno_mode_ok): New function.
17700 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17701 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17702 (pdp11_hard_regno_mode_ok): New function.
17703 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17704 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17706 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17708 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17709 (rs6000_hard_regno_mode_ok): Rename to...
17710 (rs6000_hard_regno_mode_ok_uncached): ...this.
17711 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17712 (rs6000_hard_regno_mode_ok): New function.
17713 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17714 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17715 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17716 (riscv_hard_regno_mode_ok): ...this and make static.
17717 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17718 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17719 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17720 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17721 (rl78_hard_regno_mode_ok): Make static and return bool.
17722 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17723 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17725 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17726 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17727 (rs6000_hard_regno_mode_ok): Rename to...
17728 (rs6000_hard_regno_mode_ok_uncached): ...this.
17729 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17730 (rs6000_hard_regno_mode_ok): New function.
17731 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17732 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17733 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17734 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17735 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17736 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17737 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17738 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17739 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17740 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17741 (sh_hard_regno_mode_ok): Make static.
17742 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17743 instead of HARD_REGNO_MODE_OK.
17744 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17745 (sparc_mode_class): Delete.
17746 (HARD_REGNO_MODE_OK): Delete.
17747 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17748 (hard_regno_mode_classes): Make static.
17749 (sparc_mode_class): Likewise.
17750 (sparc_hard_regno_mode_ok): New function.
17751 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17752 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17753 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17755 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17756 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17757 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17758 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17759 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17760 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17761 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17762 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17763 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17764 (visium_hard_regno_mode_ok): New function.
17765 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17766 instead of HARD_REGNO_MODE_OK.
17767 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17768 (HARD_REGNO_MODE_OK): Delete.
17769 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17770 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17771 (xtensa_option_override): Update accordingly.
17772 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17773 (xtensa_hard_regno_mode_ok): New function.
17774 * system.h (HARD_REGNO_MODE_OK): Poison.
17776 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17777 Alan Hayward <alan.hayward@arm.com>
17778 David Sherwood <david.sherwood@arm.com>
17780 * target.def (hard_regno_call_part_clobbered): New hook.
17781 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17782 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17783 * doc/tm.texi: Regenerate.
17784 * hooks.h (hook_bool_uint_mode_false): Declare.
17785 * hooks.c (hook_bool_uint_mode_false): New function.
17786 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17787 * cselib.c (cselib_process_insn): Use
17788 targetm.hard_regno_call_part_clobbered instead of
17789 HARD_REGNO_CALL_PART_CLOBBERED.
17790 * ira-conflicts.c (ira_build_conflicts): Likewise.
17791 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17792 * lra-constraints.c (need_for_call_save_p): Likewise.
17793 * lra-lives.c: Include target.h.
17794 (check_pseudos_live_through_calls): Use
17795 targetm.hard_regno_call_part_clobbered instead of
17796 HARD_REGNO_CALL_PART_CLOBBERED.
17797 * regcprop.c: Include target.h.
17798 (copyprop_hardreg_forward_1): Use
17799 targetm.hard_regno_call_part_clobbered instead of
17800 HARD_REGNO_CALL_PART_CLOBBERED.
17801 * reginfo.c (choose_hard_reg_mode): Likewise.
17802 * regrename.c (check_new_reg_p): Likewise.
17803 * reload.c (find_equiv_reg): Likewise.
17804 * reload1.c (emit_reload_insns): Likewise.
17805 * sched-deps.c (deps_analyze_insn): Likewise.
17806 * sel-sched.c (init_regs_for_mode): Likewise.
17807 (mark_unavailable_hard_regs): Likewise.
17808 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17809 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17810 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17812 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17813 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17814 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17816 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17818 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17819 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17820 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17822 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17823 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17824 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17826 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17827 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17829 * config/powerpcspe/powerpcspe.c
17830 (rs6000_hard_regno_call_part_clobbered): New function.
17831 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17832 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17833 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17835 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17836 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17837 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17839 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17840 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17841 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17843 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17844 Alan Hayward <alan.hayward@arm.com>
17845 David Sherwood <david.sherwood@arm.com>
17847 * rtl.h (subreg_memory_offset): Declare.
17848 * emit-rtl.c (subreg_memory_offset): New function.
17849 * expmed.c (store_bit_field_1): Use it.
17850 * expr.c (undefined_operand_subword_p): Likewise.
17851 * simplify-rtx.c (simplify_subreg): Likewise.
17853 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17855 PR rtl-optimization/57448
17858 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17859 using atomic_load pattern.
17860 (expand_atomic_store): Likewise.
17862 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17865 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17866 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17867 handling. Use replace_call_with_value with NULL instead of
17868 gsi_replace, unlink_stmt_vdef and release_defs.
17870 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17873 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17875 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17878 * rtl.h (emit_library_call_value_1): Declare.
17879 (emit_library_call): Replace declaration with a series of overloads.
17880 Remove the parameter count argument.
17881 (emit_library_call_value): Likewise.
17882 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17883 with an "rtx_mode_t *".
17884 (emit_library_call_value): Delete.
17885 (emit_library_call): Likewise.
17886 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17887 (asan_emit_allocas_unpoison): Likewise.
17888 * builtins.c (expand_builtin_powi): Likewise.
17889 (expand_asan_emit_allocas_unpoison): Likewise.
17890 * cfgexpand.c (expand_main_function): Likewise.
17891 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17892 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17893 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17894 * config/arm/arm.c (arm_trampoline_init): Likewise.
17895 (arm_call_tls_get_addr): Likewise.
17896 (arm_expand_divmod_libfunc): Likewise.
17897 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17898 (smulsi3_highpart): Likewise.
17899 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17900 (c6x_expand_compare): Likewise.
17901 (c6x_expand_movmem): Likewise.
17902 * config/frv/frv.c (frv_trampoline_init): Likewise.
17903 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17904 (ix86_expand_divmod_libfunc): Likewise.
17905 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17906 (ia64_expand_compare): Likewise.
17907 (ia64_profile_hook): Likewise.
17908 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17909 (nonlocal_goto): Likewise.
17910 (restore_stack_nonlocal): Likewise.
17911 * config/m32r/m32r.c (block_move_call): Likewise.
17912 (m32r_trampoline_init): Likewise.
17913 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17914 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17915 (m68k_call_m68k_read_tp): Likewise.
17916 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17917 (microblaze_expand_divide): Likewise.
17918 * config/mips/mips.h (mips_args): Likewise.
17919 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17920 (MIPS_ICACHE_SYNC): Likewise.
17921 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17922 (nios2_trampoline_init): Likewise.
17923 * config/pa/pa.c (hppa_tls_call): Likewise.
17924 (pa_trampoline_init): Likewise.
17925 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17926 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17927 (expand_strn_compare): Likewise.
17928 (rs6000_generate_compare): Likewise.
17929 (rs6000_expand_float128_convert): Likewise.
17930 (output_profile_hook): Likewise.
17931 (rs6000_trampoline_init): Likewise.
17932 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17933 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17934 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17935 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17936 (rs6000_generate_compare): Likewise.
17937 (rs6000_expand_float128_convert): Likewise.
17938 (output_profile_hook): Likewise.
17939 (rs6000_trampoline_init): Likewise.
17940 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17941 * config/sh/sh.c (sh_trampoline_init): Likewise.
17942 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17943 (sparc_emit_float_lib_cmp): Likewise.
17944 (sparc32_initialize_trampoline): Likewise.
17945 (sparc64_initialize_trampoline): Likewise.
17946 (sparc_profile_hook): Likewise.
17947 * config/spu/spu.c (ea_load_store): Likewise.
17948 * config/spu/spu.md (floatunssidf2): Likewise.
17949 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17950 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17951 * config/visium/visium.c (expand_block_move_4): Likewise.
17952 (expand_block_move_2): Likewise.
17953 (expand_block_move_1): Likewise.
17954 (expand_block_set_4): Likewise.
17955 (expand_block_set_2): Likewise.
17956 (expand_block_set_1): Likewise.
17957 (visium_trampoline_init): Likewise.
17958 (visium_profile_hook): Likewise.
17959 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17960 (xtensa_setup_frame_addresses): Likewise.
17961 (xtensa_trampoline_init): Likewise.
17962 * except.c (sjlj_emit_function_enter): Likewise.
17963 (sjlj_emit_function_exit): Likewise.
17964 * explow.c (allocate_dynamic_stack_space): Likewise.
17965 (probe_stack_range): Likewise.
17966 * expr.c (convert_mode_scalar): Likewise.
17967 * optabs.c (expand_binop): Likewise.
17968 (expand_twoval_binop_libfunc): Likewise.
17969 (expand_unop): Likewise.
17970 (prepare_cmp_insn): Likewise.
17971 (prepare_float_lib_cmp): Likewise.
17972 (expand_float): Likewise.
17973 (expand_fix): Likewise.
17974 (expand_fixed_convert): Likewise.
17975 (maybe_emit_sync_lock_test_and_set): Likewise.
17976 (expand_atomic_compare_and_swap): Likewise.
17977 (expand_mem_thread_fence): Likewise.
17978 (expand_atomic_fetch_op): Likewise.
17980 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17982 * doc/generic.texi (OpenACC): Adjust URL.
17983 * doc/invoke.texi (C Dialect Options): Ditto.
17985 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17987 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17988 predicate for operand 1. Add (m,<S>) constraint.
17989 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17990 Prevent memory operand 1 with register operand 2.
17992 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17994 PR rtl-optimization/82024
17995 * combine.c (try_combine): If the combination result is a PARALLEL,
17996 and we only need to retain the SET in there that would be placed
17997 at I2, check that we can place that at I3 instead, before doing so.
17999 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18002 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18004 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18005 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18008 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18009 Jeff Law <law@redhat.com>
18011 * varasm.c (bss_initializer_p): Do not put constants into .bss
18012 (categorize_decl_for_section): Handle bss_initializer_p returning
18013 false when DECL_INITIAL is NULL.
18015 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18018 * config/s390/s390.c (s390_can_inline_p): New function.
18020 2017-09-01 Jeff Law <law@redhat.com>
18022 PR tree-optimization/82052
18023 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18024 Always initialize the returned slot after a hash table miss
18025 when INSERT is true.
18027 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18029 * config/s390/s390.md (mem_signal_fence): Remove.
18030 * doc/md.texi (mem_signal_fence): Remove.
18031 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18033 * target-insns.def (mem_signal_fence): Remove.
18035 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18038 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18041 * asan.c (create_odr_indicator): Strip name encoding from assembler
18042 name before appending it after __odr_asan_.
18044 2017-09-01 Martin Liska <mliska@suse.cz>
18046 PR tree-optimization/82059
18047 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18048 frequency only when an edge is redirected.
18050 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18052 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18053 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18054 (arc_conditional_register_usage): Remove ARC600 lp_count
18056 (arc_file_start): Emit Tag_ARC_CPU_variation.
18057 (arc_can_use_doloop_p): New conditions to use ZOLs.
18058 (hwloop_fail): New function.
18059 (hwloop_optimize): Likewise.
18060 (hwloop_pattern_reg): Likewise.
18061 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18062 (arc_reorg_loops): New function, calls reorg_loops.
18063 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18064 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18066 (arc_loop_hazard): Remove function, functionality moved into
18068 (arc_hazard): Remove arc_loop_hazard call.
18069 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18070 into hwloop_optimize.
18071 (arc_label_align): Remove ZOL handling.
18072 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18073 * config/arc/arc.md (doloop_begin): Remove pattern.
18074 (doloop_begin_i): Likewise.
18075 (doloop_end_i): Likewise.
18076 (doloop_fallback): Likewise.
18077 (doloop_fallback_m): Likewise.
18078 (doloop_end): Reimplement expand.
18079 (arc_lp): New pattern for LP instruction.
18080 (loop_end): New pattern.
18081 (loop_fail): Likewise.
18082 (decrement_and_branch_until_zero): Likewise.
18083 * config/arc/arc.opt (mlpc-width): New option.
18084 * doc/invoke.texi (mlpc-width): Document option.
18086 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18088 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18089 (arc_ccfsm_advance): Fix checking for delay slots.
18090 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18092 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18094 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18096 (movhi_insn): Update store instruction constraint which are saving
18097 6-bit short immediates.
18098 (movsi_insn): Consider also short scaled load operations.
18099 (zero_extendhisi2_i): Use Usd constraint instead of T.
18100 (extendhisi2_i): Add q constraint.
18101 (arc_clzsi2): Add type and length attributes.
18102 (arc_ctzsi2): Likewise.
18103 * config/arc/constraints.md (Usc): Update constraint, the
18104 assembler can parse two relocations for a single instruction.
18106 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18108 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18109 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18111 2017-08-31 Olivier Hainque <hainque@adacore.com>
18113 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18114 match as powerpc-wrs-vxworks*.
18116 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18118 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18119 register constraint for by-element operand.
18120 (aarch64_mls_elt_merge<mode>): Likewise.
18122 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18124 * config/arc/arc.c (arc_can_follow_jump): Check for short
18127 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18129 * config.gcc: Use g.opt for arc.
18130 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18131 functionality moved to ...
18132 (legitimate_scaled_address_p): New function, ...here.
18133 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18134 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18135 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18137 (arc_override_options): Handle G option.
18138 (arc_output_pic_addr_const): Correct function definition.
18139 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18140 (arc_decl_anon_ns_mem_p): Delete.
18141 (arc_in_small_data_p): Overhaul this function to take into
18142 consideration the value given via G option.
18143 (arc_rewrite_small_data_1): Renamed and corrected old
18144 arc_rewrite_small_data function.
18145 (arc_rewrite_small_data): New function.
18146 (small_data_pattern): Don't use pic_offset_table_rtx.
18147 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18148 * config/arc/simdext.md (movmisalignv2hi): Use
18149 prepare_move_operands function.
18151 (movmisalign*): Likewise.
18152 * doc/invoke.texi (ARC options): Document -G option.
18154 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18156 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18158 * config/arc/arc.c (arc_print_operand): Output scalled address for
18159 sdata whenever is possible.
18160 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18161 load/stores are available.
18162 (compact_sda_memory_operand): Check for the alignment required by
18163 code density instructions.
18164 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18166 * config/arc/constraints.md (Usd): Update constraint.
18167 (Us0): New constraint.
18168 (Usc): Update constraint.
18170 2017-08-31 Richard Biener <rguenther@suse.de>
18172 PR middle-end/82054
18173 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18174 function only once.
18176 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18178 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18179 Resize type_signature.
18181 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18182 Alan Hayward <alan.hayward@arm.com>
18183 David Sherwood <david.sherwood@arm.com>
18185 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18186 subregs whose inner modes can be stored in GPRs.
18187 (aarch64_classify_index): Likewise.
18189 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18190 Alan Hayward <alan.hayward@arm.com>
18191 David Sherwood <david.sherwood@arm.com>
18193 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18194 (V_INT_EQUIV): ...this.
18195 (v_cmp_result): Rename to...
18196 (v_int_equiv): ...this.
18197 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18198 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18199 (copysign<mode>3): Likewise.
18200 (aarch64_simd_bsl<mode>_internal): Likewise.
18201 (aarch64_simd_bsl<mode>): Likewise.
18202 (vec_cmp<mode><mode>): Likewise.
18203 (vcond<mode><mode>): Likewise.
18204 (vcond<v_cmp_mixed><mode>): Likewise.
18205 (vcondu<mode><v_cmp_mixed>): Likewise.
18206 (aarch64_cm<optab><mode>): Likewise.
18207 (aarch64_cmtst<mode>): Likewise.
18208 (aarch64_fac<optab><mode>): Likewise.
18209 (vec_perm_const<mode>): Likewise.
18210 (vcond_mask_<mode><v_cmp_result>): Rename to...
18211 (vcond_mask_<mode><v_int_equiv>): ...this.
18212 (vec_cmp<mode><v_cmp_result>): Rename to...
18213 (vec_cmp<mode><v_int_equiv>): ...this.
18215 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18216 Alan Hayward <alan.hayward@arm.com>
18217 David Sherwood <david.sherwood@arm.com>
18219 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18221 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18222 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18223 (UNSPEC_LD4_DREG): New unspecs.
18224 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18225 (aarch64_ld2<mode>_dreg_be): Replace with...
18226 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18228 (aarch64_ld3<mode>_dreg_le)
18229 (aarch64_ld3<mode>_dreg_be): Replace with...
18230 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18232 (aarch64_ld4<mode>_dreg_le)
18233 (aarch64_ld4<mode>_dreg_be): Replace with...
18234 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18237 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18239 PR tree-optimization/81987
18240 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18241 insert an initializer in a location not dominated by the stride
18244 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18246 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18247 on the entire header of the finally block in the fallthru case.
18249 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18251 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18253 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18255 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18256 rs6000_emit_move_from_cr and call renamed function.
18257 (rs6000_emit_prologue): Call renamed functions.
18258 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18259 movesi_from_cr, remove volatile CRs.
18261 2017-08-30 Jon Beniston <jon@beniston.com>
18262 Richard Biener <rguenther@suse.de>
18264 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18265 of VECTOR_MODE_P check.
18266 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18267 element vector types.
18269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18271 * df.h (df_read_modify_subreg_p): Remove in favor of...
18272 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18273 const_rtx instead of an rtx.
18274 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18275 * df-problems.c (df_word_lr_mark_ref): Likewise.
18276 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18277 (mark_pseudo_reg_dead): Likewise.
18278 (mark_ref_dead): Likewise.
18279 * reginfo.c (init_subregs_of_mode): Likewise.
18280 * sched-deps.c (sched_analyze_1): Likewise.
18281 * df-scan.c (df_def_record_1): Likewise.
18282 (df_uses_record): Likewise.
18283 (df_read_modify_subreg_p): Remove in favor of...
18284 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18285 const_rtx instead of an rtx.
18287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18288 Alan Hayward <alan.hayward@arm.com>
18289 David Sherwood <david.sherwood@arm.com>
18291 * rtl.h (partial_subreg_p): New function.
18292 * caller-save.c (save_call_clobbered_regs): Use it.
18293 * calls.c (expand_call): Likewise.
18294 * combine.c (combinable_i3pat): Likewise.
18295 (simplify_set): Likewise.
18296 (make_extraction): Likewise.
18297 (make_compound_operation_int): Likewise.
18298 (gen_lowpart_or_truncate): Likewise.
18299 (force_to_mode): Likewise.
18300 (make_field_assignment): Likewise.
18301 (reg_truncated_to_mode): Likewise.
18302 (record_truncated_value): Likewise.
18303 (move_deaths): Likewise.
18304 * cse.c (record_jump_cond): Likewise.
18305 (cse_insn): Likewise.
18306 * cselib.c (cselib_lookup_1): Likewise.
18307 * expmed.c (extract_bit_field_using_extv): Likewise.
18308 * function.c (assign_parm_setup_reg): Likewise.
18309 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18310 * ira-build.c (create_insn_allocnos): Likewise.
18311 * lra-coalesce.c (merge_pseudos): Likewise.
18312 * lra-constraints.c (match_reload): Likewise.
18313 (simplify_operand_subreg): Likewise.
18314 (curr_insn_transform): Likewise.
18315 * lra-lives.c (process_bb_lives): Likewise.
18316 * lra.c (new_insn_reg): Likewise.
18317 (lra_substitute_pseudo): Likewise.
18318 * regcprop.c (mode_change_ok): Likewise.
18319 (maybe_mode_change): Likewise.
18320 (copyprop_hardreg_forward_1): Likewise.
18321 * reload.c (push_reload): Likewise.
18322 (find_reloads): Likewise.
18323 (find_reloads_subreg_address): Likewise.
18324 * reload1.c (alter_reg): Likewise.
18325 (eliminate_regs_1): Likewise.
18326 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18328 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18330 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18331 back to if statements, including unpack.
18333 2017-08-30 Martin Liska <mliska@suse.cz>
18335 PR inline-asm/82001
18336 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18338 (func_checker::compare_asm_inputs_outputs): ... this function.
18339 (func_checker::compare_gimple_asm): Use the function to compare
18340 also ASM constrains.
18341 * ipa-icf-gimple.h: Rename the function.
18343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18344 Alan Hayward <alan.hayward@arm.com>
18345 David Sherwood <david.sherwood@arm.com>
18347 * coretypes.h (complex_mode): New type.
18348 * gdbhooks.py (build_pretty_printer): Handle it.
18349 * machmode.h (complex_mode): New class.
18350 (complex_mode::includes_p): New function.
18351 (is_complex_int_mode): Likewise.
18352 (is_complex_float_mode): Likewise.
18353 * genmodes.c (get_mode_class): Handle complex mode classes.
18354 * function.c (expand_function_end): Use is_complex_int_mode.
18356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18357 Alan Hayward <alan.hayward@arm.com>
18358 David Sherwood <david.sherwood@arm.com>
18360 * coretypes.h (scalar_mode_pod): New typedef.
18361 * gdbhooks.py (build_pretty_printer): Handle it.
18362 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18363 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18364 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18365 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18366 as_a <scalar_mode>.
18368 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18369 Alan Hayward <alan.hayward@arm.com>
18370 David Sherwood <david.sherwood@arm.com>
18372 * machmode.h (mode_for_vector): Take a scalar_mode instead
18374 * stor-layout.c (mode_for_vector): Likewise.
18375 * explow.c (promote_mode): Use as_a <scalar_mode>.
18376 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18379 Alan Hayward <alan.hayward@arm.com>
18380 David Sherwood <david.sherwood@arm.com>
18382 * target.def (preferred_simd_mode): Take a scalar_mode
18383 instead of a machine_mode.
18384 * targhooks.h (default_preferred_simd_mode): Likewise.
18385 * targhooks.c (default_preferred_simd_mode): Likewise.
18386 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18387 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18388 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18389 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18390 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18391 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18392 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18393 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18395 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18396 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18397 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18398 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18399 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18400 * doc/tm.texi: Regenerate.
18401 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18405 Alan Hayward <alan.hayward@arm.com>
18406 David Sherwood <david.sherwood@arm.com>
18408 * target.def (scalar_mode_supported_p): Take a scalar_mode
18409 instead of a machine_mode.
18410 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18411 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18412 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18413 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18414 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18415 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18416 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18417 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18418 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18419 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18420 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18421 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18422 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18423 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18425 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18426 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18427 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18428 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18429 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18431 * doc/tm.texi: Regenerate.
18433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18434 Alan Hayward <alan.hayward@arm.com>
18435 David Sherwood <david.sherwood@arm.com>
18437 * coretypes.h (opt_scalar_mode): New typedef.
18438 * gdbhooks.py (build_pretty_printers): Handle it.
18439 * machmode.h (mode_iterator::get_2xwider): Add overload for
18441 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18443 * expr.c (convert_mode_scalar): Likewise.
18444 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18445 * optabs.c (expand_float): Likewise.
18446 (expand_fix): Likewise.
18447 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18450 Alan Hayward <alan.hayward@arm.com>
18451 David Sherwood <david.sherwood@arm.com>
18453 * optabs.c (expand_float): Explicitly check for scalars before
18454 using a branching expansion.
18455 (expand_fix): Likewise.
18457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18458 Alan Hayward <alan.hayward@arm.com>
18459 David Sherwood <david.sherwood@arm.com>
18461 * expr.c (convert_mode): Split scalar handling out into...
18462 (convert_mode_scalar): ...this new function. Treat the modes
18465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18466 Alan Hayward <alan.hayward@arm.com>
18467 David Sherwood <david.sherwood@arm.com>
18469 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18471 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18474 Alan Hayward <alan.hayward@arm.com>
18475 David Sherwood <david.sherwood@arm.com>
18477 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18478 rather than a machine_mode.
18479 (fixed_from_string): Likewise.
18480 (fixed_convert): Likewise.
18481 (fixed_convert_from_int): Likewise.
18482 (fixed_convert_from_real): Likewise.
18483 (real_convert_from_fixed): Likewise.
18484 * fixed-value.c (fixed_from_double_int): Likewise.
18485 (fixed_from_string): Likewise.
18486 (fixed_convert): Likewise.
18487 (fixed_convert_from_int): Likewise.
18488 (fixed_convert_from_real): Likewise.
18489 (real_convert_from_fixed): Likewise.
18490 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18493 Alan Hayward <alan.hayward@arm.com>
18494 David Sherwood <david.sherwood@arm.com>
18496 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18497 of separate mode class checks. Do not allow vector modes here.
18498 (immed_wide_int_const): Use as_a <scalar_mode>.
18499 * explow.c (trunc_int_for_mode): Likewise.
18500 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18501 (wi::shwi): Likewise.
18502 (wi::min_value): Likewise.
18503 (wi::max_value): Likewise.
18504 * dwarf2out.c (loc_descriptor): Likewise.
18505 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18506 for CONST_WIDE_INT.
18508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18509 Alan Hayward <alan.hayward@arm.com>
18510 David Sherwood <david.sherwood@arm.com>
18512 * tree.h (SCALAR_TYPE_MODE): New macro.
18513 * expr.c (expand_expr_addr_expr_1): Use it.
18514 (expand_expr_real_2): Likewise.
18515 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18516 (fold_convert_const_fixed_from_int): Likewise.
18517 (fold_convert_const_fixed_from_real): Likewise.
18518 (native_encode_fixed): Likewise
18519 (native_encode_complex): Likewise
18520 (native_encode_vector): Likewise.
18521 (native_interpret_fixed): Likewise.
18522 (native_interpret_real): Likewise.
18523 (native_interpret_complex): Likewise.
18524 (native_interpret_vector): Likewise.
18525 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18526 (simd_clone_adjust_argument_types): Likewise.
18527 (simd_clone_init_simd_arrays): Likewise.
18528 (simd_clone_adjust): Likewise.
18529 * stor-layout.c (layout_type): Likewise.
18530 * tree.c (build_minus_one_cst): Likewise.
18531 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18532 * tree-inline.c (estimate_move_cost): Likewise.
18533 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18534 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18535 (vectorizable_reduction): Likewise.
18536 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18537 (vect_recog_mixed_size_cond_pattern): Likewise.
18538 (check_bool_pattern): Likewise.
18539 (adjust_bool_pattern): Likewise.
18540 (search_type_for_mask_1): Likewise.
18541 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18542 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18543 (vectorizable_load): Likewise.
18544 (vectorizable_store): Likewise.
18545 * ubsan.c (ubsan_encode_value): Likewise.
18546 * varasm.c (output_constant): Likewise.
18548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18549 Alan Hayward <alan.hayward@arm.com>
18550 David Sherwood <david.sherwood@arm.com>
18552 * coretypes.h (scalar_mode): New class.
18553 * machmode.h (scalar_mode): Likewise.
18554 (scalar_mode::includes_p): New function.
18555 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18556 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18557 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18558 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18559 * expmed.c (store_bit_field_1): Likewise.
18560 (extract_bit_field_1): Likewise.
18561 * expr.c (write_complex_part): Likewise.
18562 (read_complex_part): Likewise.
18563 (emit_move_complex_push): Likewise.
18564 (expand_expr_real_2): Likewise.
18565 * function.c (assign_parm_setup_reg): Likewise.
18566 (assign_parms_unsplit_complex): Likewise.
18567 * optabs.c (expand_binop): Likewise.
18568 * rtlanal.c (subreg_get_info): Likewise.
18569 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18570 * varasm.c (output_constant_pool_2): Likewise.
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 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18578 (expmed_mult_highpart_optab): Likewise.
18579 (expmed_mult_highpart): Likewise.
18581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18582 Alan Hayward <alan.hayward@arm.com>
18583 David Sherwood <david.sherwood@arm.com>
18585 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18586 instead of a machine_mode.
18587 (builtin_memset_read_str): Likewise.
18588 * builtins.c (c_readstr): Likewise.
18589 (builtin_memcpy_read_str): Likewise.
18590 (builtin_strncpy_read_str): Likewise.
18591 (builtin_memset_read_str): Likewise.
18592 (builtin_memset_gen_str): Likewise.
18593 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18594 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18595 instead of a machine_mode.
18596 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18598 (make_extraction): Likewise.
18599 (try_widen_shift_mode): Take and return scalar_int_modes instead
18601 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18602 a scalar_int_mode instead of a machine_mode.
18603 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18604 (avr_addr_space_pointer_mode): Likewise.
18605 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18606 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18607 (msp430_unwind_word_mode): Likewise.
18608 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18609 (spu_addr_space_pointer_mode): Likewise.
18610 (spu_addr_space_address_mode): Likewise.
18611 (spu_libgcc_cmp_return_mode): Likewise.
18612 (spu_libgcc_shift_count_mode): Likewise.
18613 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18614 (rl78_addr_space_pointer_mode): Likewise.
18615 (fl78_unwind_word_mode): Likewise.
18616 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18618 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18619 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18620 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18621 (mips_valid_pointer_mode): Likewise.
18622 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18623 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18624 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18626 (ft32_addr_space_address_mode): Likewise.
18627 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18628 scalar_int_mode instead of a machine_mode.
18629 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18631 (m32c_addr_space_address_mode): Likewise.
18632 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18633 (rs6000_eh_return_filter_mode): Likewise.
18634 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18635 (rs6000_eh_return_filter_mode): Likewise.
18636 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18637 (s390_libgcc_shift_count_mode): Likewise.
18638 (s390_unwind_word_mode): Likewise.
18639 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18641 * target.def (mode_rep_extended): Likewise.
18642 (valid_pointer_mode): Likewise.
18643 (addr_space.valid_pointer_mode): Likewise.
18644 (eh_return_filter_mode): Return a scalar_int_mode rather than
18646 (libgcc_cmp_return_mode): Likewise.
18647 (libgcc_shift_count_mode): Likewise.
18648 (unwind_word_mode): Likewise.
18649 (addr_space.pointer_mode): Likewise.
18650 (addr_space.address_mode): Likewise.
18651 * doc/tm.texi: Regenerate.
18652 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18654 (do_jump): Use scalar_int_mode for local variables.
18655 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18656 rather than a machine_mode.
18657 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18658 (scompare_loc_descriptor_wide): Likewise.
18659 (scompare_loc_descriptor_narrow): Likewise.
18660 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18662 * except.c (sjlj_emit_dispatch_table): Likewise.
18663 (expand_builtin_eh_copy_values): Likewise.
18664 * explow.c (convert_memory_address_addr_space_1): Likewise.
18665 Take a scalar_int_mode rather than a machine_mode.
18666 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18667 than a machine_mode.
18668 (memory_address_addr_space): Use scalar_int_mode for local variables.
18669 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18670 rather than a machine_mode.
18671 * expmed.c (mask_rtx): Likewise.
18672 (init_expmed_one_conv): Likewise.
18673 (expand_mult_highpart_adjust): Likewise.
18674 (extract_high_half): Likewise.
18675 (expmed_mult_highpart_optab): Likewise.
18676 (expmed_mult_highpart): Likewise.
18677 (expand_smod_pow2): Likewise.
18678 (expand_sdiv_pow2): Likewise.
18679 (emit_store_flag_int): Likewise.
18680 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18682 (extract_low_bits): Likewise.
18683 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18685 * expr.c (pieces_addr::adjust): Likewise.
18686 (can_store_by_pieces): Likewise.
18687 (store_by_pieces): Likewise.
18688 (clear_by_pieces_1): Likewise.
18689 (expand_expr_addr_expr_1): Likewise.
18690 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18691 (expand_expr_real_1): Likewise.
18692 (try_casesi): Likewise.
18693 * final.c (shorten_branches): Likewise.
18694 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18695 type of "mode" to machine_mode.
18696 * internal-fn.c (expand_arith_overflow_result_store): Take a
18697 scalar_int_mode rather than a machine_mode.
18698 (expand_mul_overflow): Use scalar_int_mode for local variables.
18699 * loop-doloop.c (doloop_modify): Likewise.
18700 (doloop_optimize): Likewise.
18701 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18702 than a machine_mode.
18703 (expand_doubleword_shift_condmove): Likewise.
18704 (expand_doubleword_shift): Likewise.
18705 (expand_doubleword_clz): Likewise.
18706 (expand_doubleword_popcount): Likewise.
18707 (expand_doubleword_parity): Likewise.
18708 (expand_absneg_bit): Use scalar_int_mode for local variables.
18709 (prepare_float_lib_cmp): Likewise.
18710 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18711 rather than a machine_mode.
18712 (convert_memory_address_addr_space): Likewise.
18713 (get_mode_bounds): Likewise.
18714 (get_address_mode): Return a scalar_int_mode rather than a
18716 * rtlanal.c (get_address_mode): Likewise.
18717 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18718 than a machine_mode.
18719 * targhooks.c (default_mode_rep_extended): Likewise.
18720 (default_valid_pointer_mode): Likewise.
18721 (default_addr_space_valid_pointer_mode): Likewise.
18722 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18723 than a machine_mode.
18724 (default_libgcc_cmp_return_mode): Likewise.
18725 (default_libgcc_shift_count_mode): Likewise.
18726 (default_unwind_word_mode): Likewise.
18727 (default_addr_space_pointer_mode): Likewise.
18728 (default_addr_space_address_mode): Likewise.
18729 * targhooks.h (default_eh_return_filter_mode): Likewise.
18730 (default_libgcc_cmp_return_mode): Likewise.
18731 (default_libgcc_shift_count_mode): Likewise.
18732 (default_unwind_word_mode): Likewise.
18733 (default_addr_space_pointer_mode): Likewise.
18734 (default_addr_space_address_mode): Likewise.
18735 (default_mode_rep_extended): Take a scalar_int_mode rather than
18737 (default_valid_pointer_mode): Likewise.
18738 (default_addr_space_valid_pointer_mode): Likewise.
18739 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18741 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18742 rather than a machine_mode.
18743 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18744 for local variables.
18745 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18746 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18747 than a machine_mode.
18749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18750 Alan Hayward <alan.hayward@arm.com>
18751 David Sherwood <david.sherwood@arm.com>
18753 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18754 the mode argument to scalar_int_mode.
18755 (do_jump_by_parts_zero_rtx): Likewise.
18756 (do_jump_by_parts_equality_rtx): Likewise.
18757 (do_jump_by_parts_greater): Take a mode argument.
18758 (do_jump_by_parts_equality): Likewise.
18759 (do_jump_1): Update calls accordingly.
18761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18762 Alan Hayward <alan.hayward@arm.com>
18763 David Sherwood <david.sherwood@arm.com>
18765 * is-a.h (safe_dyn_cast): New function.
18766 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18767 (jump_table_for_label): Likewise.
18768 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18769 instead of an rtx_insn *.
18770 (shorten_branches): Use dyn_cast instead of LABEL_P and
18771 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18772 rtx_jump_table_data::get_data_mode.
18773 (final_scan_insn): Likewise.
18775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18776 Alan Hayward <alan.hayward@arm.com>
18777 David Sherwood <david.sherwood@arm.com>
18779 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18780 trying to combine a full-register integer set with a subreg
18783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18784 Alan Hayward <alan.hayward@arm.com>
18785 David Sherwood <david.sherwood@arm.com>
18787 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18788 that is always either address_mode or pointer_mode.
18790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18791 Alan Hayward <alan.hayward@arm.com>
18792 David Sherwood <david.sherwood@arm.com>
18794 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18795 when the two are known to be equal.
18797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18798 Alan Hayward <alan.hayward@arm.com>
18799 David Sherwood <david.sherwood@arm.com>
18801 * simplify-rtx.c (simplify_const_unary_operation): Use
18802 is_a <scalar_int_mode> instead of checking for a nonzero
18803 precision. Forcibly convert op_mode to a scalar_int_mode
18804 in that case. More clearly differentiate the operand and
18805 result modes and use the former when deciding what the value
18806 of a count-bits operation should be. Use is_int_mode instead
18807 of checking for a MODE_INT. Remove redundant check for whether
18808 this mode has a zero precision.
18810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18811 Alan Hayward <alan.hayward@arm.com>
18812 David Sherwood <david.sherwood@arm.com>
18814 * optabs.c (widen_leading): Change the type of the mode argument
18815 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18816 (widen_bswap): Likewise.
18817 (expand_parity): Likewise.
18818 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18819 (expand_ffs): Likewise.
18820 (epand_unop): Check for scalar integer modes before calling the
18823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18824 Alan Hayward <alan.hayward@arm.com>
18825 David Sherwood <david.sherwood@arm.com>
18827 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18829 (expand_expr_real_1): Update call accordingly.
18831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18832 Alan Hayward <alan.hayward@arm.com>
18833 David Sherwood <david.sherwood@arm.com>
18835 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18836 value_mode arguments. Use scalar_int_mode internally.
18837 (store_bit_field_1): Rename the new integer mode from imode
18838 to op0_mode and use it instead of GET_MODE (op0). Update calls
18839 to store_split_bit_field, store_bit_field_using_insv and
18840 store_fixed_bit_field.
18841 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18842 Use scalar_int_mode internally. Use a bit count rather than a mode
18843 when calculating the largest bit size for get_best_mode.
18844 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18845 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18846 Remove assertion that OP0 has a scalar integer mode.
18847 (store_split_bit_field): Add op0_mode and value_mode arguments.
18848 Update calls to extract_fixed_bit_field.
18849 (extract_bit_field_using_extv): Add an op0_mode argument.
18850 Use scalar_int_mode internally.
18851 (extract_bit_field_1): Rename the new integer mode from imode to
18852 op0_mode and use it instead of GET_MODE (op0). Update calls to
18853 extract_split_bit_field, extract_bit_field_using_extv and
18854 extract_fixed_bit_field.
18855 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18856 to extract_split_bit_field and extract_fixed_bit_field_1.
18857 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18858 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18859 on the target mode.
18860 (extract_split_bit_field): Add an op0_mode argument. Update call
18861 to extract_fixed_bit_field.
18863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18864 Alan Hayward <alan.hayward@arm.com>
18865 David Sherwood <david.sherwood@arm.com>
18867 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18868 * explow.c (hard_function_value): Likewise.
18869 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18870 convert_to_mode call outside the loop.
18871 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18872 for the mode iterator. Require the mode specified by max_pieces
18874 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18876 (copy_blkmode_to_reg): Likewise.
18877 (set_storage_via_setmem): Likewise.
18878 * optabs.c (prepare_cmp_insn): Likewise.
18879 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18880 * stor-layout.c (finish_bitfield_representative): Likewise.
18882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18883 Alan Hayward <alan.hayward@arm.com>
18884 David Sherwood <david.sherwood@arm.com>
18886 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18887 * expr.c (convert_move): Use them.
18888 (convert_modes): Likewise.
18889 (store_expr_with_bounds): Likewise.
18891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18892 Alan Hayward <alan.hayward@arm.com>
18893 David Sherwood <david.sherwood@arm.com>
18895 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18896 parameter for the mode of "x". Remove the "known_x", "known_mode"
18897 and "known_ret" arguments. Change the type of the mode argument
18898 to scalar_int_mode.
18899 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18900 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18901 (reg_num_sign_bit_copies_for_combine): Likewise.
18902 * rtlanal.c (nonzero_bits1): Likewise.
18903 (num_sign_bit_copies1): Likewise.
18904 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18905 (reg_num_sign_bit_copies_general): Likewise.
18906 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18907 (reg_nonzero_bits_general): Likewise.
18909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18910 Alan Hayward <alan.hayward@arm.com>
18911 David Sherwood <david.sherwood@arm.com>
18913 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18914 than in subroutines. Return 1 for non-integer modes.
18915 (cached_num_sign_bit_copies): Change the type of the mode parameter
18916 to scalar_int_mode.
18917 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18918 classes. Handle CONST_INT_P first and then check whether X also
18919 has a scalar integer mode. Check the same thing for inner registers
18920 of a SUBREG and for values that are being extended or truncated.
18922 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18923 Alan Hayward <alan.hayward@arm.com>
18924 David Sherwood <david.sherwood@arm.com>
18926 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18927 in subroutines. Return the mode mask for non-integer modes.
18928 (cached_nonzero_bits): Change the type of the mode parameter
18929 to scalar_int_mode.
18930 (nonzero_bits1): Likewise. Remove early exit for other mode
18931 classes. Handle CONST_INT_P first and then check whether X
18932 also has a scalar integer mode.
18934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18935 Alan Hayward <alan.hayward@arm.com>
18936 David Sherwood <david.sherwood@arm.com>
18938 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18939 Return a scalar_int_mode and assert that the size is greater than
18941 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18942 (op_by_pieces_d::op_by_pieces_d): Likewise.
18943 (op_by_pieces_d::run): Likewise.
18944 (can_store_by_pieces): Likewise.
18946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18947 Alan Hayward <alan.hayward@arm.com>
18948 David Sherwood <david.sherwood@arm.com>
18950 * combine.c (extract_left_shift): Add a mode argument and update
18952 (make_compound_operation_int): Change the type of the mode parameter
18953 to scalar_int_mode and update the call to extract_left_shift.
18955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18956 Alan Hayward <alan.hayward@arm.com>
18957 David Sherwood <david.sherwood@arm.com>
18959 * combine.c (simplify_and_const_int): Change the type of the mode
18960 parameter to scalar_int_mode.
18961 (simplify_and_const_int_1): Likewise. Update recursive call.
18963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18964 Alan Hayward <alan.hayward@arm.com>
18965 David Sherwood <david.sherwood@arm.com>
18967 * combine.c (simplify_compare_const): Check that the mode is a
18968 scalar_int_mode (rather than VOIDmode) before testing its
18970 (simplify_comparison): Move COMPARISON_P handling out of the
18971 loop and restrict the latter part of the loop to scalar_int_modes.
18972 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18973 and when considering SUBREG_REGs. Use is_int_mode instead of
18974 checking GET_MODE_CLASS against MODE_INT.
18976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18977 Alan Hayward <alan.hayward@arm.com>
18978 David Sherwood <david.sherwood@arm.com>
18980 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18981 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18982 shift_unit_mode and for modes involved in scalar shifts.
18984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18985 Alan Hayward <alan.hayward@arm.com>
18986 David Sherwood <david.sherwood@arm.com>
18988 * combine.c (force_int_to_mode): New function, split out from...
18989 (force_to_mode): ...here. Keep xmode up-to-date and use it
18990 instead of GET_MODE (x).
18992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18993 Alan Hayward <alan.hayward@arm.com>
18994 David Sherwood <david.sherwood@arm.com>
18996 * optabs-query.h (extraction_insn::struct_mode): Change type to
18997 opt_scalar_int_mode and update comment.
18998 (extraction_insn::field_mode): Change type to scalar_int_mode.
18999 (extraction_insn::pos_mode): Likewise.
19000 * combine.c (make_extraction): Update accordingly.
19001 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19002 (get_optab_extraction_insn): Likewise.
19003 * recog.c (simplify_while_replacing): Likewise.
19004 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19005 parameter to opt_scalar_int_mode.
19007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19008 Alan Hayward <alan.hayward@arm.com>
19009 David Sherwood <david.sherwood@arm.com>
19011 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19012 to a scalar_int_mode instead of a machine_mode.
19013 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19014 (get_best_mode): Return a boolean and use a pointer argument to store
19015 the selected mode. Replace the limit mode parameter with a bit limit.
19016 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19017 for the values returned by bit_field_mode_iterator::next_mode.
19018 (store_bit_field): Update call to get_best_mode.
19019 (store_fixed_bit_field): Likewise.
19020 (extract_fixed_bit_field): Likewise.
19021 * expr.c (optimize_bitfield_assignment_op): Likewise.
19022 * fold-const.c (optimize_bit_field_compare): Likewise.
19023 (fold_truth_andor_1): Likewise.
19024 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19025 Update for new type of m_mode.
19026 (get_best_mode): As above.
19028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19029 Alan Hayward <alan.hayward@arm.com>
19030 David Sherwood <david.sherwood@arm.com>
19032 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19033 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19034 (store_bit_field): Check is_a <scalar_int_mode> before calling
19035 strict_volatile_bitfield_p.
19036 (extract_bit_field): Likewise.
19038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19039 Alan Hayward <alan.hayward@arm.com>
19040 David Sherwood <david.sherwood@arm.com>
19042 * target.def (cstore_mode): Return a scalar_int_mode.
19043 * doc/tm.texi: Regenerate.
19044 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19045 * targhooks.h (default_cstore_mode): Likewise.
19046 * targhooks.c (default_cstore_mode): Likewise, using a forced
19048 * expmed.c (emit_cstore): Expect the target of the cstore to be
19051 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19052 Alan Hayward <alan.hayward@arm.com>
19053 David Sherwood <david.sherwood@arm.com>
19055 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19057 (niter_desc): Likewise mode.
19058 (iv_analyze): Add a mode parameter.
19060 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19061 and change its type to scalar_int_mode.
19062 * loop-iv.c: Update commentary at head of file.
19063 (iv_constant): Pass the mode paraeter before the rtx it describes
19064 and change its type to scalar_int_mode. Remove VOIDmode handling.
19065 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19066 (iv_extend): Likewise.
19067 (shorten_into_mode): Likewise.
19068 (iv_add): Use scalar_int_mode.
19069 (iv_mult): Likewise.
19070 (iv_shift): Likewise.
19071 (canonicalize_iv_subregs): Likewise.
19072 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19073 it describes and change its mode to scalar_int_mode. Also change
19074 the type of the returned inner_mode to scalar_int_mode.
19075 (get_biv_step): Likewise, turning outer_mode from a pointer
19076 into a direct parameter. Update call to get_biv_step_1.
19077 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19078 iv_constant and get_biv_step.
19079 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19080 and change its type to scalar_int_mode. Don't initialise iv->mode
19081 to VOIDmode and remove later checks for its still being VOIDmode.
19082 Update calls to iv_analyze_op and iv_analyze_expr. Check
19083 is_a <scalar_int_mode> when changing the mode under consideration.
19084 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19085 Update call to iv_analyze_expr.
19086 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19087 inner register is not also a scalar_int_mode. Update call to
19089 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19090 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19091 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19092 separate mode class checks. Update calls to iv_analyze. Remove
19093 fix-up of VOIDmodes after iv_analyze_biv.
19094 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19095 don't have a scalar_int_mode. Update call to biv_p.
19097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19098 Alan Hayward <alan.hayward@arm.com>
19099 David Sherwood <david.sherwood@arm.com>
19101 * cfgexpand.c (convert_debug_memory_address): Use
19102 as_a <scalar_int_mode>.
19103 * combine.c (expand_compound_operation): Likewise.
19104 (make_extraction): Likewise.
19105 (change_zero_ext): Likewise.
19106 (simplify_comparison): Likewise.
19107 * cse.c (cse_insn): Likewise.
19108 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19109 (mem_loc_descriptor): Likewise.
19110 (loc_descriptor): Likewise.
19111 * expmed.c (init_expmed_one_mode): Likewise.
19112 (synth_mult): Likewise.
19113 (emit_store_flag_1): Likewise.
19114 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19115 of a comparison with size.
19116 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19117 (reduce_to_bit_field_precision): Likewise.
19118 * function.c (expand_function_end): Likewise.
19119 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19120 * loop-doloop.c (doloop_modify): Likewise.
19121 * optabs.c (expand_binop): Likewise.
19122 (expand_unop): Likewise.
19123 (expand_copysign_absneg): Likewise.
19124 (prepare_cmp_insn): Likewise.
19125 (maybe_legitimize_operand): Likewise.
19126 * recog.c (const_scalar_int_operand): Likewise.
19127 * rtlanal.c (get_address_mode): Likewise.
19128 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19129 (simplify_cond_clz_ctz): Likewise.
19130 * tree-nested.c (get_nl_goto_field): Likewise.
19131 * tree.c (build_vector_type_for_mode): Likewise.
19132 * var-tracking.c (use_narrower_mode): Likewise.
19134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19135 Alan Hayward <alan.hayward@arm.com>
19136 David Sherwood <david.sherwood@arm.com>
19138 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19139 * builtins.c (expand_builtin_signbit): Use it.
19140 * cfgexpand.c (expand_debug_expr): Likewise.
19141 * dojump.c (do_jump): Likewise.
19142 (do_compare_and_jump): Likewise.
19143 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19144 * expmed.c (make_tree): Likewise.
19145 * expr.c (expand_expr_real_2): Likewise.
19146 (expand_expr_real_1): Likewise.
19147 (try_casesi): Likewise.
19148 * fold-const-call.c (fold_const_call_ss): Likewise.
19149 * fold-const.c (unextend): Likewise.
19150 (extract_muldiv_1): Likewise.
19151 (fold_single_bit_test): Likewise.
19152 (native_encode_int): Likewise.
19153 (native_encode_string): Likewise.
19154 (native_interpret_int): Likewise.
19155 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19156 * internal-fn.c (expand_addsub_overflow): Likewise.
19157 (expand_neg_overflow): Likewise.
19158 (expand_mul_overflow): Likewise.
19159 (expand_arith_overflow): Likewise.
19160 * match.pd: Likewise.
19161 * stor-layout.c (layout_type): Likewise.
19162 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19163 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19164 * tree-ssanames.c (get_range_info): Likewise.
19165 * tree-switch-conversion.c (array_value_type) Likewise.
19166 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19167 (vect_recog_divmod_pattern): Likewise.
19168 (vect_recog_mixed_size_cond_pattern): Likewise.
19169 * tree-vrp.c (extract_range_basic): Likewise.
19170 (simplify_float_conversion_using_ranges): Likewise.
19171 * tree.c (int_fits_type_p): Likewise.
19172 * ubsan.c (instrument_bool_enum_load): Likewise.
19173 * varasm.c (mergeable_string_section): Likewise.
19174 (narrowing_initializer_constant_valid_p): Likewise.
19175 (output_constant): Likewise.
19177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19178 Alan Hayward <alan.hayward@arm.com>
19179 David Sherwood <david.sherwood@arm.com>
19181 * machmode.h (NARROWEST_INT_MODE): New macro.
19182 * expr.c (alignment_for_piecewise_move): Use it instead of
19183 GET_CLASS_NARROWEST_MODE (MODE_INT).
19184 (push_block): Likewise.
19185 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19187 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19190 Alan Hayward <alan.hayward@arm.com>
19191 David Sherwood <david.sherwood@arm.com>
19193 * postreload.c (move2add_valid_value_p): Change the type of the
19194 mode parameter to scalar_int_mode.
19195 (move2add_use_add2_insn): Add a mode parameter and use it instead
19197 (move2add_use_add3_insn): Likewise.
19198 (reload_cse_move2add): Update accordingly.
19200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19201 Alan Hayward <alan.hayward@arm.com>
19202 David Sherwood <david.sherwood@arm.com>
19204 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19206 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19207 * optabs.c (expand_unop): Likewise.
19209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19210 Alan Hayward <alan.hayward@arm.com>
19211 David Sherwood <david.sherwood@arm.com>
19213 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19214 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19215 (popcount_loc_descriptor): Likewise.
19216 (bswap_loc_descriptor): Likewise.
19217 (rotate_loc_descriptor): Likewise.
19218 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19219 calling the functions above.
19221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19222 Alan Hayward <alan.hayward@arm.com>
19223 David Sherwood <david.sherwood@arm.com>
19225 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19227 (try_combine): Likewise.
19228 (simplify_if_then_else): Likewise.
19229 * cse.c (cse_insn): Likewise.
19230 * dwarf2out.c (mem_loc_descriptor): Likewise.
19231 * emit-rtl.c (gen_lowpart_common): Likewise.
19232 * simplify-rtx.c (simplify_truncation): Likewise.
19233 (simplify_binary_operation_1): Likewise.
19234 (simplify_const_relational_operation): Likewise.
19235 (simplify_ternary_operation): Likewise.
19236 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19239 Alan Hayward <alan.hayward@arm.com>
19240 David Sherwood <david.sherwood@arm.com>
19242 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19243 * reload.c (push_reload): Likewise.
19244 (find_reloads): Likewise.
19246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19247 Alan Hayward <alan.hayward@arm.com>
19248 David Sherwood <david.sherwood@arm.com>
19250 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19251 (make_compound_operation_int): Likewise.
19252 (change_zero_ext): Likewise.
19253 * expr.c (convert_move): Likewise.
19254 (convert_modes): Likewise.
19255 * fwprop.c (forward_propagate_subreg): Likewise.
19256 * loop-iv.c (get_biv_step_1): Likewise.
19257 * optabs.c (widen_operand): Likewise.
19258 * postreload.c (move2add_valid_value_p): Likewise.
19259 * recog.c (simplify_while_replacing): Likewise.
19260 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19261 (simplify_binary_operation_1): Likewise. Remove redundant
19262 mode equality check.
19264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19265 Alan Hayward <alan.hayward@arm.com>
19266 David Sherwood <david.sherwood@arm.com>
19268 * combine.c (combine_simplify_rtx): Add checks for
19269 is_a <scalar_int_mode>.
19270 (simplify_if_then_else): Likewise.
19271 (make_field_assignment): Likewise.
19272 (simplify_comparison): Likewise.
19273 * ifcvt.c (noce_try_bitop): Likewise.
19274 * loop-invariant.c (canonicalize_address_mult): Likewise.
19275 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19278 Alan Hayward <alan.hayward@arm.com>
19279 David Sherwood <david.sherwood@arm.com>
19281 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19282 is_a <scalar_int_mode> instead of != BLKmode.
19284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19285 Alan Hayward <alan.hayward@arm.com>
19286 David Sherwood <david.sherwood@arm.com>
19288 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19289 instead of != VOIDmode.
19290 * combine.c (if_then_else_cond): Likewise.
19291 (change_zero_ext): Likewise.
19292 * dwarf2out.c (mem_loc_descriptor): Likewise.
19293 (loc_descriptor): Likewise.
19294 * rtlanal.c (canonicalize_condition): Likewise.
19295 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19298 Alan Hayward <alan.hayward@arm.com>
19299 David Sherwood <david.sherwood@arm.com>
19301 * simplify-rtx.c (simplify_binary_operation_1): Use
19302 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19305 Alan Hayward <alan.hayward@arm.com>
19306 David Sherwood <david.sherwood@arm.com>
19308 * wide-int.h (int_traits<unsigned char>) New class.
19309 (int_traits<unsigned short>) Likewise.
19310 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19311 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19313 * combine.c (set_nonzero_bits_and_sign_copies): Use
19314 is_a <scalar_int_mode>.
19315 (find_split_point): Likewise.
19316 (combine_simplify_rtx): Likewise.
19317 (simplify_logical): Likewise.
19318 (expand_compound_operation): Likewise.
19319 (expand_field_assignment): Likewise.
19320 (make_compound_operation): Likewise.
19321 (extended_count): Likewise.
19322 (change_zero_ext): Likewise.
19323 (simplify_comparison): Likewise.
19324 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19325 (ucompare_loc_descriptor): Likewise.
19326 (minmax_loc_descriptor): Likewise.
19327 (mem_loc_descriptor): Likewise.
19328 (loc_descriptor): Likewise.
19329 * expmed.c (init_expmed_one_mode): Likewise.
19330 * lra-constraints.c (lra_constraint_offset): Likewise.
19331 * optabs.c (prepare_libcall_arg): Likewise.
19332 * postreload.c (move2add_note_store): Likewise.
19333 * reload.c (operands_match_p): Likewise.
19334 * rtl.h (load_extend_op): Likewise.
19335 * rtlhooks.c (gen_lowpart_general): Likewise.
19336 * simplify-rtx.c (simplify_truncation): Likewise.
19337 (simplify_unary_operation_1): Likewise.
19338 (simplify_binary_operation_1): Likewise.
19339 (simplify_const_binary_operation): Likewise.
19340 (simplify_const_relational_operation): Likewise.
19341 (simplify_subreg): Likewise.
19342 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19343 * var-tracking.c (adjust_mems): Likewise.
19344 (prepare_call_arguments): Likewise.
19346 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19347 Alan Hayward <alan.hayward@arm.com>
19348 David Sherwood <david.sherwood@arm.com>
19350 * machmode.h (is_int_mode): New fuction.
19351 * combine.c (find_split_point): Use it.
19352 (combine_simplify_rtx): Likewise.
19353 (simplify_if_then_else): Likewise.
19354 (simplify_set): Likewise.
19355 (simplify_shift_const_1): Likewise.
19356 (simplify_comparison): Likewise.
19357 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19358 * cse.c (notreg_cost): Likewise.
19359 (cse_insn): Likewise.
19360 * cselib.c (cselib_lookup_1): Likewise.
19361 * dojump.c (do_jump_1): Likewise.
19362 (do_compare_rtx_and_jump): Likewise.
19363 * dse.c (get_call_args): Likewise.
19364 * dwarf2out.c (rtl_for_decl_init): Likewise.
19365 (native_encode_initializer): Likewise.
19366 * expmed.c (emit_store_flag_1): Likewise.
19367 (emit_store_flag): Likewise.
19368 * expr.c (convert_modes): Likewise.
19369 (store_field): Likewise.
19370 (expand_expr_real_1): Likewise.
19371 * fold-const.c (fold_read_from_constant_string): Likewise.
19372 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19373 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19374 * optabs.c (expand_binop): Likewise.
19375 (expand_unop): Likewise.
19376 (expand_abs_nojump): Likewise.
19377 (expand_one_cmpl_abs_nojump): Likewise.
19378 * simplify-rtx.c (mode_signbit_p): Likewise.
19379 (val_signbit_p): Likewise.
19380 (val_signbit_known_set_p): Likewise.
19381 (val_signbit_known_clear_p): Likewise.
19382 (simplify_relational_operation_1): Likewise.
19383 * tree.c (vector_type_mode): Likewise.
19385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19386 Alan Hayward <alan.hayward@arm.com>
19387 David Sherwood <david.sherwood@arm.com>
19389 * machmode.h (smallest_mode_for_size): Fix formatting.
19390 (smallest_int_mode_for_size): New function.
19391 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19392 instead of smallest_mode_for_size.
19393 * combine.c (make_extraction): Likewise.
19394 * config/arc/arc.c (arc_expand_movmem): Likewise.
19395 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19396 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19397 * config/s390/s390.c (s390_expand_insv): Likewise.
19398 * config/sparc/sparc.c (assign_int_registers): Likewise.
19399 * config/spu/spu.c (spu_function_value): Likewise.
19400 (spu_function_arg): Likewise.
19401 * coverage.c (get_gcov_type): Likewise.
19402 (get_gcov_unsigned_t): Likewise.
19403 * dse.c (find_shift_sequence): Likewise.
19404 * expmed.c (store_bit_field_1): Likewise.
19405 * expr.c (convert_move): Likewise.
19406 (store_field): Likewise.
19407 * internal-fn.c (expand_arith_overflow): Likewise.
19408 * optabs-query.c (get_best_extraction_insn): Likewise.
19409 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19410 * stor-layout.c (layout_type): Likewise.
19411 (initialize_sizetypes): Likewise.
19412 * targhooks.c (default_get_mask_mode): Likewise.
19413 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19416 Alan Hayward <alan.hayward@arm.com>
19417 David Sherwood <david.sherwood@arm.com>
19419 * machmode.h (opt_mode::else_blk): New function.
19420 (int_mode_for_mode): Declare.
19421 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19422 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19424 * cfgexpand.c (expand_debug_expr): Likewise.
19425 * combine.c (gen_lowpart_or_truncate): Likewise.
19426 (gen_lowpart_for_combine): Likewise.
19427 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19428 * config/avr/avr.c (avr_to_int_mode): Likewise.
19429 (avr_out_plus_1): Likewise.
19430 (avr_out_plus): Likewise.
19431 (avr_out_round): Likewise.
19432 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19433 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19434 (s390_expand_vcond): Likewise.
19435 * config/spu/spu.c (spu_split_immediate): Likewise.
19436 (spu_expand_mov): Likewise.
19437 * dse.c (get_stored_val): Likewise.
19438 * expmed.c (store_bit_field_1): Likewise.
19439 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19441 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19442 (extract_low_bits): Likewise.
19443 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19444 handling rather than repeating the check.
19445 (emit_group_store): Likewise.
19446 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19447 * optabs.c (expand_absneg_bit): Likewise.
19448 (expand_copysign_absneg): Likewise.
19449 (expand_copysign_bit): Likewise.
19450 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19451 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19452 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19453 * var-tracking.c (prepare_call_arguments): Likewise.
19454 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19455 int_mode_for_mode instead of mode_for_size.
19456 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19459 Alan Hayward <alan.hayward@arm.com>
19460 David Sherwood <david.sherwood@arm.com>
19462 * machmode.h (int_mode_for_size): New function.
19463 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19464 instead of mode_for_size.
19465 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19467 * combine.c (expand_field_assignment): Use int_mode_for_size
19468 instead of mode_for_size.
19469 (make_extraction): Likewise.
19470 (simplify_shift_const_1): Likewise.
19471 (simplify_comparison): Likewise.
19472 * dojump.c (do_jump): Likewise.
19473 * dwarf2out.c (mem_loc_descriptor): Likewise.
19474 * emit-rtl.c (init_derived_machine_modes): Likewise.
19475 * expmed.c (flip_storage_order): Likewise.
19476 (convert_extracted_bit_field): Likewise.
19477 * expr.c (copy_blkmode_from_reg): Likewise.
19478 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19479 * internal-fn.c (expand_mul_overflow): Likewise.
19480 * lower-subreg.c (simple_move): Likewise.
19481 * optabs-libfuncs.c (init_optabs): Likewise.
19482 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19483 * tree.c (vector_type_mode): Likewise.
19484 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19485 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19486 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19487 * tree-vect-stmts.c (vectorizable_load): Likewise.
19488 (vectorizable_store): Likewise.
19490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19491 Alan Hayward <alan.hayward@arm.com>
19492 David Sherwood <david.sherwood@arm.com>
19494 * coretypes.h (pod_mode): New type.
19495 (scalar_int_mode_pod): New typedef.
19496 * machmode.h (pod_mode): New class.
19497 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19498 * genmodes.c (emit_mode_int_n): Update accordingly.
19499 * lower-subreg.h (target_lower_subreg): Change type to
19500 scalar_int_mode_pod.
19501 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19502 scalar_int_mode_pod.
19504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19505 Alan Hayward <alan.hayward@arm.com>
19506 David Sherwood <david.sherwood@arm.com>
19508 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19509 machine_mode to scalar_int_mode.
19510 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19511 (rs6000_option_override_internal): Remove cast to int.
19512 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19513 machine_mode to scalar_int_mode.
19514 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19515 (rs6000_option_override_internal): Remove cast to int.
19516 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19517 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19519 * config/s390/s390.c (s390_expand_builtin): Likewise.
19520 * coretypes.h (scalar_int_mode): New type.
19521 (opt_scalar_int_mode): New typedef.
19522 * machmode.h (scalar_int_mode): New class.
19523 (scalar_int_mode::includes_p): New function.
19524 (byte_mode): Change type to scalar_int_mode.
19525 (word_mode): Likewise.
19526 (ptr_mode): Likewise.
19527 * emit-rtl.c (byte_mode): Likewise.
19528 (word_mode): Likewise.
19529 (ptr_mode): Likewise.
19530 (init_derived_machine_modes): Update accordingly.
19531 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19532 and MODE_PARTIAL_INT.
19533 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19534 opt_scalar_int_mode.
19536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19537 Alan Hayward <alan.hayward@arm.com>
19538 David Sherwood <david.sherwood@arm.com>
19540 * target.def (libgcc_floating_mode_supported_p): Take a
19542 * doc/tm.texi: Regenerate.
19543 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19545 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19546 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19550 Alan Hayward <alan.hayward@arm.com>
19551 David Sherwood <david.sherwood@arm.com>
19553 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19554 * doc/tm.texi: Regenerate.
19555 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19556 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19557 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19558 * targhooks.h (default_floatn_mode): Likewise.
19559 * targhooks.c (default_floatn_mode): Likewise.
19560 * tree.c (build_common_tree_nodes): Update accordingly.
19562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19563 Alan Hayward <alan.hayward@arm.com>
19564 David Sherwood <david.sherwood@arm.com>
19566 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19567 (mode_iterator::iterate_p): Likewise.
19568 (mode_iterator::get_wider): Likewise.
19569 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19572 Alan Hayward <alan.hayward@arm.com>
19573 David Sherwood <david.sherwood@arm.com>
19575 * coretypes.h (opt_scalar_float_mode): New typedef.
19576 * machmode.h (float_mode_for_size): New function.
19577 * emit-rtl.c (double_mode): Delete.
19578 (init_emit_once): Use float_mode_for_size.
19579 * stor-layout.c (layout_type): Likewise.
19580 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19583 Alan Hayward <alan.hayward@arm.com>
19584 David Sherwood <david.sherwood@arm.com>
19586 * output.h (assemble_real): Take a scalar_float_mode.
19587 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19588 * config/arm/arm.md (consttable_4): Likewise.
19589 (consttable_8): Likewise.
19590 (consttable_16): Likewise.
19591 * config/mips/mips.md (consttable_float): Likewise.
19592 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19593 * varasm.c (assemble_real): Take a scalar_float_mode.
19594 (output_constant_pool_2): Update accordingly.
19595 (output_constant): Likewise.
19597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19598 Alan Hayward <alan.hayward@arm.com>
19599 David Sherwood <david.sherwood@arm.com>
19601 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19602 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19603 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19604 (native_encode_real): Likewise.
19605 (native_interpret_real): Likewise.
19606 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19607 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19610 Alan Hayward <alan.hayward@arm.com>
19611 David Sherwood <david.sherwood@arm.com>
19613 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19614 <scalar_float_mode>. Simplify.
19615 (gen_extend_conv_libfunc): Likewise.
19617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19618 Alan Hayward <alan.hayward@arm.com>
19619 David Sherwood <david.sherwood@arm.com>
19621 * coretypes.h (scalar_float_mode): New type.
19622 * machmode.h (mode_traits::from_int): Use machine_mode if
19623 USE_ENUM_MODES is defined.
19624 (is_a): New function.
19626 (dyn_cast): Likewise.
19627 (scalar_float_mode): New class.
19628 (scalar_float_mode::includes_p): New function.
19629 (is_float_mode): Likewise.
19630 * gdbhooks.py (MachineModePrinter): New class.
19631 (build_pretty_printer): Use it for scalar_float_mode.
19632 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19633 (format_helper::format_helper): Turn into a template.
19634 * genmodes.c (get_mode_class): New function.
19635 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19636 or machine_mode if none.
19637 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19638 as_a <scalar_float_mode>.
19639 * dwarf2out.c (mem_loc_descriptor): Likewise.
19640 (insert_float): Likewise.
19641 (add_const_value_attribute): Likewise.
19642 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19643 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19644 (expand_unop): Update accordingly.
19645 (expand_abs_nojump): Likewise.
19646 (expand_copysign_absneg): Take a scalar_float_mode.
19647 (expand_copysign_bit): Likewise.
19648 (expand_copysign): Update accordingly.
19650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19651 Alan Hayward <alan.hayward@arm.com>
19652 David Sherwood <david.sherwood@arm.com>
19654 * coretypes.h (opt_mode): New class.
19655 * machmode.h (opt_mode): Likewise.
19656 (opt_mode::else_void): New function.
19657 (opt_mode::require): Likewise.
19658 (opt_mode::exists): Likewise.
19659 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19660 (GET_MODE_2XWIDER_MODE): Likewise.
19661 (mode_iterator::get_wider): Update accordingly.
19662 (mode_iterator::get_2xwider): Likewise.
19663 (mode_iterator::get_known_wider): Likewise, turning into a template.
19664 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19665 forcing a wider mode to exist.
19666 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19667 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19668 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19669 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19670 * lower-subreg.c (init_lower_subreg): Likewise.
19671 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19672 on the final iteration.
19673 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19674 a wider mode exists before asking for a move pattern.
19675 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19676 forcing a wider mode to exist.
19677 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19678 returning false if no such mode exists.
19679 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19680 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19681 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19682 Avoid checking for a MODE_INT if we already know the mode is not a
19684 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19685 forcing a wider mode to exist.
19686 (expmed_mult_highpart_optab): Likewise.
19687 (expmed_mult_highpart): Likewise.
19688 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19690 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19691 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19692 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19693 * internal-fn.c (expand_mul_overflow): Update use of
19694 GET_MODE_2XWIDER_MODE.
19695 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19696 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19697 GET_MODE_WIDER_MODE.
19698 (convert_plusminus_to_widen): Likewise.
19699 * tree-switch-conversion.c (array_value_type): Likewise.
19700 * var-tracking.c (emit_note_insn_var_location): Likewise.
19701 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19702 Return false inside rather than outside the loop if no wider mode
19704 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19705 and GET_MODE_2XWIDER_MODE
19706 (can_compare_p): Use else_void.
19707 * gdbhooks.py (OptMachineModePrinter): New class.
19708 (build_pretty_printer): Use it for opt_mode.
19710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19711 Alan Hayward <alan.hayward@arm.com>
19712 David Sherwood <david.sherwood@arm.com>
19714 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19715 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19718 Alan Hayward <alan.hayward@arm.com>
19719 David Sherwood <david.sherwood@arm.com>
19721 * machmode.h (mode_traits): New structure.
19722 (get_narrowest_mode): New function.
19723 (mode_iterator::start): Likewise.
19724 (mode_iterator::iterate_p): Likewise.
19725 (mode_iterator::get_wider): Likewise.
19726 (mode_iterator::get_known_wider): Likewise.
19727 (mode_iterator::get_2xwider): Likewise.
19728 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19729 (FOR_EACH_MODE): Likewise.
19730 (FOR_EACH_MODE_FROM): Likewise.
19731 (FOR_EACH_MODE_UNTIL): Likewise.
19732 (FOR_EACH_WIDER_MODE): Likewise.
19733 (FOR_EACH_2XWIDER_MODE): Likewise.
19734 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19735 * combine.c (simplify_comparison): Likewise
19736 * config/i386/i386.c (type_natural_mode): Likewise.
19737 * cse.c (cse_insn): Likewise.
19738 * dse.c (find_shift_sequence): Likewise.
19739 * emit-rtl.c (init_derived_machine_modes): Likewise.
19740 (init_emit_once): Likewise.
19741 * explow.c (hard_function_value): Likewise.
19742 * expmed.c (extract_fixed_bit_field_1): Likewise.
19743 (extract_bit_field_1): Likewise.
19744 (expand_divmod): Likewise.
19745 (emit_store_flag_1): Likewise.
19746 * expr.c (init_expr_target): Likewise.
19747 (convert_move): Likewise.
19748 (alignment_for_piecewise_move): Likewise.
19749 (widest_int_mode_for_size): Likewise.
19750 (emit_block_move_via_movmem): Likewise.
19751 (copy_blkmode_to_reg): Likewise.
19752 (set_storage_via_setmem): Likewise.
19753 (compress_float_constant): Likewise.
19754 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19755 * optabs-query.c (get_best_extraction_insn): Likewise.
19756 * optabs.c (expand_binop): Likewise.
19757 (expand_twoval_unop): Likewise.
19758 (expand_twoval_binop): Likewise.
19759 (widen_leading): Likewise.
19760 (widen_bswap): Likewise.
19761 (expand_parity): Likewise.
19762 (expand_unop): Likewise.
19763 (prepare_cmp_insn): Likewise.
19764 (prepare_float_lib_cmp): Likewise.
19765 (expand_float): Likewise.
19766 (expand_fix): Likewise.
19767 (expand_sfix_optab): Likewise.
19768 * postreload.c (move2add_use_add2_insn): Likewise.
19769 * reg-stack.c (reg_to_stack): Likewise.
19770 * reginfo.c (choose_hard_reg_mode): Likewise.
19771 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19772 * stor-layout.c (mode_for_size): Likewise.
19773 (smallest_mode_for_size): Likewise.
19774 (mode_for_vector): Likewise.
19775 (finish_bitfield_representative): Likewise.
19776 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19777 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19778 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19779 * var-tracking.c (prepare_call_arguments): Likewise.
19781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19782 Alan Hayward <alan.hayward@arm.com>
19783 David Sherwood <david.sherwood@arm.com>
19785 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19786 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19787 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19788 * machmode.h (mode_size): Move earlier in file.
19789 (mode_precision): Likewise.
19790 (mode_inner): Likewise.
19791 (mode_nunits): Likewise.
19792 (mode_unit_size): Likewise.
19793 (unit_unit_precision): Likewise.
19794 (mode_wider): Likewise.
19795 (mode_2xwider): Likewise.
19796 (machine_mode): New class.
19797 (mode_to_bytes): New function.
19798 (mode_to_bits): Likewise.
19799 (mode_to_precision): Likewise.
19800 (mode_to_inner): Likewise.
19801 (mode_to_unit_size): Likewise.
19802 (mode_to_unit_precision): Likewise.
19803 (mode_to_nunits): Likewise.
19804 (GET_MODE_SIZE): Use mode_to_bytes.
19805 (GET_MODE_BITSIZE): Use mode_to_bits.
19806 (GET_MODE_PRECISION): Use mode_to_precision.
19807 (GET_MODE_INNER): Use mode_to_inner.
19808 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19809 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19810 (GET_MODE_NUNITS): Use mode_to_nunits.
19811 * system.h (ALWAYS_INLINE): New macro.
19812 * config/powerpcspe/powerpcspe-c.c
19813 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19814 int for arg1_mode and arg2_mode.
19816 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19817 Alan Hayward <alan.hayward@arm.com>
19818 David Sherwood <david.sherwood@arm.com>
19820 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19821 Prefix mode names with E_ in case statements.
19822 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19823 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19824 (aarch64_split_simd_move): Likewise.
19825 (aarch64_gen_storewb_pair): Likewise.
19826 (aarch64_gen_loadwb_pair): Likewise.
19827 (aarch64_gen_store_pair): Likewise.
19828 (aarch64_gen_load_pair): Likewise.
19829 (aarch64_get_condition_code_1): Likewise.
19830 (aarch64_constant_pool_reload_icode): Likewise.
19831 (get_rsqrte_type): Likewise.
19832 (get_rsqrts_type): Likewise.
19833 (get_recpe_type): Likewise.
19834 (get_recps_type): Likewise.
19835 (aarch64_gimplify_va_arg_expr): Likewise.
19836 (aarch64_simd_container_mode): Likewise.
19837 (aarch64_emit_load_exclusive): Likewise.
19838 (aarch64_emit_store_exclusive): Likewise.
19839 (aarch64_expand_compare_and_swap): Likewise.
19840 (aarch64_gen_atomic_cas): Likewise.
19841 (aarch64_emit_bic): Likewise.
19842 (aarch64_emit_atomic_swap): Likewise.
19843 (aarch64_emit_atomic_load_op): Likewise.
19844 (aarch64_evpc_trn): Likewise.
19845 (aarch64_evpc_uzp): Likewise.
19846 (aarch64_evpc_zip): Likewise.
19847 (aarch64_evpc_ext): Likewise.
19848 (aarch64_evpc_rev): Likewise.
19849 (aarch64_evpc_dup): Likewise.
19850 (aarch64_gen_ccmp_first): Likewise.
19851 (aarch64_gen_ccmp_next): Likewise.
19852 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19853 (alpha_emit_xfloating_libcall): Likewise.
19854 (emit_insxl): Likewise.
19855 (alpha_arg_type): Likewise.
19856 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19857 (arc_preferred_simd_mode): Likewise.
19858 (arc_secondary_reload): Likewise.
19859 (get_arc_condition_code): Likewise.
19860 (arc_print_operand): Likewise.
19861 (arc_legitimate_constant_p): Likewise.
19862 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19863 * config/arc/arc.md (casesi_load): Likewise.
19864 (casesi_compact_jump): Likewise.
19865 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19866 (cc_use_register): Likewise.
19867 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19868 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19869 (arm_init_iwmmxt_builtins): Likewise.
19870 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19871 (neon_expand_vector_init): Likewise.
19872 (arm_attr_length_move_neon): Likewise.
19873 (maybe_get_arm_condition_code): Likewise.
19874 (arm_emit_vector_const): Likewise.
19875 (arm_preferred_simd_mode): Likewise.
19876 (arm_output_iwmmxt_tinsr): Likewise.
19877 (thumb1_output_casesi): Likewise.
19878 (thumb2_output_casesi): Likewise.
19879 (arm_emit_load_exclusive): Likewise.
19880 (arm_emit_store_exclusive): Likewise.
19881 (arm_expand_compare_and_swap): Likewise.
19882 (arm_evpc_neon_vuzp): Likewise.
19883 (arm_evpc_neon_vzip): Likewise.
19884 (arm_evpc_neon_vrev): Likewise.
19885 (arm_evpc_neon_vtrn): Likewise.
19886 (arm_evpc_neon_vext): Likewise.
19887 (arm_validize_comparison): Likewise.
19888 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19889 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19890 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19891 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19892 (c6x_preferred_simd_mode): Likewise.
19893 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19894 (epiphany_rtx_costs): Likewise.
19895 * config/epiphany/predicates.md (proper_comparison_operator):
19897 * config/frv/frv.c (condexec_memory_operand): Likewise.
19898 (frv_emit_move): Likewise.
19899 (output_move_single): Likewise.
19900 (output_condmove_single): Likewise.
19901 (frv_hard_regno_mode_ok): Likewise.
19902 (frv_matching_accg_mode): Likewise.
19903 * config/h8300/h8300.c (split_adds_subs): Likewise.
19904 (h8300_rtx_costs): Likewise.
19905 (h8300_print_operand): Likewise.
19906 (compute_mov_length): Likewise.
19907 (output_logical_op): Likewise.
19908 (compute_logical_op_length): Likewise.
19909 (compute_logical_op_cc): Likewise.
19910 (h8300_shift_needs_scratch_p): Likewise.
19911 (output_a_shift): Likewise.
19912 (compute_a_shift_length): Likewise.
19913 (compute_a_shift_cc): Likewise.
19914 (expand_a_rotate): Likewise.
19915 (output_a_rotate): Likewise.
19916 * config/i386/i386.c (classify_argument): Likewise.
19917 (function_arg_advance_32): Likewise.
19918 (function_arg_32): Likewise.
19919 (function_arg_64): Likewise.
19920 (function_value_64): Likewise.
19921 (ix86_gimplify_va_arg): Likewise.
19922 (ix86_legitimate_constant_p): Likewise.
19923 (put_condition_code): Likewise.
19924 (split_double_mode): Likewise.
19925 (ix86_avx256_split_vector_move_misalign): Likewise.
19926 (ix86_expand_vector_logical_operator): Likewise.
19927 (ix86_split_idivmod): Likewise.
19928 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19929 (ix86_build_const_vector): Likewise.
19930 (ix86_build_signbit_mask): Likewise.
19931 (ix86_match_ccmode): Likewise.
19932 (ix86_cc_modes_compatible): Likewise.
19933 (ix86_expand_branch): Likewise.
19934 (ix86_expand_sse_cmp): Likewise.
19935 (ix86_expand_sse_movcc): Likewise.
19936 (ix86_expand_int_sse_cmp): Likewise.
19937 (ix86_expand_vec_perm_vpermi2): Likewise.
19938 (ix86_expand_vec_perm): Likewise.
19939 (ix86_expand_sse_unpack): Likewise.
19940 (ix86_expand_int_addcc): Likewise.
19941 (ix86_split_to_parts): Likewise.
19942 (ix86_vectorize_builtin_gather): Likewise.
19943 (ix86_vectorize_builtin_scatter): Likewise.
19944 (avx_vpermilp_parallel): Likewise.
19945 (inline_memory_move_cost): Likewise.
19946 (ix86_tieable_integer_mode_p): Likewise.
19947 (x86_maybe_negate_const_int): Likewise.
19948 (ix86_expand_vector_init_duplicate): Likewise.
19949 (ix86_expand_vector_init_one_nonzero): Likewise.
19950 (ix86_expand_vector_init_one_var): Likewise.
19951 (ix86_expand_vector_init_concat): Likewise.
19952 (ix86_expand_vector_init_interleave): Likewise.
19953 (ix86_expand_vector_init_general): Likewise.
19954 (ix86_expand_vector_set): Likewise.
19955 (ix86_expand_vector_extract): Likewise.
19956 (emit_reduc_half): Likewise.
19957 (ix86_emit_i387_round): Likewise.
19958 (ix86_mangle_type): Likewise.
19959 (ix86_expand_round_sse4): Likewise.
19960 (expand_vec_perm_blend): Likewise.
19961 (canonicalize_vector_int_perm): Likewise.
19962 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19963 (expand_vec_perm_1): Likewise.
19964 (expand_vec_perm_interleave3): Likewise.
19965 (expand_vec_perm_even_odd_pack): Likewise.
19966 (expand_vec_perm_even_odd_1): Likewise.
19967 (expand_vec_perm_broadcast_1): Likewise.
19968 (ix86_vectorize_vec_perm_const_ok): Likewise.
19969 (ix86_expand_vecop_qihi): Likewise.
19970 (ix86_expand_mul_widen_hilo): Likewise.
19971 (ix86_expand_sse2_abs): Likewise.
19972 (ix86_expand_pextr): Likewise.
19973 (ix86_expand_pinsr): Likewise.
19974 (ix86_preferred_simd_mode): Likewise.
19975 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19976 * config/i386/sse.md (*andnot<mode>3): Likewise.
19977 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19978 (*<code><mode>3): Likewise.
19979 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19980 (ia64_expand_atomic_op): Likewise.
19981 (ia64_arg_type): Likewise.
19982 (ia64_mode_to_int): Likewise.
19983 (ia64_scalar_mode_supported_p): Likewise.
19984 (ia64_vector_mode_supported_p): Likewise.
19985 (expand_vec_perm_broadcast): Likewise.
19986 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19987 (iq2000_function_arg_advance): Likewise.
19988 (iq2000_function_arg): Likewise.
19989 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19990 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19991 (m68k_libcall_value): Likewise.
19992 (m68k_function_value): Likewise.
19993 (sched_attr_op_type): Likewise.
19994 * config/mcore/mcore.c (mcore_output_move): Likewise.
19995 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19997 (microblaze_function_arg): Likewise.
19998 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19999 (mips_print_operand): Likewise.
20000 (mips_mode_ok_for_mov_fmt_p): Likewise.
20001 (mips_vector_mode_supported_p): Likewise.
20002 (mips_preferred_simd_mode): Likewise.
20003 (mips_expand_vpc_loongson_even_odd): Likewise.
20004 (mips_expand_vec_unpack): Likewise.
20005 (mips_expand_vi_broadcast): Likewise.
20006 (mips_expand_vector_init): Likewise.
20007 (mips_expand_vec_reduc): Likewise.
20008 (mips_expand_msa_cmp): Likewise.
20009 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20010 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20011 (cc_flags_for_mode): Likewise.
20012 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20013 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20014 (nds32_output_casesi_pc_relative): Likewise.
20015 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20016 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20017 (nvptx_gen_unpack): Likewise.
20018 (nvptx_gen_pack): Likewise.
20019 (nvptx_gen_shuffle): Likewise.
20020 (nvptx_gen_wcast): Likewise.
20021 (nvptx_preferred_simd_mode): Likewise.
20022 * config/pa/pa.c (pa_secondary_reload): Likewise.
20023 * config/pa/predicates.md (base14_operand): Likewise.
20024 * config/powerpcspe/powerpcspe-c.c
20025 (altivec_resolve_overloaded_builtin): Likewise.
20026 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20028 (rs6000_preferred_simd_mode): Likewise.
20029 (output_vec_const_move): Likewise.
20030 (rs6000_expand_vector_extract): Likewise.
20031 (rs6000_split_vec_extract_var): Likewise.
20032 (reg_offset_addressing_ok_p): Likewise.
20033 (rs6000_legitimate_offset_address_p): Likewise.
20034 (rs6000_legitimize_address): Likewise.
20035 (rs6000_emit_set_const): Likewise.
20036 (rs6000_const_vec): Likewise.
20037 (rs6000_emit_move): Likewise.
20038 (spe_build_register_parallel): Likewise.
20039 (rs6000_darwin64_record_arg_recurse): Likewise.
20040 (swap_selector_for_mode): Likewise.
20041 (spe_init_builtins): Likewise.
20042 (paired_init_builtins): Likewise.
20043 (altivec_init_builtins): Likewise.
20044 (do_load_for_compare): Likewise.
20045 (rs6000_generate_compare): Likewise.
20046 (rs6000_expand_float128_convert): Likewise.
20047 (emit_load_locked): Likewise.
20048 (emit_store_conditional): Likewise.
20049 (rs6000_output_function_epilogue): Likewise.
20050 (rs6000_handle_altivec_attribute): Likewise.
20051 (rs6000_function_value): Likewise.
20052 (emit_fusion_gpr_load): Likewise.
20053 (emit_fusion_p9_load): Likewise.
20054 (emit_fusion_p9_store): Likewise.
20055 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20056 (fusion_gpr_mem_load): Likewise.
20057 (fusion_addis_mem_combo_load): Likewise.
20058 (fusion_addis_mem_combo_store): Likewise.
20059 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20060 (fusion_gpr_mem_load): Likewise.
20061 (fusion_addis_mem_combo_load): Likewise.
20062 (fusion_addis_mem_combo_store): Likewise.
20063 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20065 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20066 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20067 (rs6000_preferred_simd_mode): Likewise.
20068 (output_vec_const_move): Likewise.
20069 (rs6000_expand_vector_extract): Likewise.
20070 (rs6000_split_vec_extract_var): Likewise.
20071 (reg_offset_addressing_ok_p): Likewise.
20072 (rs6000_legitimate_offset_address_p): Likewise.
20073 (rs6000_legitimize_address): Likewise.
20074 (rs6000_emit_set_const): Likewise.
20075 (rs6000_const_vec): Likewise.
20076 (rs6000_emit_move): Likewise.
20077 (rs6000_darwin64_record_arg_recurse): Likewise.
20078 (swap_selector_for_mode): Likewise.
20079 (paired_init_builtins): Likewise.
20080 (altivec_init_builtins): Likewise.
20081 (rs6000_expand_float128_convert): Likewise.
20082 (emit_load_locked): Likewise.
20083 (emit_store_conditional): Likewise.
20084 (rs6000_output_function_epilogue): Likewise.
20085 (rs6000_handle_altivec_attribute): Likewise.
20086 (rs6000_function_value): Likewise.
20087 (emit_fusion_gpr_load): Likewise.
20088 (emit_fusion_p9_load): Likewise.
20089 (emit_fusion_p9_store): Likewise.
20090 * config/rx/rx.c (rx_gen_move_template): Likewise.
20091 (flags_from_mode): Likewise.
20092 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20093 (s390_slb_comparison): Likewise.
20094 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20095 (s390_vector_mode_supported_p): Likewise.
20096 (s390_cc_modes_compatible): Likewise.
20097 (s390_match_ccmode_set): Likewise.
20098 (s390_canonicalize_comparison): Likewise.
20099 (s390_emit_compare_and_swap): Likewise.
20100 (s390_branch_condition_mask): Likewise.
20101 (s390_rtx_costs): Likewise.
20102 (s390_secondary_reload): Likewise.
20103 (__SECONDARY_RELOAD_CASE): Likewise.
20104 (s390_expand_cs): Likewise.
20105 (s390_preferred_simd_mode): Likewise.
20106 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20107 * config/sh/sh.c (sh_print_operand): Likewise.
20108 (dump_table): Likewise.
20109 (sh_secondary_reload): Likewise.
20110 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20111 * config/sh/sh.md (casesi_worker_1): Likewise.
20112 (casesi_worker_2): Likewise.
20113 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20114 (fcc_comparison_operator): Likewise.
20115 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20116 (emit_soft_tfmode_cvt): Likewise.
20117 (sparc_preferred_simd_mode): Likewise.
20118 (output_cbranch): Likewise.
20119 (sparc_print_operand): Likewise.
20120 (sparc_expand_vec_perm_bmask): Likewise.
20121 (vector_init_bshuffle): Likewise.
20122 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20123 (spu_vector_mode_supported_p): Likewise.
20124 (spu_expand_insv): Likewise.
20125 (spu_emit_branch_or_set): Likewise.
20126 (spu_handle_vector_attribute): Likewise.
20127 (spu_builtin_splats): Likewise.
20128 (spu_builtin_extract): Likewise.
20129 (spu_builtin_promote): Likewise.
20130 (spu_expand_sign_extend): Likewise.
20131 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20132 (tilegx_simd_int): Likewise.
20133 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20134 (tilepro_simd_int): Likewise.
20135 * config/v850/v850.c (const_double_split): Likewise.
20136 (v850_print_operand): Likewise.
20137 (ep_memory_offset): Likewise.
20138 * config/vax/vax.c (vax_rtx_costs): Likewise.
20139 (vax_output_int_move): Likewise.
20140 (vax_output_int_add): Likewise.
20141 (vax_output_int_subtract): Likewise.
20142 * config/visium/predicates.md (visium_branch_operator): Likewise.
20143 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20144 (visium_print_operand_address): Likewise.
20145 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20146 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20147 (xtensa_expand_conditional_branch): Likewise.
20148 (xtensa_copy_incoming_a7): Likewise.
20149 (xtensa_output_literal): Likewise.
20150 * dfp.c (decimal_real_maxval): Likewise.
20151 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20154 Alan Hayward <alan.hayward@arm.com>
20155 David Sherwood <david.sherwood@arm.com>
20157 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20158 (mode_nunits_inline): Likewise.
20159 (mode_inner_inline): Likewise.
20160 (mode_unit_size_inline): Likewise.
20161 (mode_unit_precision_inline): Likewise.
20162 (emit_insn_modes_h): Likewise. Also emit a #define of the
20164 (emit_mode_wider): Add an E_ prefix to mode names.
20165 (emit_mode_complex): Likewise.
20166 (emit_mode_inner): Likewise.
20167 (emit_mode_adjustments): Likewise.
20168 (emit_mode_int_n): Likewise.
20169 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20170 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20171 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20172 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20173 (CRC32_BUILTIN, ENTRY): Likewise.
20174 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20175 (aarch64_pop_regs): Likewise.
20176 (aarch64_process_components): Likewise.
20177 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20178 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20179 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20180 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20181 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20182 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20184 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20185 (pcmpistr): Likewise.
20186 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20187 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20188 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20190 * config/rl78/rl78.c (mduc_regs): Likewise.
20191 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20192 (htm_expand_builtin): Likewise.
20193 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20194 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20195 * config/xtensa/xtensa.c (print_operand): Likewise.
20196 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20197 (NUM_MODE_VECTOR_INT): Likewise.
20198 * genoutput.c (null_operand): Likewise.
20199 (output_operand_data): Likewise.
20200 * genrecog.c (print_parameter_value): Likewise.
20201 * lra.c (debug_operand_data): Likewise.
20203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20205 * dwarf2out.c (scompare_loc_descriptor_wide)
20206 (scompare_loc_descriptor_narrow): New functions, split out from...
20207 (scompare_loc_descriptor): ...here.
20208 * expmed.c (emit_store_flag_int): New function, split out from...
20209 (emit_store_flag): ...here.
20211 2017-08-30 Richard Biener <rguenther@suse.de>
20213 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20214 (dwarf2out_early_finish): Move setting of AT_pubnames from
20215 early debug output to early finish.
20217 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20219 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20220 and -mdata-region to the assembler.
20222 2017-08-30 Richard Biener <rguenther@suse.de>
20224 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20226 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20227 (dwarf2out_early_global_decl): Only generate a DIE for the
20228 abstract origin if it doesn't already exist or is a declaration DIE.
20229 (resolve_addr): Do not add the linkage name twice when
20230 generating a stub DIE for the DW_TAG_GNU_call_site target.
20232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20234 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20235 Use machine_mode rather than int for arg1_mode.
20237 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20240 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20241 that the second argument of the built-in functions to unpack
20242 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20243 switch statement instead a lot of if statements.
20244 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20245 Allow 64-bit values to be in Altivec registers as well as
20246 traditional floating point registers.
20247 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20249 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20251 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20252 MAX_REGS_PER_ADDRESS == 1.
20254 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20256 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20257 * config/i386/i386.c (ix86_option_override_internal): Simplify
20258 setting of opts->x_flag_entry.
20260 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20261 Jakub Jelinek <jakub@redhat.com>
20262 Richard Biener <rguenther@suse.de>
20264 PR tree-optimization/81503
20265 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20266 folded constant fits in the target type; reorder tests for clarity.
20268 2017-08-29 Martin Liska <mliska@suse.cz>
20270 * passes.def: Include pass_lower_switch.
20271 * stmt.c (dump_case_nodes): Remove and move to
20272 tree-switch-conversion.
20273 (case_values_threshold): Likewise.
20274 (expand_switch_as_decision_tree_p): Likewise.
20275 (emit_case_decision_tree): Likewise.
20276 (expand_case): Likewise.
20277 (balance_case_nodes): Likewise.
20278 (node_has_low_bound): Likewise.
20279 (node_has_high_bound): Likewise.
20280 (node_is_bounded): Likewise.
20281 (emit_case_nodes): Likewise.
20282 (struct simple_case_node): New struct.
20283 (add_case_node): Remove.
20284 (emit_case_dispatch_table): Use vector instead of case_list.
20285 (reset_out_edges_aux): Remove.
20286 (compute_cases_per_edge): Likewise.
20287 (expand_case): Build list of simple_case_node.
20288 (expand_sjlj_dispatch_table): Use it.
20289 * tree-switch-conversion.c (struct case_node): Moved from
20290 stmt.c and adjusted.
20291 (emit_case_nodes): Likewise.
20292 (node_has_low_bound): Likewise.
20293 (node_has_high_bound): Likewise.
20294 (node_is_bounded): Likewise.
20295 (case_values_threshold): Likewise.
20296 (reset_out_edges_aux): Likewise.
20297 (compute_cases_per_edge): Likewise.
20298 (add_case_node): Likewise.
20299 (dump_case_nodes): Likewise.
20300 (balance_case_nodes): Likewise.
20301 (expand_switch_as_decision_tree_p): Likewise.
20302 (emit_jump): Likewise.
20303 (emit_case_decision_tree): Likewise.
20304 (try_switch_expansion): Likewise.
20305 (do_jump_if_equal): Likewise.
20306 (emit_cmp_and_jump_insns): Likewise.
20307 (fix_phi_operands_for_edge): New function.
20308 (record_phi_operand_mapping): Likewise.
20309 (class pass_lower_switch): New pass.
20310 (pass_lower_switch::execute): New function.
20311 (make_pass_lower_switch): Likewise.
20312 (conditional_probability):
20313 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20314 * tree-pass.h: Add make_pass_lower_switch.
20316 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20319 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20322 2017-08-29 Richard Biener <rguenther@suse.de>
20324 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20325 we do not add a DW_AT_inline attribute twice.
20326 (gen_subprogram_die): Remove code setting DW_AT_inline on
20327 DECL_ABSTRACT_P nodes.
20329 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20331 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20332 calls to internal functions.
20333 (gimplify_modify_expr): Likewise.
20334 * tree-call-cdce.c (use_internal_fn): Likewise.
20335 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20336 (convert_to_divmod): Set the nothrow flag.
20337 * tree-if-conv.c (predicate_mem_writes): Likewise.
20338 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20339 (vectorizable_call): Likewise.
20340 (vectorizable_store): Likewise.
20341 (vectorizable_load): Likewise.
20342 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20343 (vect_recog_mask_conversion_pattern): Likewise.
20345 2017-08-29 Martin Liska <mliska@suse.cz>
20348 * gcc.c (driver_handle_option): Add new argument.
20349 * opts-common.c (handle_option): Pass
20350 target_option_override_hook.
20351 * opts-global.c (lang_handle_option): Add new option.
20352 (set_default_handlers): Add new argument.
20353 (decode_options): Likewise.
20354 * opts.c (target_handle_option): Likewise.
20355 (common_handle_option): Call target_option_override_hook.
20356 * opts.h (struct cl_option_handler_func): Add hook for
20357 target option override.
20358 (struct cl_option_handlers): Likewise.
20359 (set_default_handlers): Add new argument.
20360 (decode_options): Likewise.
20361 (common_handle_option): Likewise.
20362 (target_handle_option): Likewise.
20363 * toplev.c (toplev::main): Pass targetm.target_option.override
20366 2017-08-29 Richard Biener <rguenther@suse.de>
20367 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20369 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20370 life to the active subtree.
20372 2017-08-28 Jeff Law <law@redhat.com>
20374 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20375 derive_equivalences.
20376 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20377 Code moved into....
20378 (edge_info::derive_equivalences): New private member function
20380 * tree-ssa-dom.c (class edge_info): Changed from a struct
20381 to a class. Add ctor/dtor, methods and data members.
20382 (edge_info::edge_info): Renamed from allocate_edge_info.
20383 Initialize additional members.
20384 (edge_info::~edge_info): New.
20385 (free_dom_edge_info): Delete the edge info.
20386 (record_edge_info): Use new class & associated member functions.
20387 Tighten forms for testing for edge equivalences.
20388 (record_temporary_equivalences): Iterate over the simple
20389 equivalences rather than assuming there's only one per edge.
20390 (cprop_into_successor_phis): Iterate over the simple
20391 equivalences rather than assuming there's only one per edge.
20392 (optimize_stmt): Use operand_equal_p rather than pointer
20393 equality for mini-DSE code.
20395 2017-08-28 Nathan Sidwell <nathan@acm.org>
20397 * gcc.c (execute): Fold SIGPIPE handling into switch
20398 statement. Adjust internal error message.
20400 2017-08-28 Richard Biener <rguenther@suse.de>
20403 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20404 Do nothing for removed DIEs.
20406 2017-08-28 Richard Biener <rguenther@suse.de>
20408 PR tree-optimization/81977
20409 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20412 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20415 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20416 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20417 using targetm.gen_mem_thread_fence.
20419 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20422 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20423 predicate to register_operand. Reorder operands.
20424 (*btr<mode>): Ditto.
20425 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20426 (*btr<mode>_mask): Ditto.
20428 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20430 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20431 * config/rs6000/xmmintrin.h: New file.
20432 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20434 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20437 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20438 parameter and_insn and return it.
20439 (recombine_lvx_pattern): Insert a copy to ensure availability of
20440 the base register of the copied masking operation at the point of
20441 the instruction replacement.
20442 (recombine_stvx_pattern): Likewise.
20444 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20446 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20447 undocumented switches.
20448 (-mpower9-dform-vector): Likewise.
20449 (-mpower9-dform): Likewise.
20450 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20451 comments to delete references to -mpower9-dform* switches.
20452 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20453 Delete reference to -mpower9-dform* switches, test for
20454 -mpower9-vector instead.
20455 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20456 (OTHER_P9_VECTOR_MASKS): Likewise.
20457 (POWERPC_MASKS): Likewise.
20458 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20459 tests against -mpower9-dform* to -mpower9-vector. Delete code
20460 that checked for -mpower9-dform* consistancy with other options.
20461 Add test for -mpower9-misc to enable other power9 switches.
20462 (rs6000_init_hard_regno_mode_ok): Likewise.
20463 (rs6000_option_override_internal): Likewise.
20464 (rs6000_emit_prologue): Likewise.
20465 (rs6000_emit_epilogue): Likewise.
20466 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20467 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20468 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20470 (emit_fusion_p9_store): Likewise.
20471 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20472 resetting these macros if the assembler does not support ISA 3.0
20474 (TARGET_P9_DFORM_VECTOR): Likewise.
20475 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20476 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20478 2017-08-25 Alan Modra <amodra@gmail.com>
20480 PR rtl-optimization/81747
20481 * cse.c (cse_extended_basic_block): Don't attempt to record
20482 equivalences for degenerate conditional branches that branch
20483 to their fall-through.
20485 2017-08-24 Martin Sebor <msebor@redhat.com>
20487 PR middle-end/81908
20488 * gimple-fold.c (size_must_be_zero_p): New function.
20489 (gimple_fold_builtin_memory_op): Call it.
20491 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20493 * config/rs6000/mm_malloc.h: New file.
20495 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20497 PR tree-optimization/81913
20498 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20499 analysis when either IVs in condition can wrap.
20501 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20503 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20504 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20506 2017-08-24 Richard Biener <rguenther@suse.de>
20509 * targhooks.c (default_target_can_inline_p): Properly
20510 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20511 is present and always compare.
20512 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20513 infer -mfpmath=sse from TARGET_SSE_P.
20514 (ix86_can_inline_p): Properly use target_option_default_node when
20515 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20517 2017-08-24 Richard Biener <rguenther@suse.de>
20520 * dwarf2out.c (output_die): Handle flag_generate_offload like
20522 (output_comp_unit): Likewise.
20523 (gen_array_type_die): Likewise.
20524 (dwarf2out_early_finish): Likewise.
20525 (note_variable_value_in_expr): Likewise.
20526 (dwarf2out_finish): Likewise. Adjust assert.
20527 * cgraphunit.c (symbol_table::compile): Move setting of
20528 flag_generate_offload earlier ...
20529 (symbol_table::finalize_compilation_unit): ... here, before
20530 early debug finalization.
20532 2017-08-24 Richard Biener <rguenther@suse.de>
20534 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20535 and ipa-fnsummary.h.
20536 (ix86_can_inline_p): When ix86_fpmath flags do not match
20537 check whether the callee uses FP math at all.
20539 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20541 PR middle-end/81931
20542 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20543 instead of TYPE_PRECISION.
20545 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20546 Alan Hayward <alan.hayward@arm.com>
20547 David Sherwood <david.sherwood@arm.com>
20549 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20551 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20553 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20555 (ix86_frame::stack_realign_allocate): New field.
20556 (struct machine_frame_state): Modify comments.
20557 (machine_frame_state::sp_realigned_fp_end): New field.
20558 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20559 layout calculation.
20560 (sp_valid_at): Add assertion to assure no attempt to access invalid
20561 offset of a realigned stack.
20562 (fp_valid_at): Likewise.
20563 (choose_baseaddr): Modify comments.
20564 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20565 ix86_expand_prologue.
20566 (ix86_expand_prologue): Modify stack realignment and allocation.
20567 (ix86_expand_epilogue): Modify comments.
20568 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20569 avx2_runtime, avx512f, and avx512f_runtime.
20571 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20573 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20574 (mstackrealign): Do not init to -1.
20575 * config/i386/i386.c (ix86_option_override_internal):
20576 Check opts_set, not opts when setting default value of
20577 opts->x_ix86_force_align_arg_pointer.
20579 2017-08-23 Richard Biener <rguenther@suse.de>
20581 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20582 lang_hooks.decl_printable_name.
20583 * print-rtl-function.c (print_rtx_function): Likewise.
20584 * tree-pretty-print.c (dump_function_header): Likewise.
20586 2017-08-23 Richard Biener <rguenther@suse.de>
20589 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20590 -g0 at compile-time.
20592 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20594 PR middle-end/19706
20595 * doc/sourcebuild.texi (Other hardware attributes):
20598 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20600 PR middle-end/19706
20601 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20602 Fix single-use check.
20604 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20606 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20608 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20610 * doc/install.texi: Modify to add more details on running selected
20613 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20615 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20616 is combined with -mabi=ms.
20617 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20620 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20622 PR tree-optimization/81488
20623 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20624 and cached_basis fields.
20625 (MAX_SPREAD): New constant.
20626 (alloc_cand_and_find_basis): Initialize new fields.
20627 (clear_visited): New function.
20628 (create_phi_basis_1): Rename from create_phi_basis, set visited
20629 and cached_basis fields.
20630 (create_phi_basis): New wrapper function.
20631 (phi_add_costs_1): Rename from phi_add_costs, add spread
20632 parameter, set visited field, short-circuit when limits reached.
20633 (phi_add_costs): New wrapper function.
20634 (record_phi_increments_1): Rename from record_phi_increments, set
20636 (record_phi_increments): New wrapper function.
20637 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20638 (phi_incr_cost): New wrapper function.
20639 (all_phi_incrs_profitable_1): Rename from
20640 all_phi_incrs_profitable, set visited field.
20641 (all_phi_incrs_profitable): New wrapper function.
20643 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20644 Alan Hayward <alan.hayward@arm.com>
20645 David Sherwood <david.sherwood@arm.com>
20647 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20648 that takes the outer and inner modes.
20649 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20650 comparison as the canonical test for a paradoxical subreg.
20651 * combine.c (simplify_set): Use paradoxical_subreg_p.
20652 (make_extraction): Likewise.
20653 (force_to_mode): Likewise.
20654 (rtx_equal_for_field_assignment_p): Likewise.
20655 (gen_lowpart_for_combine): Likewise.
20656 (simplify_comparison): Likewise.
20657 * cse.c (equiv_constant): Likewise.
20658 * expmed.c (store_bit_field_1): Likewise.
20659 * final.c (alter_subreg): Likewise.
20660 * fwprop.c (propagate_rtx): Likewise.
20661 (forward_propagate_subreg): Likewise.
20662 * ira-conflicts.c (ira_build_conflicts): Likewise.
20663 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20664 * lra-constraints.c (curr_insn_transform): Likewise.
20665 (split_reg): Likewise.
20666 * lra-eliminations.c (move_plus_up): Likewise.
20667 (lra_eliminate_regs_1): Likewise.
20668 * recog.c (general_operand): Likewise.
20669 * ree.c (combine_reaching_defs): Likewise.
20670 * reload.c (push_reload): Likewise.
20671 (find_reloads): Likewise.
20672 * reload1.c (elimination_effects): Likewise.
20673 (compute_reload_subreg_offset): Likewise.
20674 (choose_reload_regs): Likewise.
20675 * rtlanal.c (subreg_lsb_1): Likewise.
20676 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20677 (simplify_subreg): Likewise.
20678 * var-tracking.c (track_loc_p): Likewise.
20679 * emit-rtl.c (byte_lowpart_offset): Likewise.
20680 (paradoxical_subreg_p): Delete out-of-line definition.
20682 2017-08-22 Jeff Law <law@redhat.com>
20684 PR tree-optimization/81741
20685 PR tree-optimization/71947
20686 * tree-ssa-dom.c: Include tree-inline.h.
20687 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20688 equivalences if one is more expensive to compute than the other.
20689 * tree-ssa-scopedtables.h (class const_or_copies): Make
20690 record_const_or_copy_raw method private.
20691 (class avail_exprs_stack): New method simplify_binary_operation.
20692 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20693 avail_exprs_stack::simplify_binary_operation as needed.
20694 (avail_exprs_stack::simplify_binary_operation): New function.
20696 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20698 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20699 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20700 (DOT_SYMBOLS): Likewise.
20701 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20702 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20703 (RS6000_ABI_NAME): Likewise.
20704 (TARGET_CMODEL): Likewise.
20705 (TOC_SECTION_ASM_OP): Likewise.
20706 (SET_CMODEL): New macro.
20707 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20709 2017-08-22 Richard Biener <rguenther@suse.de>
20711 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20712 to allow for free-lang-data replacements similar to verify_type_variant.
20714 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20715 Alan Hayward <alan.hayward@arm.com>
20716 David Sherwood <david.sherwood@arm.com>
20718 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20719 VOIDmode for the LABEL_REF.
20721 2017-08-22 Richard Biener <rguenther@suse.de>
20723 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20725 2017-08-22 Richard Biener <rguenther@suse.de>
20727 * common.opt (feliminate-dwarf2-dups): Ignore.
20728 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20729 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20730 same_die_p_wrap, compute_section_prefix,
20731 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20732 (comdat_symbol_id, comdat_symbol_number): Likewise.
20733 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20735 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20736 (output_die): Mark unreachable path unreachable.
20737 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20738 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20739 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20740 (dwarf2out_early_finish): Likewise.
20742 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20744 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20746 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20749 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20750 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20751 (avr_attribute_table) <io, io_low, address>: Initialize
20752 .decl_required with true.
20754 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20756 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20757 undocumented debugging options.
20758 (-mvsx-scalar-double): Likewise.
20759 (-mallow-df-permute): Likewise.
20760 (-mvectorize-builtins): Likewise.
20761 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20762 (rs6000_builtin_vectorized_function): Likewise.
20763 (rs6000_builtin_md_vectorized_function): Likewise.
20764 (rs6000_opt_vars): Likewise.
20766 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20769 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20770 (*btrq_imm): Rename from *btrq.
20771 (*btcq_imm): Rename from *btcq.
20772 (btsc): New code attribute.
20773 (*<btsc><mode>): New insn pattern.
20774 (*btr<mode>): Ditto.
20775 (*<btsc><mode>_mask): New insn_and_split pattern.
20776 (*btr<mode>_mask): Ditto.
20778 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20779 Alan Hayward <alan.hayward@arm.com>
20780 David Sherwood <david.sherwood@arm.com>
20782 * function.c (pad_below): Simplify padding calculation.
20784 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20785 Alan Hayward <alan.hayward@arm.com>
20786 David Sherwood <david.sherwood@arm.com>
20788 * target.def (function_prologue): Remove frame size argument.
20789 (function_epilogue): Likewise.
20790 * doc/tm.texi: Regenerate.
20791 * final.c (final_start_function): Update call to function_prologue.
20792 (final_end_function): Update call to function_epilogue.
20793 (default_function_pro_epilogue): Remove frame size argument.
20794 * output.h (default_function_pro_epilogue): Likewise.
20795 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20796 (arm_output_function_prologue): Likewise.
20797 * config/frv/frv.c (frv_function_prologue): Likewise.
20798 (frv_function_epilogue): Likewise.
20799 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20800 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20801 (ia64_output_function_epilogue): Likewise.
20802 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20803 (m32r_output_function_epilogue): Likewise.
20804 * config/microblaze/microblaze.c (microblaze_function_prologue)
20805 (microblaze_function_epilogue): Likewise.
20806 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20807 (mips_output_function_epilogue): Likewise.
20808 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20809 (mmix_target_asm_function_epilogue): Likewise.
20810 * config/msp430/msp430.c (msp430_start_function): Likewise.
20811 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20812 (nds32_asm_function_epilogue): Likewise.
20813 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20814 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20815 (pa_output_function_epilogue): Likewise.
20816 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20817 (rs6000_output_function_epilogue): Likewise.
20818 * config/rl78/rl78.c (rl78_start_function): Likewise.
20819 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20820 (rs6000_output_function_epilogue): Likewise.
20821 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20822 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20823 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20824 (sparc_asm_function_epilogue): Likewise.
20826 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20828 * tree.h (type_has_mode_precision_p): New function.
20829 * convert.c (convert_to_integer_1): Use it.
20830 * expr.c (expand_expr_real_2): Likewise.
20831 (expand_expr_real_1): Likewise.
20832 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20833 * match.pd: Likewise.
20834 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20835 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20836 * tree-tailcall.c (process_assignment): Likewise.
20837 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20838 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20839 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20840 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20841 (vectorizable_assignment): Likewise.
20842 (vectorizable_shift): Likewise.
20843 (vectorizable_operation): Likewise.
20844 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20846 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20848 * match.pd: Add pow (C, x) simplification.
20850 2017-08-21 Richard Biener <rguenther@suse.de>
20852 PR tree-optimization/81900
20853 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20854 for blocks with abnormal predecessors.
20855 (compute_antic): Do not set visited flag prematurely.
20857 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20860 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20862 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20864 * stor-layout.h (vector_type_mode): Move to...
20865 * tree.h (vector_type_mode): ...here.
20866 * stor-layout.c (vector_type_mode): Move to...
20867 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20869 2017-08-21 Richard Biener <rguenther@suse.de>
20871 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20872 register_external_die hooks.
20873 (debug_false_tree_charstarstar_uhwistar): Declare.
20874 (debug_nothing_tree_charstar_uhwi): Likewise.
20875 * debug.c (do_nothing_debug_hooks): Adjust.
20876 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20877 (debug_nothing_tree_charstar_uhwi): Likewise.
20878 * dbxout.c (dbx_debug_hooks): Adjust.
20879 (xcoff_debug_hooks): Likewise.
20880 * sdbout.c (sdb_debug_hooks): Likewise.
20881 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20882 * dwarf2out.c (macinfo_label_base): New global.
20883 (dwarf2out_register_external_die): New function for the
20884 register_external_die hook.
20885 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20886 (dwarf2_debug_hooks): Use them.
20887 (dwarf2_lineno_debug_hooks): Adjust.
20888 (struct die_struct): Add with_offset flag.
20889 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20890 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20891 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20892 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20893 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20894 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20895 defining section names for the early LTO debug variants.
20896 (reset_indirect_string): New helper.
20897 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20898 (print_dw_val): Add support for offsetted symbol references.
20899 (get_ultimate_context): Split out from is_cxx.
20900 (is_cxx): Use get_ultimate_context.
20901 (is_fortran): Add decl overload.
20902 (compute_comp_unit_symbol): Split out worker from
20903 compute_section_prefix.
20904 (compute_section_prefix): Call compute_comp_unit_symbol and
20905 set comdat_type_p here.
20906 (output_die): Skip DIE symbol output for the LTO added one.
20907 Handle DIE symbol references with offset.
20908 (output_comp_unit): Guard section name mangling properly.
20909 For LTO debug sections emit a symbol at the section beginning
20910 which we use to refer to its DIEs.
20911 (add_abstract_origin_attribute): For DIEs registered via
20912 dwarf2out_register_external_die directly refer to the early
20913 DIE rather than indirectly through the shadow one we created.
20914 Remove obsolete call to dwarf2out_abstract_function for
20915 non-function/block origins.
20916 (gen_array_type_die): When generating early LTO debug do
20917 not emit DW_AT_string_length.
20918 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20919 late when in LTO. As suggested place a gcc_unreachable for
20920 the DECL_ABSTRACT_P case.
20921 (gen_subprogram_die): Avoid another specification DIE
20922 for early built declarations/definitions for the late LTO case.
20923 (gen_variable_die): Add type references for late duplicated VLA dies
20925 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20926 we have the abstract instance already.
20927 (process_scope_var): Adjust decl DIE contexts in LTO which
20928 first puts them in limbo.
20929 (gen_decl_die): Do not generate type DIEs late apart from
20930 types for VLAs or for decls we do not yet have a DIE. Do not
20931 call dwarf2out_abstract_function late.
20932 (dwarf2out_early_global_decl): Make sure to create DIEs
20933 for abstract instances of a decl first.
20934 (dwarf2out_late_global_decl): Adjust comment.
20935 (output_macinfo_op): With multiple macro sections use
20936 macinfo_label_base to distinguish labels.
20937 (output_macinfo): Likewise. Update macinfo_label_base.
20938 Pass in the line info label.
20939 (note_variable_value_in_expr): When generating LTO resolve
20940 all variable values here by generating DIEs as needed.
20941 (init_sections_and_labels): Add early LTO debug flag parameter
20942 and generate different sections and names if set. Add generation
20943 counter for the labels so we can have multiple of them.
20944 (reset_dies): Helper to allow DIEs to be output multiple times.
20945 (dwarf2out_finish): When outputting DIEs to the fat part of an
20946 LTO object first reset DIEs.
20947 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20948 (modified_type_die): Check for decl_ultimate_origin being self
20950 (gen_type_die_with_usage): Likewise.
20951 (gen_typedef_die): Allow decl_ultimate_origin being self.
20952 (set_decl_abstract_flags): Remove.
20953 (set_block_abstract_flags): Likewise.
20954 (dwarf2out_abstract_function): Treat the early generated DIEs
20955 as the abstract copy and only add DW_AT_inline and
20956 DW_AT_artificial here and call set_decl_origin_self.
20957 If the DIE has an abstract origin don't do anything.
20958 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20959 if we have none yet (Go fails to build one, PR78628).
20960 (variably_modified_type_p): Prevent endless recursion for Ada
20961 cyclic pointer types.
20962 * lto-streamer-in.c: Include debug.h.
20963 (dref_queue): New global.
20964 (lto_read_tree_1): Stream in DIE references.
20965 (lto_input_tree): Register DIE references.
20966 (input_function): Stream DECL_DEBUG_ARGS.
20967 * lto-streamer-out.c: Include debug.h.
20968 (lto_write_tree_1): Output DIE references.
20969 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20970 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20971 (output_function): Stream DECL_DEBUG_ARGS.
20972 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20973 Stream DECL_ABSTRACT_ORIGIN.
20974 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20975 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20976 DECL_CONTEXT for file-scope decls.
20977 * lto-streamer.h (struct dref_entry): Declare.
20978 (dref_queue): Likewise.
20979 * cfgexpand.c (pass_expand::execute): Do not call the
20980 outlining_inline_function hook here.
20981 * lto-wrapper.c (debug_obj): New global.
20982 (tool_cleanup): Unlink it if required.
20983 (debug_objcopy): New function.
20984 (run_gcc): Handle early debug sections in the IL files by
20985 extracting them to separate files, partially linkin them and
20986 feeding the result back as result to the linker.
20987 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20988 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20989 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20990 sections into a separate segment.
20991 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20993 (darwin_asm_dwarf_section): Likewise.
20994 (darwin_asm_output_dwarf_offset): Likewise.
20995 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20997 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20998 Alan Hayward <alan.hayward@arm.com>
20999 David Sherwood <david.sherwood@arm.com>
21001 * read-md.h (md_reader::record_potential_iterator_use): Replace
21002 pointer argument with an rtx and an index.
21003 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21004 (apply_mode_iterator): Likewise.
21005 (apply_code_iterator): Likewise.
21006 (apply_int_iterator): Likewise.
21007 (apply_subst_iterator): Likewise.
21008 (record_iterator_use): Likewise.
21009 (record_attribute_use): Likewise.
21010 (md_reader::record_potential_iterator_use): Likewise. Update calls
21011 to record_iterator_use and apply_iterator.
21012 (iterator_use): Replace ptr with x and index.
21013 (attribute_use): Likewise.
21014 (apply_attribute_uses): Update calls to apply_iterator.
21015 (apply_iterators): Likewise. Update initialization of iterator_use.
21016 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21017 and record_potential_iterator_use.
21018 (rtx_reader::read_rtx_operand): Likewise.
21020 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21021 Alan Hayward <alan.hayward@arm.com>
21022 David Sherwood <david.sherwood@arm.com>
21024 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21027 2017-08-21 Richard Biener <rguenther@suse.de>
21029 PR middle-end/81884
21030 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21031 at struct end conservatively when comparing common bases.
21033 2017-08-21 Richard Biener <rguenther@suse.de>
21035 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21036 (mem_ref_in_stmt): Remove.
21037 (determine_max_movement): Use ref index to get at the reference.
21038 (invariantness_dom_walker::before_dom_children): Deal with
21039 lim data already initialized.
21040 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21042 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21044 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21045 (TARGET_ISA_ROUND): Ditto.
21046 (TARGET_ROUND): Ditto.
21047 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21048 * config/i386/i386.md: Ditto.
21049 * config/i386/sse.md: Ditto.
21050 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21051 with OPTION_MASK_ISA_SSE4_1.
21053 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21056 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21057 __builtin_ia32_lzcnt_u16.
21059 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21062 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21063 (rs6000_set_current_function): Rewrite function to use it.
21065 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21068 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21069 and TYPE_WARN_IF_NOT_ALIGN.
21070 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21071 (handle_warn_if_not_align): New.
21072 (place_union_field): Call handle_warn_if_not_align.
21073 (place_field): Call handle_warn_if_not_align.
21074 Copy TYPE_WARN_IF_NOT_ALIGN.
21075 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21076 (layout_type): Likewise.
21077 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21079 (tree_decl_common): Add warn_if_not_align.
21080 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21081 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21082 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21083 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21084 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21085 * doc/extend.texi: Document warn_if_not_aligned attribute.
21086 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21088 2017-08-17 Martin Liska <mliska@suse.cz>
21091 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21092 (get_data_dependence): Use it as pointer type.
21093 (distribute_loop): Likewise.
21095 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21097 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21098 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21099 (p8_vmrgow_<mode>_direct): New define_insn.
21100 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21101 handle endianness for vmrgew and vmrgow permute patterns.
21103 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21105 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21106 * config/rs6000/rs6000-cpus.def: Remove comment.
21107 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21108 (POWERPC_MASKS): Likewise.
21109 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21110 use of TARGET_VSX_TIMODE.
21111 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21112 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21113 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21114 (rs6000_option_override_internal): Remove dead code.
21115 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21116 (rs6000_legitimize_reload_address): Likewise.
21117 (rs6000_legitimate_address_p): Likewise.
21118 (rs6000_opt_masks): Delete "vsx-timode".
21119 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21120 from function comment.
21121 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21122 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21123 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21125 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21126 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21127 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21128 (VSX_TI): Likewise.
21130 (define_peephole2): Likewise.
21132 2017-08-17 Martin Sebor <msebor@redhat.com>
21135 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21136 past the end of an array.
21137 (test_pp_format): Add test cases.
21139 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21141 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21142 missing ECF_NOTHROW flags.
21144 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21147 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21148 operands residing in integer registers.
21149 (*vsx_le_perm_load_<mode>): Likewise.
21150 (*vsx_le_perm_store_<mode>): Likewise.
21151 (define_peephole2): Add peepholes to optimize the above.
21153 2017-08-17 Marek Polacek <polacek@redhat.com>
21155 PR middle-end/81814
21156 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21157 to mimic what shorten_compare did. Change the return type to bool.
21158 (fold_cond_expr_with_comparison): Update call to
21159 operand_equal_for_comparison_p.
21160 (fold_ternary_loc): Likewise.
21162 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21164 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21166 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21168 2017-08-17 Richard Biener <rguenther@suse.de>
21170 * tree-ssa-structalias.c (solve_graph): When propagating
21171 to successors update the graphs succ edges and avoid duplicate work.
21173 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21176 * config/i386/i386.c (ix86_option_override_internal): Save target
21177 specific options after ix86_stack_protector_guard_reg was changed.
21179 2017-08-17 Richard Biener <rguenther@suse.de>
21181 PR tree-optimization/81827
21182 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21183 (new_var_info): Initialize it conservatively.
21184 (get_call_vi): Mark register vars.
21185 (new_scalar_tmp_constraint_exp): Likewise.
21186 (handle_rhs_call): Likewise.
21187 (handle_const_call): Likewise.
21188 (create_function_info_for): Likewise.
21189 (solve_constraints): Sort varinfos to separate register from
21190 non-register vars to pack points-to solution bitmaps during
21193 2017-08-17 Marek Polacek <polacek@redhat.com>
21195 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21197 2017-08-17 Richard Biener <rguenther@suse.de>
21199 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21200 to true when overflow is undefined and we saturated the result.
21202 2017-08-17 Alan Modra <amodra@gmail.com>
21205 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21206 Don't use store multiple if only one reg needs saving.
21207 (interesting_frame_related_regno): New function.
21208 (rs6000_frame_related): Don't emit frame info for regs that
21210 (rs6000_emit_epilogue): Likewise.
21212 2017-08-16 Nathan Sidwell <nathan@acm.org>
21214 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21215 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21216 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21217 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21218 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21219 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21220 (verify_type): Adjust for TYPE_BINFO move.
21221 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21222 process TYPE_BINFO directly.
21223 (hash_tree): Likewise.
21224 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21226 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21229 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21231 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21233 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21236 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21237 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21238 Add ix86_binary_operator_ok to insn constraint.
21239 (*iordi_1_bts): Ditto.
21240 (*xordi_1_btc): Ditto.
21241 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21242 Update corresponding peephole2 pattern.
21246 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21248 PR tree-optimization/81832
21249 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21250 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21252 2017-08-16 Marek Polacek <polacek@redhat.com>
21255 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21256 regarding size_zero_node.
21258 2017-08-16 Martin Liska <mliska@suse.cz>
21261 * config.gcc: Respect previously set extra_objs in case
21264 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21266 PR tree-optimization/81835
21267 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21268 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21269 not depend on the phi.
21271 2017-08-16 Alan Modra <amodra@gmail.com>
21273 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21276 2017-08-16 Alan Modra <amodra@gmail.com>
21278 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21279 (save_reg_p): ..into this. Update all callers.
21280 (first_reg_to_save): Simplify.
21282 2017-08-16 Alan Modra <amodra@gmail.com>
21284 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21287 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21291 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21292 constant count if that count is less than 32.
21294 2017-08-15 Nathan Sidwell <nathan@acm.org>
21296 * gcc.c (execute): Emit friendlier message if inferior is killed
21297 by an external cause.
21299 2017-08-15 Richard Biener <rguenther@suse.de>
21301 PR tree-optimization/81790
21302 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21303 CONSTRUCTORs from simplifying and VN.
21305 2017-08-14 Martin Sebor <msebor@redhat.com>
21307 * builtin-attrs.def: Add comments.
21309 2017-08-14 Martin Sebor <msebor@redhat.com>
21312 * doc/extend.texi (attribute nonstring): Document new attribute.
21314 2017-08-14 Martin Sebor <msebor@redhat.com>
21317 * tree-diagnostic.c (default_tree_printer): Handle %G.
21318 * gimple-pretty-print.h (percent_G_format): Declare new function.
21319 * gimple-pretty-print.c (percent_G_format): Define.
21320 * tree-pretty-print.c (percent_K_format): Add argument.
21322 2017-08-14 Martin Sebor <msebor@redhat.com>
21324 PR translation/79998
21325 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21326 Remove a stray space.
21328 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21331 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21332 (*iordi_1_bts): Ditto.
21333 (*xordi_1_btc): Ditto.
21335 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21338 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21339 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21341 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21342 quoted strings, and make more translator-friendly.
21343 (darwin_rs6000_override_options): Likewise.
21344 (rs6000_option_override_internal): Likewise.
21345 (rs6000_return_in_memory): Fix overlong line.
21346 (init_cmulative_args): Use quoted strings, and make more
21347 translator-friendly.
21348 (rs6000_pass_by_reference): Fix overlong line.
21349 (def_builtin): Use quoted strings.
21350 (altivec_expand_predicate_builtin): Use quoted strings, and make
21351 more translator-friendly.
21352 (htm_expand_builtin): Use quoted strings.
21353 (cpu_expand_builtin): Use quoted strings, and make more
21354 translator-friendly.
21355 (altivec_expand_builtin): Likewise.
21356 (paired_expand_predicate_builtin): Likewise.
21357 (rs6000_invalid_builtin): Likewise.
21358 (builtin_function_type): Use quoted strings.
21359 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21360 more translator-friendly.
21361 (rs6000_trampoline_init): Likewise.
21362 (rs6000_handle_altivec_attribute): Likewise.
21363 (rs6000_inner_target_options): Use quoted strings.
21364 (rs6000_disable_incompatible_switches): Likewise.
21365 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21366 strings, and make more translator-friendly.
21367 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21369 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21371 PR tree-optimization/81799
21372 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21373 cond_expr to simple gimple operand.
21375 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21377 PR middle-end/46932
21378 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21380 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21384 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21385 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21386 TARGET_GASISR_PROLOGUES.
21387 * config/avr/avr.c (avr_option_override): Same.
21388 (avr_pass_pre_proep::execute): Same.
21390 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21393 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21394 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21396 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21398 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21399 define_insn_and_split. Split to a memory load from 0 in
21400 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21401 using PTR mode iterator.
21402 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21403 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21404 (*add_tp_<mode>): Redefine as define_insn_and_split.
21405 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21406 address space. Merge with *add_tp_x32 using PTR mode iterator.
21407 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21408 Split to an add with a memory load from 0 in
21409 DEFAULT_TLS_SEG_REG address space.
21411 2017-08-12 Andrew Pinski <apinski@cavium.com>
21413 * config/aarch64/aarch64-option-extensions.def (rdma):
21414 Fix feature string to what Linux prints out in /proc/cpuinfo.
21416 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21419 * dwarf2out.c (modified_type_die): For C typedef types that have
21420 an ultimate origin, process the ultimate origin instead of the
21422 (gen_typedef_die): Assert that input DECLs have no ultimate
21424 (gen_type_die_with_usage): For typedef variants that have an
21425 ultimate origin, just call gen_decl_die on the original DECL.
21426 (process_scope_var): Avoid creating DIEs for local typedefs and
21427 concrete static variables.
21429 2017-08-12 Alan Modra <amodra@gmail.com>
21433 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21434 match gnu-user.h startfile.
21435 (ENDFILE_LINUX_SPEC): Similarly.
21437 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21440 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21442 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21444 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21445 * config/aarch64/aarch64.md (mov<mode>): Change.
21446 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21447 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21448 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21450 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21452 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21453 for storage order barriers.
21455 2017-08-11 Martin Liska <mliska@suse.cz>
21458 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21459 variables of void type.
21461 2017-08-11 Martin Liska <mliska@suse.cz>
21463 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21464 TARGET_SUPPORTS_ALIASES.
21465 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21466 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21467 (optimize_weakref): Likewise.
21468 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21469 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21470 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21472 2017-08-11 Martin Liska <mliska@suse.cz>
21475 * config/i386/i386.c (make_resolver_func): Do complete
21476 refactoring of the function.
21478 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21481 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21482 * config/i386/i386.c (ix86_stack_protect_guard): Use
21483 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21484 * doc/invoke.texi (x86 Options): Document
21485 -mstack-protector-guard-symbol= option.
21487 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21489 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21490 * config/i386/i386.c (ix86_split_stack_guard): New function.
21491 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21492 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21493 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21494 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21495 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21496 (split_stack_space_check): Call ix86_split_stack_guard.
21498 2017-08-10 Martin Sebor <msebor@redhat.com>
21500 * print-tree.c (print_node): Print location using the established
21502 Replace spaces with colons.
21503 (debug_raw, debug): Ditto.
21505 2017-08-10 Martin Sebor <msebor@redhat.com>
21508 * pretty-print.c (pp_format): Correct the handling of %s precision.
21510 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21513 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21515 (ix86_finalize_stack_frame_flags): This. Also clear
21516 frame_pointer_needed if -fno-omit-frame-pointer is used without
21518 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21519 with ix86_finalize_stack_frame_flags.
21520 (ix86_expand_epilogue): Likewise.
21521 (ix86_expand_split_stack_prologue): Likewise.
21522 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21524 2017-08-10 Martin Liska <mliska@suse.cz>
21527 * c-attribs.c (handle_target_attribute):
21528 Report warning for an empty string argument of target attribute.
21530 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21533 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21535 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21536 or DECL_NONLOCAL labels.
21537 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21538 or DECL_NONLOCAL labels here.
21540 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21542 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21543 to indicate when early gimple folding has been disabled.
21544 (rs6000_gimple_fold_builtin): Add debug content.
21545 (rs6000_invalid_builtin): Fix whitespace.
21546 (rs6000_expand_builtin): Fix whitespace.
21547 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21549 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21552 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21553 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21555 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21557 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21558 (qdf24xx): Likewise.
21559 * config/aarch64/aarch64.md: Include falkor.md.
21560 * config/aarch64/falkor.md: New.
21562 2017-08-09 Marek Polacek <polacek@redhat.com>
21565 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21566 * diagnostic.c (emit_diagnostic): Add a comment.
21567 (emit_diagnostic_valist): New function.
21569 2017-08-09 Marek Polacek <polacek@redhat.com>
21572 * input.c (make_location): New overload.
21573 * input.h (make_location): Declare.
21575 2017-08-08 Alan Modra <amodra@gmail.com>
21576 H.J. Lu <hongjiu.lu@intel.com>
21579 * gcc.c (NO_PIE_SPEC): Delete.
21580 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21582 (LINK_PIE_SPEC): ..to here.
21583 (LINK_COMMAND_SPEC): Support -no-pie.
21584 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21585 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21587 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21588 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21589 (ENDFILE_CRTEND_SPEC): Similarly.
21591 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21594 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21595 (mstack-protector-guard-offset=): Ditto.
21596 * config/i386/i386.c (ix86_option_override): Handle
21597 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21599 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21600 ix86_stack_protect_guard_offset variables.
21601 (TARGET_STACK_PROTECT_GUARD): Always define.
21602 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21603 and -mstack-protector-guard-offset= options.
21605 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21607 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21608 boundary case for the last candidate.
21610 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21612 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21613 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21615 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21617 PR middle-end/19706
21618 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21619 * config/aarch64/aarch64-builtins.c
21620 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21621 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21622 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21624 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21625 Andrew Pinski <pinskia@gmail.com>
21627 PR middle-end/19706
21628 * internal-fn.def (XORSIGN): New.
21629 * optabs.def (xorsign_optab): New.
21630 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21631 (convert_expand_mult_copysign): New.
21632 (pass_optimize_widening_mul::execute): Call
21633 convert_expand_mult_copysign.
21635 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21637 PR tree-optimization/81354
21638 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21639 Insert on edges rather than explicitly creating landing pads.
21640 (analyze_candidates_and_replace): Commit edge inserts.
21642 2017-08-08 Richard Biener <rguenther@suse.de>
21644 PR middle-end/81719
21645 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21646 (expand_simple_operations): Also look through ADDR_EXPRs with
21647 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21649 2017-08-08 Richard Biener <rguenther@suse.de>
21651 PR tree-optimization/81723
21652 * tree-vect-slp.c (struct bst_traits): New hash traits.
21653 (bst_fail): New global.
21654 (vect_build_slp_tree_2): New worker, split out from ...
21655 (vect_build_slp_tree): ... this now wrapping it with using
21656 bst_fail set to cache SLP tree build fails. Properly handle
21658 (vect_analyze_slp_instance): Allocate and free bst_fail.
21660 2017-08-08 Martin Liska <mliska@suse.cz>
21663 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21664 LABEL_DECLs that can be from a different function.
21666 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21668 PR tree-optimization/81744
21669 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21670 loop's number of iterations.
21672 2017-08-08 Martin Liska <mliska@suse.cz>
21674 * asan.c: Include header files.
21675 * attribs.c (build_decl_attribute_variant): New function moved
21677 (build_type_attribute_qual_variant): Likewise.
21678 (cmp_attrib_identifiers): Likewise.
21679 (simple_cst_list_equal): Likewise.
21680 (omp_declare_simd_clauses_equal): Likewise.
21681 (attribute_value_equal): Likewise.
21682 (comp_type_attributes): Likewise.
21683 (build_type_attribute_variant): Likewise.
21684 (lookup_ident_attribute): Likewise.
21685 (remove_attribute): Likewise.
21686 (merge_attributes): Likewise.
21687 (merge_type_attributes): Likewise.
21688 (merge_decl_attributes): Likewise.
21689 (merge_dllimport_decl_attributes): Likewise.
21690 (handle_dll_attribute): Likewise.
21691 (attribute_list_equal): Likewise.
21692 (attribute_list_contained): Likewise.
21693 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21694 (lookup_attribute_by_prefix): Likewise.
21695 * bb-reorder.c: Include header files.
21696 * builtins.c: Likewise.
21697 * calls.c: Likewise.
21698 * cfgexpand.c: Likewise.
21699 * cgraph.c: Likewise.
21700 * cgraphunit.c: Likewise.
21701 * convert.c: Likewise.
21702 * dwarf2out.c: Likewise.
21703 * final.c: Likewise.
21704 * fold-const.c: Likewise.
21705 * function.c: Likewise.
21706 * gimple-expr.c: Likewise.
21707 * gimple-fold.c: Likewise.
21708 * gimple-pretty-print.c: Likewise.
21709 * gimple.c: Likewise.
21710 * gimplify.c: Likewise.
21711 * hsa-common.c: Likewise.
21712 * hsa-gen.c: Likewise.
21713 * internal-fn.c: Likewise.
21714 * ipa-chkp.c: Likewise.
21715 * ipa-cp.c: Likewise.
21716 * ipa-devirt.c: Likewise.
21717 * ipa-fnsummary.c: Likewise.
21718 * ipa-inline.c: Likewise.
21719 * ipa-visibility.c: Likewise.
21721 * lto-cgraph.c: Likewise.
21722 * omp-expand.c: Likewise.
21723 * omp-general.c: Likewise.
21724 * omp-low.c: Likewise.
21725 * omp-offload.c: Likewise.
21726 * omp-simd-clone.c: Likewise.
21727 * opts-global.c: Likewise.
21728 * passes.c: Likewise.
21729 * predict.c: Likewise.
21730 * sancov.c: Likewise.
21731 * sanopt.c: Likewise.
21732 * symtab.c: Likewise.
21733 * toplev.c: Likewise.
21734 * trans-mem.c: Likewise.
21735 * tree-chkp.c: Likewise.
21736 * tree-eh.c: Likewise.
21737 * tree-into-ssa.c: Likewise.
21738 * tree-object-size.c: Likewise.
21739 * tree-parloops.c: Likewise.
21740 * tree-profile.c: Likewise.
21741 * tree-ssa-ccp.c: Likewise.
21742 * tree-ssa-live.c: Likewise.
21743 * tree-ssa-loop.c: Likewise.
21744 * tree-ssa-sccvn.c: Likewise.
21745 * tree-ssa-structalias.c: Likewise.
21746 * tree-ssa.c: Likewise.
21747 * tree-streamer-in.c: Likewise.
21748 * tree-vectorizer.c: Likewise.
21749 * tree-vrp.c: Likewise.
21750 * tsan.c: Likewise.
21751 * ubsan.c: Likewise.
21752 * varasm.c: Likewise.
21753 * varpool.c: Likewise.
21754 * tree.c: Remove functions moved to attribs.[ch].
21755 * tree.h: Likewise.
21756 * config/aarch64/aarch64.c: Add attrs.h header file.
21757 * config/alpha/alpha.c: Likewise.
21758 * config/arc/arc.c: Likewise.
21759 * config/arm/arm.c: Likewise.
21760 * config/avr/avr.c: Likewise.
21761 * config/bfin/bfin.c: Likewise.
21762 * config/c6x/c6x.c: Likewise.
21763 * config/cr16/cr16.c: Likewise.
21764 * config/cris/cris.c: Likewise.
21765 * config/darwin.c: Likewise.
21766 * config/epiphany/epiphany.c: Likewise.
21767 * config/fr30/fr30.c: Likewise.
21768 * config/frv/frv.c: Likewise.
21769 * config/ft32/ft32.c: Likewise.
21770 * config/h8300/h8300.c: Likewise.
21771 * config/i386/winnt.c: Likewise.
21772 * config/ia64/ia64.c: Likewise.
21773 * config/iq2000/iq2000.c: Likewise.
21774 * config/lm32/lm32.c: Likewise.
21775 * config/m32c/m32c.c: Likewise.
21776 * config/m32r/m32r.c: Likewise.
21777 * config/m68k/m68k.c: Likewise.
21778 * config/mcore/mcore.c: Likewise.
21779 * config/microblaze/microblaze.c: Likewise.
21780 * config/mips/mips.c: Likewise.
21781 * config/mmix/mmix.c: Likewise.
21782 * config/mn10300/mn10300.c: Likewise.
21783 * config/moxie/moxie.c: Likewise.
21784 * config/msp430/msp430.c: Likewise.
21785 * config/nds32/nds32-isr.c: Likewise.
21786 * config/nds32/nds32.c: Likewise.
21787 * config/nios2/nios2.c: Likewise.
21788 * config/nvptx/nvptx.c: Likewise.
21789 * config/pa/pa.c: Likewise.
21790 * config/pdp11/pdp11.c: Likewise.
21791 * config/powerpcspe/powerpcspe.c: Likewise.
21792 * config/riscv/riscv.c: Likewise.
21793 * config/rl78/rl78.c: Likewise.
21794 * config/rx/rx.c: Likewise.
21795 * config/s390/s390.c: Likewise.
21796 * config/sh/sh.c: Likewise.
21797 * config/sol2.c: Likewise.
21798 * config/sparc/sparc.c: Likewise.
21799 * config/spu/spu.c: Likewise.
21800 * config/stormy16/stormy16.c: Likewise.
21801 * config/tilegx/tilegx.c: Likewise.
21802 * config/tilepro/tilepro.c: Likewise.
21803 * config/v850/v850.c: Likewise.
21804 * config/vax/vax.c: Likewise.
21805 * config/visium/visium.c: Likewise.
21806 * config/xtensa/xtensa.c: Likewise.
21808 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21811 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21812 constraints since the -mupper-regs-* switches have been
21814 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21815 into a vector from a double word element that was extracted from
21816 another vector, and eliminate extra XXPERMDI instructions.
21817 (vsx_concat_<mode>_2): Likewise.
21818 (vsx_concat_<mode>_3): Likewise.
21819 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21820 concat to allow optimizing inserts from previous extracts.
21822 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21824 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21825 memory reference to a SSP offset in TLS address space.
21826 (ix86_print_operand) <case '@'>: Remove.
21827 (ix86_print_operand_punct_valid_p): Remove '@' code.
21828 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21829 UNSPEC_SP_TLS_TEST.
21830 (stack_tls_protect_set_<mode>): Remove.
21831 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21832 (stack_tls_protect_test_<mode>): Remove.
21833 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21835 2017-08-07 Olivier Hainque <hainque@adacore.com>
21838 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21840 2017-08-07 Douglas Rupp <rupp@adacore.com>
21842 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21843 variable was referenced as multidir in command.
21845 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21848 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21851 2017-08-07 Martin Liska <mliska@suse.cz>
21853 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21854 * config/rl78/rl78.c: Add include of attribs.h.
21855 * config/sh/sh.c: Likewise.
21856 * config/v850/v850.c: Likewise.
21858 2017-08-07 Tom de Vries <tom@codesourcery.com>
21860 PR middle-end/78266
21861 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21863 2017-08-07 Martin Liska <mliska@suse.cz>
21865 * config/mips/mips.c: Include attribs.h.
21867 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21870 * doc/invoke.texi: Document change in behvaior for -Ofast for
21873 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21875 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21877 (aarch64_pop_regs): Likewise.
21878 (aarch64_gen_load_pair): Likewise.
21879 (aarch64_save_callee_saves): Likewise.
21880 (aarch64_restore_callee_saves): Likewise.
21882 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21884 * config/i386/i386.c: Revert the last change.
21886 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21889 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21891 (ix86_finalize_stack_frame_flags): This. Also clear
21892 frame_pointer_needed if -fno-omit-frame-pointer is used without
21894 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21895 with ix86_finalize_stack_frame_flags.
21896 (ix86_expand_epilogue): Likewise.
21897 (ix86_expand_split_stack_prologue): Likewise.
21899 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21902 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21903 to P_AES for Westmere.
21905 2017-08-07 Jonathan Yong <10walls@gmail.com>
21907 * config/i386/mingw.opt (fset-stack-executable): Removed.
21908 * config/i386/cygming.opt (fset-stack-executable): Moved
21910 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21912 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21914 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21916 2017-08-07 Marek Polacek <polacek@redhat.com>
21918 PR middle-end/81737
21919 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21921 2017-08-07 Martin Liska <mliska@suse.cz>
21923 * attribs.h (canonicalize_attr_name): New function.
21924 (cmp_attribs): Move from c-format.c and adjusted.
21925 (is_attribute_p): Moved from tree.h.
21926 * tree-inline.c: Add new includes.
21927 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21928 (private_is_attribute_p): Remove.
21929 (private_lookup_attribute): Likewise.
21930 (private_lookup_attribute_by_prefix): Simplify.
21931 (remove_attribute): Use is_attribute_p.
21932 * tree.h: Remove removed declarations.
21934 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21936 PR middle-end/81698
21937 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21938 instead of computing it in the function. Formatting fix.
21939 (expand_case): Don't rely on default_edge being the first edge,
21940 clear it if removing it, pass default_edge to
21941 emit_case_dispatch_table.
21942 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21945 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21947 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21948 insn in the function, emit NOP after the insn.
21950 2017-08-06 Tom de Vries <tom@codesourcery.com>
21952 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21955 2017-08-06 Tom de Vries <tom@codesourcery.com>
21957 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21960 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21962 PR tree-optimization/57371
21963 * match.pd: New pattern.
21965 2017-08-04 Marek Polacek <polacek@redhat.com>
21967 PR middle-end/81695
21968 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21969 perform the computation in offset_int.
21971 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21973 PR tree-optimization/81136
21974 * tree-vectorizer.h: Include tree-hash-traits.h.
21975 (vec_base_alignments): New typedef.
21976 (vec_info): Add a base_alignments field.
21977 (vect_record_base_alignments): Declare.
21978 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21980 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21981 (create_data_ref): Add an is_conditional_in_stmt argument.
21982 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21983 the is_conditional_in_stmt field.
21984 (data_ref_loc): Add an is_conditional_in_stmt field.
21985 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21986 (find_data_references_in_stmt): Update call to create_data_ref.
21987 (graphite_find_data_references_in_stmt): Likewise.
21988 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21989 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21990 (vect_record_base_alignment): New function.
21991 (vect_record_base_alignments): Likewise.
21992 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21993 for nested statements even if we fail to compute a misalignment.
21994 Use pooled base alignments for unconditional references.
21995 (vect_find_same_alignment_drs): Compare base addresses instead
21997 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21998 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22000 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22002 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22003 Add an explicit name for the enum. Use auto_vec for slp_instances
22004 and grouped_stores.
22005 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22007 (_bb_vec_info): Add a constructor and destructor.
22008 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22009 (destroy_loop_vec_info): Delete.
22010 (vect_destroy_datarefs): Likewise.
22011 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22012 (vec_info::vec_info): New function.
22013 (vec_info::~vec_info): Likewise.
22014 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22015 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22016 destroy_loop_vec_info.
22017 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22018 (_loop_vec_info::_loop_vec_info): ...this.
22019 (destroy_loop_vec_info): Replace with...
22020 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22021 the stmt_vec_infos. Leave handling of vec_info information to its
22022 destructor. Remove explicit vector releases.
22023 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22024 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22025 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22026 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22027 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22028 (destroy_bb_vec_info): Replace with...
22029 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22030 information to its destructor.
22031 (vect_slp_analyze_bb_1): Use new and delete instead of
22032 new_bb_vec_info and destroy_bb_vec_info.
22033 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22036 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22038 * tree-data-ref.h (subscript): Add access_fn field.
22039 (data_dependence_relation): Add could_be_independent_p.
22040 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22041 (same_access_functions): Move to tree-data-ref.c.
22042 * tree-data-ref.c (ref_contains_union_access_p): New function.
22043 (access_fn_component_p): Likewise.
22044 (access_fn_components_comparable_p): Likewise.
22045 (dr_analyze_indices): Add a reference to access_fn_component_p.
22046 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22048 (constant_access_functions): Likewise.
22049 (add_other_self_distances): Likewise.
22050 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22051 (initialize_data_dependence_relation): Use XCNEW and remove
22052 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22053 of access functions that have the same type. Allow the
22054 subsequence to end with different bases in some circumstances.
22055 Record the chosen access functions in SUB_ACCESS_FN.
22056 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22057 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22058 (subscript_dependence_tester_1): Likewise dra and drb.
22059 (build_classic_dist_vector): Update calls accordingly.
22060 (subscript_dependence_tester): Likewise.
22061 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22062 DDR_COULD_BE_INDEPENDENT_P.
22063 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22064 comp_alias_ddrs instead of may_alias_ddrs.
22065 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22067 (vect_analyze_data_ref_dependence): Use it if
22068 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22069 distance vectors if that fails.
22070 (dependence_distance_ge_vf): New function.
22071 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22072 LOOP_VINFO_MAY_ALIAS_DDRS.
22074 2017-08-04 Richard Biener <rguenther@suse.de>
22076 PR middle-end/81705
22077 * fold-const.c (fold_binary_loc): Properly restrict
22078 minus_var0 && minus_var1 case when associating undefined overflow
22081 2017-08-04 Tom de Vries <tom@codesourcery.com>
22083 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22085 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22087 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22088 Don't start diagnostic messages with a capital letter.
22089 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22091 (rs6000_invalid_builtin): Likewise.
22092 (rs6000_trampoline_init): Likewise.
22094 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22097 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22098 after setting changeable df flags.
22100 2017-08-03 Richard Biener <rguenther@suse.de>
22102 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22103 up if the use is in USE - X.
22105 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22107 * toplev.c (dumpfile.h): New include.
22108 (internal_error_reentered): New static function. Use it...
22109 (internal_error_function): ...here to handle reentered internal_error.
22111 2017-08-03 Richard Biener <rguenther@suse.de>
22113 PR middle-end/81148
22114 * fold-const.c (split_tree): Add minus_var and minus_con
22115 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22116 here but always use minus_*.
22117 (associate_trees): Assert we never associate with MINUS_EXPR
22118 and NULL first operand. Do not recurse for PLUS_EXPR operands
22119 when associating as MINUS_EXPR either.
22120 (fold_binary_loc): Track minus_var and minus_con.
22122 2017-08-03 Tom de Vries <tom@codesourcery.com>
22125 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22126 ACCEL_COMPILER, apply finish_options on
22127 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22129 2017-08-03 Tom de Vries <tom@codesourcery.com>
22132 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22133 function_entry_patch_area_size > 0.
22135 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22138 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22139 instead of 10??LU, perform unit multiplication in wide_int,
22140 don't change alloc_object_size_limit if the limit is larger
22143 PR tree-optimization/81655
22144 PR tree-optimization/81588
22145 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22146 the case when ranges[i].low and high are 1 for unsigned type with
22149 PR middle-end/81052
22150 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22151 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22153 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22155 * tree-vrp.h: Add include guard.
22157 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22160 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22161 (ud2): New insn pattern.
22162 * config/i386/i386.c (ix86_expand_epilogue):
22163 For naked functions, generate ud2 instead of trap insn.
22165 2017-08-02 Marek Polacek <polacek@redhat.com>
22168 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22170 2017-08-02 Tom de Vries <tom@codesourcery.com>
22171 Cesar Philippidis <cesar@codesourcery.com>
22173 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22174 Add missing edge probabilities.
22176 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22178 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22179 Correct endianness.
22181 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22183 PR middle-end/79499
22184 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22185 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22186 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22187 prologue_seq sequences - if any.
22189 2017-08-02 Richard Biener <rguenther@suse.de>
22191 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22192 via vectors if supported, integer extracts via punning if supported
22193 or otherwise vector extracts.
22195 2017-08-02 Richard Biener <rguenther@suse.de>
22197 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22199 (bitmap_insert_into_set): ... this.
22201 2017-08-02 Richard Biener <rguenther@suse.de>
22203 PR tree-optimization/81633
22205 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22207 PR tree-optimization/71752
22208 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22210 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22212 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22213 (machine_function::call_ms2sysv_pad_out): Remove field.
22214 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22215 (ix86_compute_frame_layout): Likewise.
22217 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22220 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22221 attribute with interrupt attribute.
22223 2017-08-01 Andrew Pinski <apinski@cavium.com>
22225 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22226 BIT_INSERT_EXPR's operand 1
22227 to see if the types precision matches.
22229 2017-08-01 Martin Liska <mliska@suse.cz>
22231 PR middle-end/70140
22232 * builtins.c (expand_builtin_memcpy_args): Remove.
22233 (expand_builtin_memcpy): Call newly added function
22234 expand_builtin_memory_copy_args.
22235 (expand_builtin_memcpy_with_bounds): Likewise.
22236 (expand_builtin_mempcpy): Remove last argument.
22237 (expand_builtin_mempcpy_with_bounds): Likewise.
22238 (expand_builtin_memory_copy_args): New function created from
22239 expand_builtin_mempcpy_args with small modifications.
22240 (expand_builtin_mempcpy_args): Remove.
22241 (expand_builtin_stpcpy): Remove unused argument.
22242 (expand_builtin): Likewise.
22243 (expand_builtin_with_bounds): Likewise.
22245 2017-08-01 Martin Liska <mliska@suse.cz>
22248 Make mempcpy more optimal (PR middle-end/70140).
22250 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22253 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22254 __builtin_vec_cmpne verify both arguments are compatible vectors
22255 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22256 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22257 move computation of aligned to after checking the argument types.
22261 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22262 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22265 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22266 Doug Rupp <rupp@adacore.com>
22267 Olivier Hainque <hainque@adacore.com>
22269 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22270 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22272 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22273 for TARGET_OS_CPP_BUILTIN.
22274 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22275 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22277 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22278 passing required abi options to the assembler for EABI configurations.
22279 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22280 of .text.hot and .text.unlikely sections for kernel modules when
22281 using ARM style exceptions.
22282 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22283 options. Add EXTRA_CC1_SPEC.
22284 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22286 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22288 (ARM_TARGET2_DWARF_FORMAT): Define.
22289 * config/arm/t-vxworks: Adjust multilib control to removal of the
22290 Diab command line options.
22292 2017-08-01 Martin Liska <mliska@suse.cz>
22294 PR gcov-profile/81561
22295 * gcov.c (unblock): Make unblocking safe as we need to preserve
22296 index correspondence of blocks and block_lists.
22298 2017-08-01 Richard Biener <rguenther@suse.de>
22300 PR tree-optimization/81181
22301 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22302 (compute_antic): ... end of iteration here.
22304 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22306 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22307 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22308 (ftree-slp-vectorize): Likewise.
22309 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22310 can no longer be set independent of flag_tree_loop_vectorize.
22311 * omp-general.c (emp_max_vf): Likewise.
22312 * opts.c (enable_fdo_optimizations): Remove references to
22313 flag_tree_vectorize, these are now implicit.
22314 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22315 and leave it for the options machinery.
22317 2017-08-01 Martin Liska <mliska@suse.cz>
22319 PR middle-end/70140
22320 * builtins.c (expand_builtin_memcpy_args): Remove.
22321 (expand_builtin_memcpy): Call newly added function
22322 expand_builtin_memory_copy_args.
22323 (expand_builtin_memcpy_with_bounds): Likewise.
22324 (expand_builtin_mempcpy): Remove last argument.
22325 (expand_builtin_mempcpy_with_bounds): Likewise.
22326 (expand_builtin_memory_copy_args): New function created from
22327 expand_builtin_mempcpy_args with small modifications.
22328 (expand_builtin_mempcpy_args): Remove.
22329 (expand_builtin_stpcpy): Remove unused argument.
22330 (expand_builtin): Likewise.
22331 (expand_builtin_with_bounds): Likewise.
22333 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22336 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22337 print "ds:" only for immediates in generic address space.
22339 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22342 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22343 (ix86_function_ok_for_sibcall): Return false for naked functions.
22345 2017-08-01 Richard Biener <rguenther@suse.de>
22347 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22348 (compute_antic): Seed worklist with exit block predecessors.
22349 * cfganal.c (dfs_find_deadend): For a cycle return the source
22350 of the edge closing it.
22352 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22354 * config/aarch64/aarch64.c
22355 (aarch64_can_const_movi_rtx_p): Move 0 check.
22357 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22359 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22360 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22362 * match.pd: Ditto in address comparison pattern.
22364 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22366 PR tree-optimization/81627
22367 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22368 closed ssa form for store-store chain.
22370 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22372 PR tree-optimization/81620
22373 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22374 for store-store chain.
22376 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22378 PR tree-optimization/81588
22379 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22380 ranges[i].in_p, invert comparison code ccode. For >/>=,
22381 swap rhs1 and rhs2 and comparison code unconditionally,
22382 for </<= don't do that. Don't swap rhs1/rhs2 again if
22383 ranges[i].in_p, instead invert comparison code ccode if
22384 opcode or oe->rank is BIT_IOR_EXPR.
22387 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22388 a direct optab to conversion optab.
22389 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22390 with GET_MODE_INNER as last argument instead of optab_handler.
22391 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22392 vector extraction if possible and optab is available.
22393 * expr.c (store_constructor): Use convert_optab_handler instead
22394 of optab_handler. Use vector initialization from smaller
22395 vectors if possible and optab is available.
22396 * tree-vect-stmts.c (vectorizable_load): Likewise.
22397 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22398 now have two modes.
22399 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22400 of vec_init from half-sized vectors with the same element mode.
22401 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22402 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22403 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22404 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22405 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22406 after mode in gen_vec_extract* calls.
22407 (vec_extract<mode>): Renamed to ...
22408 (vec_extract<mode><ssescalarmodelower>): ... this.
22409 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22410 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22411 element mode after mode in gen_vec_init* calls.
22412 (VEC_INIT_HALF_MODE): New mode iterator.
22413 (vec_init<mode>): Renamed to ...
22414 (vec_init<mode><ssescalarmodelower>): ... this.
22415 (vec_init<mode><ssehalfvecmodelower>): New expander.
22416 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22417 (vec_extractv2sfsf): ... this.
22418 (vec_initv2sf): Renamed to ...
22419 (vec_initv2sfsf): ... this.
22420 (vec_extractv2si): Renamed to ...
22421 (vec_extractv2sisi): ... this.
22422 (vec_initv2si): Renamed to ...
22423 (vec_initv2sisi): ... this.
22424 (vec_extractv4hi): Renamed to ...
22425 (vec_extractv4hihi): ... this.
22426 (vec_initv4hi): Renamed to ...
22427 (vec_initv4hihi): ... this.
22428 (vec_extractv8qi): Renamed to ...
22429 (vec_extractv8qiqi): ... this.
22430 (vec_initv8qi): Renamed to ...
22431 (vec_initv8qiqi): ... this.
22432 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22433 (vec_init<mode>): Renamed to ...
22434 (vec_init<mode><VEC_base_l>): ... this.
22435 (vec_extract<mode>): Renamed to ...
22436 (vec_extract<mode><VEC_base_l>): ... this.
22437 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22438 (vec_initv2sfsf): ... this.
22439 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22440 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22441 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22442 element mode after mode in gen_vec_init* calls.
22443 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22444 (vec_init<mode><Vel>): ... this.
22445 (vec_extract<mode>): Renamed to ...
22446 (vec_extract<mode><Vel>): ... this.
22447 * config/aarch64/iterators.md (Vel): New mode attribute.
22448 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22449 Add element mode after mode in gen_vec_extract* calls.
22450 * config/s390/vector.md (non_vec_l): New mode attribute.
22451 (vec_extract<mode>): Renamed to ...
22452 (vec_extract<mode><non_vec_l>): ... this.
22453 (vec_init<mode>): Renamed to ...
22454 (vec_init<mode><non_vec_l>): ... this.
22455 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22456 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22458 * config/arm/iterators.md (V_elem_l): New mode attribute.
22459 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22460 (vec_extract<mode><V_elem_l>): ... this.
22461 (vec_extractv2di): Renamed to ...
22462 (vec_extractv2didi): ... this.
22463 (vec_init<mode>): Renamed to ...
22464 (vec_init<mode><V_elem_l>): ... this.
22465 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22466 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22467 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22468 Add element mode after gen_vec_extract* calls.
22469 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22470 (vec_init<mode><unitmode>): ... this.
22471 (vec_extract<mode>): Renamed to ...
22472 (vec_extract<mode><unitmode>): ... this.
22473 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22474 (vec_init<mode><unitmode>): ... this.
22475 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22476 (vec_initv2sfsf): ... this.
22477 (vec_extractv2sf): Renamed to ...
22478 (vec_extractv2sfsf): ... this.
22479 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22480 Add element mode after gen_vec_extract* calls.
22481 * config/mips/mips.md (unitmode): New mode iterator.
22482 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22483 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22484 * config/spu/spu.md (inner_l): New mode attribute.
22485 (vec_init<mode>): Renamed to ...
22486 (vec_init<mode><inner_l>): ... this.
22487 (vec_extract<mode>): Renamed to ...
22488 (vec_extract<mode><inner_l>): ... this.
22489 * config/sparc/sparc.md (veltmode): New mode iterator.
22490 (vec_init<VMALL:mode>): Renamed to ...
22491 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22492 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22493 (vec_initv2sisi): ... this.
22494 (vec_initv2sf): Renamed to ...
22495 (vec_initv2sfsf): ... this.
22496 (vec_extractv2sf): Renamed to ...
22497 (vec_extractv2sfsf): ... this.
22498 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22499 (vec_init<mode>): Renamed to ...
22500 (vec_init<mode><VEC_base_l>): ... this.
22501 (vec_extract<mode>): Renamed to ...
22502 (vec_extract<mode><VEC_base_l>): ... this.
22503 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22504 (vec_initv2sfsf): ... this.
22505 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22506 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22507 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22508 gen_vec_init* calls.
22510 2017-08-01 Richard Biener <rguenther@suse.de>
22512 PR tree-optimization/81297
22513 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22514 TREE_OVERFLOW from INTEGER_CSTs.
22516 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22518 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22520 2017-07-31 Carl Love <cel@us.ibm.com>
22522 * config/rs6000/rs6000-c: Add support for built-in functions
22523 vector signed char vec_xl_be (signed long long, signed char *);
22524 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22525 vector signed int vec_xl_be (signed long long, signed int *);
22526 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22527 vector signed long long vec_xl_be (signed long long, signed long long *);
22528 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22529 vector signed short vec_xl_be (signed long long, signed short *);
22530 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22531 vector double vec_xl_be (signed long long, double *);
22532 vector float vec_xl_be (signed long long, float *);
22533 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22534 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22535 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22537 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22538 (altivec_expand_builtin): Add switch statement to call
22539 altivec_expand_xl_be for each builtin.
22540 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22541 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22542 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22543 __builtin_vsx_le_be_v16qi.
22544 * doc/extend.texi: Update the built-in documentation file for the
22545 new built-in functions.
22547 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22550 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22552 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22554 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22556 * config.gcc: Add z14.
22557 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22558 CPU model numbers for z13s and z14.
22559 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22561 * config/s390/s390-opts.h (enum processor_type): Rename
22562 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22563 * config/s390/s390.c (processor_table): Add field for CPU name to
22564 be passed to Binutils.
22565 (s390_asm_output_machine_for_arch): Use the new field in
22566 processor_table for Binutils.
22567 (s390_expand_builtin): Replace arch12 with z14.
22568 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22569 (s390_get_sched_attrmask): Likewise.
22570 (s390_get_unit_mask): Likewise.
22571 * config/s390/s390.opt: Add z14 to processor_type enum.
22573 2017-07-31 Martin Jambor <mjambor@suse.cz>
22576 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22577 regardless of optimization level.
22579 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22580 Martin Liska <mliska@suse.cz>
22582 * predict.def: Remove old comment and adjust probability.
22583 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22584 PREDICT statements.
22586 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22589 * config/i386/i386.c (ix86_function_naked): New function.
22590 (ix86_can_use_return_insn_p): Return false for naked functions.
22591 (ix86_expand_prologue): Skip prologue for naked functions.
22592 (ix86_expand_epilogue): Skip epilogue for naked functions
22593 and emit trap instruction.
22594 (ix86_warn_func_return): New function.
22595 (ix86_attribute_table): Add "naked" attribute specification.
22596 (TARGET_WARN_FUNC_RETURN): Define.
22597 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22599 2017-07-31 Martin Liska <mliska@suse.cz>
22601 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22602 (dump_gimple_bb_header): Always dump BB info.
22603 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22605 2017-07-31 Martin Liska <mliska@suse.cz>
22608 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22609 also with current_function_decl non-null equality.
22611 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22614 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22615 change type to the element type, instead add eltype variable and
22616 use it where we are interested in the element type.
22618 PR tree-optimization/81603
22619 * ipa-polymorphic-call.c
22620 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22621 offset arithmetic in offset_int, bail out if the resulting bit offset
22622 doesn't fit into shwi.
22624 2017-07-31 Martin Liska <mliska@suse.cz>
22626 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22627 (gimplify_save_expr): Fix comment.
22629 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22632 * config/i386/i386.c (ix86_function_arg): Update arguments for
22634 (ix86_compute_frame_layout): Set the initial stack offset to
22635 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22636 INCOMING_FRAME_SP_OFFSET.
22637 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22638 stack before exception handler returns.
22639 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22640 the 'ERROR_CODE' for exception handler.
22642 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22644 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22645 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22646 (ASM_OUTPUT_REG_POP): Ditto.
22647 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22648 instead of asm_fprintf to output pure string.
22650 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22652 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22653 to imported_module_or_decl hook.
22654 (debug_nothing_tree_tree_tree_bool): Remove.
22655 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22656 * debug.c (do_nothing_debug_hooks): Use
22657 debug_nothing_tree_tree_tree_bool_bool instead of
22658 debug_nothing_tree_tree_tree_bool.
22659 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22660 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22661 * sdbout.c (sdb_debug_hooks): Likewise.
22662 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22663 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22665 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22666 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22667 attribute, don't add anything.
22669 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22671 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22672 (fold_build2_stat_loc): Likewise.
22673 (fold_build3_stat_loc): Likewise.
22674 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22675 (fold_build1_loc): Remove macro.
22676 (fold_build2_loc): Likewise.
22677 (fold_build3_loc): Likewise.
22679 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22681 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22682 (gimple_build_debug_bind_source_stat): Likewise.
22683 * gimple.h (gimple_build_debug_bind): Remove macro.
22684 (gimple_build_debug_bind_source): Likewise.
22686 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22688 * bitmap.c (bitmap_alloc): Adjust.
22689 (bitmap_gc_alloc): Likewise.
22690 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22692 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22694 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22695 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22696 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22697 (bitmap_gc_alloc_stat): Likewise.
22698 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22700 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22702 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22703 * rtl.h (shallow_copy_rtx): Remove macro.
22705 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22707 * emit-rtl.c (gen_raw_REG): Adjust.
22708 * gengenrtl.c (gendef): Likewise.
22709 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22710 * rtl.h (rtx_alloc): Remove macro.
22712 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22714 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22715 (build_tree_list_stat): Likewise.
22716 * tree.h (build_tree_list): Remove macro.
22717 (build_tree_list_vec): Likewise.
22719 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22721 * tree.c (make_vector_stat): Remove _stat from name.
22722 (build_vector_stat): Likewise.
22723 * tree.h (make_vector_stat): Remove macro.
22724 (build_vector_stat): Likewise.
22726 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22728 * tree.h (build_var_debug_value): Remove prototype.
22730 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22732 * tree.c (tree_cons_stat): Remove _stat from name.
22733 * tree.h (tree_cons): Remove macro.
22735 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22737 * tree.c (build_vl_exp_stat): Remove _stat from name.
22738 * tree.h (build_vl_exp): Remove macro.
22740 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22742 * tree.c (build_decl_stat): Remove _stat from name.
22743 * tree.h (build_decl): Remove macro.
22745 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22747 * gimple.c (gimple_build_with_ops_stat): Adjust.
22748 (gimple_alloc_stat): Remove _stat from name.
22749 * gimple.h (gimple_alloc): Remove macro.
22751 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22753 * tree.c (make_tree_vec_stat): Remove _stat from name.
22754 (grow_tree_vec_stat): Likewise.
22755 * tree.h (make_tree_vec_stat): Adjust prototype.
22756 (grow_tree_vec_stat): Likewise.
22757 (make_tree_vec): Remove macro.
22758 (grow_tree_vec): Likewise.
22760 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22762 * fold-const.c (fold_build1_stat_loc): Adjust.
22763 (fold_build2_stat_loc): Likewise.
22764 (fold_build3_stat_loc): Likewise.
22765 * tree.c (build0_stat): Remove _stat from name.
22766 (build1_stat): Likewise.
22767 (build2_stat): Likewise.
22768 (build3_stat): Likewise.
22769 (build4_stat): Likewise.
22770 (build5_stat): Likewise.
22771 * tree.h (build1_loc): Remove macro, and rename _stat function
22773 (build2_loc): Likewise.
22774 (build3_loc): Likewise.
22775 (build4_loc): Likewise.
22776 (build5_loc): Likewise.
22778 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22780 * tree.c (make_int_cst_stat): Remove _stat from name.
22781 * tree.h (make_int_cst_stat): Adjust prototype.
22782 (make_int_cst): Remove macro.
22784 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22786 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22787 * tree.h (make_tree_binfo_stat): Adjust prototype.
22788 (make_tree_binfo): Remove.
22790 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22792 * tree.c (copy_node_stat): Rename to copy_node.
22793 (build_distinct_type_copy): Adjust.
22794 * tree.h (copy_node_stat): Adjust prototype.
22795 (copy_node): Remove macro.
22797 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22799 * tree.c (make_node_stat): rename to make_node.
22800 (build_tree_list_stat): Adjust.
22801 (build0_stat): Likewise.
22802 (build2_stat): Likewise.
22803 (build3_stat): Likewise.
22804 (build4_stat): Likewise.
22805 (build5_stat): Likewise.
22806 (build_decl_stat): Likewise.
22807 * tree.h (make_node_stat): Adjust prototype.
22808 (make_node): remove macro.
22810 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22812 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22813 (PPC_FEATURE2_SCV): Likewise.
22814 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22816 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22818 * config/aarch64/aarch64.c
22819 (aarch64_internal_mov_immediate): Add new special pattern.
22820 * config/aarch64/aarch64.md (*movdi_aarch64):
22821 Add reg/32bit const mov case.
22823 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22824 Richard Sandiford <richard.sandiford@linaro.org>
22826 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22827 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22828 Add integer and movi cases.
22829 (movi-split-hf-df-sf split, fp16): New.
22830 (enabled): Added TARGET_FP_F16INST.
22831 * config/aarch64/iterators.md (GPF_HF): New.
22832 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22834 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22836 * config/aarch64/aarch64.c
22837 (aarch64_simd_container_mode): Add prototype.
22838 (aarch64_expand_mov_immediate): Add HI support.
22839 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22840 (aarch64_can_const_movi_rtx_p): New.
22841 (aarch64_preferred_reload_class):
22842 Remove restrictions of using FP registers for certain SIMD operations.
22843 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22844 (aarch64_valid_floating_const): Add integer move validation.
22845 (aarch64_simd_imm_scalar_p): Remove.
22846 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22847 (aarch64_legitimate_constant_p): Expand list of supported cases.
22848 * config/aarch64/aarch64-protos.h
22849 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22850 (aarch64_reinterpret_float_as_int): New.
22851 (aarch64_simd_imm_scalar_p): Remove.
22852 * config/aarch64/constraints.md (Uvi): New.
22853 (Dd): Split into Ds and new Dd.
22854 * config/aarch64/aarch64.md (*movsi_aarch64):
22856 (*movdi_aarch64): Add SIMD mov case.
22858 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22860 * tree-predcom.c: (struct chain): Handle store-store chain in which
22861 stores for elimination only store loop invariant values.
22862 (execute_pred_commoning_chain): Ditto.
22863 (prepare_initializers_chain_store_elim): Ditto.
22864 (prepare_finalizers): Ditto.
22865 (is_inv_store_elimination_chain): New function.
22866 (initialize_root_vars_store_elim_1): New function.
22868 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22870 * tree-predcom.c: Revise general description of the pass.
22871 (enum chain_type): New enum type for store elimination.
22872 (struct chain): New field supporting store elimination.
22873 (struct component): Ditto.
22874 (dump_chain): Dump store-stores chain.
22875 (release_chain): Release resources.
22876 (split_data_refs_to_components): Compute and create component
22877 contains only stores for elimination.
22878 (get_chain_last_ref_at): New function.
22879 (make_invariant_chain): Initialization.
22880 (make_rooted_chain): Specify chain type in parameter and record it.
22881 (add_looparound_copies): Skip for store-stores chain.
22882 (determine_roots_comp): Compute type of chain and pass it to
22884 (initialize_root_vars_store_elim_2): New function.
22885 (finalize_eliminated_stores): New function.
22886 (remove_stmt): Handle store for elimination.
22887 (execute_pred_commoning_chain): Execute predictive commoning on
22888 store-store chains.
22889 (determine_unroll_factor): Skip unroll for store-stores chain.
22890 (prepare_initializers_chain_store_elim): New function.
22891 (prepare_initializers_chain): Hanlde store-store chain.
22892 (prepare_finalizers_chain, prepare_finalizers): New function.
22893 (tree_predictive_commoning_loop): Return integer value indicating
22894 if loop is unrolled or lcssa form is corrupted.
22895 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22897 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22899 * tree-predcom.c (initialize_root): Delete.
22900 (execute_pred_commoning_chain): Initialize root vars and replace
22901 reference of non-combined chain directly, rather than call above
22904 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22906 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22907 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22909 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22911 * tree-predcom.c (struct chain): New field init_seq.
22912 (release_chain): Release init_seq.
22913 (prepare_initializers_chain): Record intialization stmts in above
22915 (insert_init_seqs): New function.
22916 (tree_predictive_commoning_loop): Call insert_init_seqs.
22918 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22920 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22922 2017-07-28 Richard Biener <rguenther@suse.de>
22924 * match.pd: Remove superfluous :c.
22925 * genmatch.c (simplify::id): Add member.
22926 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22928 (current_id): New global.
22929 (dt_node::parent): Move from ...
22930 (dt_operand::parent): ... here. Add for_id member.
22931 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22932 (decision_tree::find_node): Relax order requirement when
22933 merging DT_TRUE nodes to ones inbetween the current simplify
22934 and the one we try to merge with. Add diagnostic whenever
22935 we need to enforce pattern order by not merging.
22936 (decision_tree::insert): Set current_id.
22937 (decision_tree::print_node): Dump parent node and for_id.
22938 (parser::last_id): Add member.
22939 (parser::push_simplify): Assign unique id.
22940 (parser::parser): Initialize last_id.
22942 2017-07-28 Martin Liska <mliska@suse.cz>
22945 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22946 gimple_build_debug_bind.
22948 2017-07-28 Richard Biener <rguenther@suse.de>
22950 PR tree-optimization/81502
22951 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22953 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22955 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22956 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22957 for BIT_FIELD_REF args.
22958 * fold-const.c (make_bit_field_ref): Likewise.
22959 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22961 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22964 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22965 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22966 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22967 Or it into SANITIZER_UNDEFINED.
22968 * ubsan.c: Include gimple-fold.h and varasm.h.
22969 (ubsan_expand_ptr_ifn): New function.
22970 (instrument_pointer_overflow): New function.
22971 (maybe_instrument_pointer_overflow): New function.
22972 (instrument_object_size): Formatting fix.
22973 (pass_ubsan::execute): Call instrument_pointer_overflow
22974 and maybe_instrument_pointer_overflow.
22975 * internal-fn.c (expand_UBSAN_PTR): New function.
22976 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22977 * sanitizer.def (__ubsan_handle_pointer_overflow,
22978 __ubsan_handle_pointer_overflow_abort): New builtins.
22979 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22980 * internal-fn.def (UBSAN_PTR): New internal function.
22981 * opts.c (sanitizer_opts): Add pointer-overflow.
22982 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22983 * fold-const.c (build_range_check): Compute pointer range check in
22984 integral type if pointer arithmetics would be needed. Formatting
22987 2017-07-28 Martin Liska <mliska@suse.cz>
22990 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22991 parameters that are of a variable-length.
22993 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22995 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22997 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22998 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22999 (CRT_CALL_STATIC_FUNCTION): Likewise.
23000 (ASM_DEFAULT_SPEC): New define.
23001 (ASM_SPEC32): Likewise.
23002 (ASM_SPEC64): Likewise.
23003 (ASM_SPEC_COMMON): Likewise.
23004 (ASM_SPEC): Likewise.
23005 (INVALID_64BIT): Likewise.
23006 (LINK_OS_DEFAULT_SPEC): Likewise.
23007 (LINK_OS_SPEC32): Likewise.
23008 (LINK_OS_SPEC64): Likewise.
23009 (POWERPC_LINUX): Likewise.
23010 (PTRDIFF_TYPE): Likewise.
23011 (RESTORE_FP_PREFIX): Likewise.
23012 (RESTORE_FP_SUFFIX): Likewise.
23013 (SAVE_FP_PREFIX): Likewise.
23014 (SAVE_FP_SUFFIX): Likewise.
23015 (SIZE_TYPE): Likewise.
23016 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23017 (TARGET_64BIT): Likewise.
23018 (TARGET_64BIT): Likewise.
23019 (TARGET_AIX): Likewise.
23020 (WCHAR_TYPE_SIZE): Likewise.
23021 (WCHAR_TYPE): Undefine.
23022 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23023 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23024 (CPP_OS_RTEMS_SPEC): Delete.
23025 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23026 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23028 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23030 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23032 PR tree-optimization/81578
23033 * tree-parloops.c (build_new_reduction): Bail out if
23034 reduction_code isn't one of the standard OpenMP reductions.
23035 Move the details printing after that decision.
23037 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23039 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23040 related to reload_in_progress.
23041 (splat_input_operand): Likewise.
23042 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23044 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23046 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23047 (TARGET_INSTANTIATE_DECLS): Likewise.
23048 (legitimate_indexed_address_p): Delete reload_in_progress code.
23049 (rs6000_debug_legitimate_address_p): Likewise.
23050 (rs6000_eliminate_indexed_memrefs): Likewise.
23051 (rs6000_emit_le_vsx_store): Likewise.
23052 (rs6000_emit_move_si_sf_subreg): Likewise.
23053 (rs6000_emit_move): Likewise.
23054 (register_to_reg_type): Likewise.
23055 (rs6000_pre_atomic_barrier): Likewise.
23056 (rs6000_machopic_legitimize_pic_address): Likewise.
23057 (rs6000_allocate_stack_temp): Likewise.
23058 (rs6000_address_for_fpconvert): Likewise.
23059 (rs6000_address_for_altivec): Likewise.
23060 (rs6000_secondary_memory_needed_rtx): Delete function.
23061 (rs6000_check_sdmode): Likewise.
23062 (rs6000_alloc_sdmode_stack_slot): Likewise.
23063 (rs6000_instantiate_decls): Likewise.
23064 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23065 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23066 Delete reload_in_progress.
23067 (*vec_reload_and_plus_<mptrsize>): Likewise.
23068 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23069 (vsx_div_v2di): Likewise.
23070 (vsx_udiv_v2di): Likewise.
23072 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23074 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23075 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23076 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23077 (rs6000_debug_reg_global): Delete print of LRA status.
23078 (rs6000_option_override_internal): Delete dead LRA related code.
23079 (rs6000_lra_p): Delete function.
23080 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23082 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23084 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23085 * config/riscv/rtems.h: New file.
23087 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23088 Sudakshina Das <sudi.das@arm.com>
23090 * config/aarch64/aarch64.md
23091 (define_split for and<mode>3nr_compare): Move
23092 non aarch64_logical_operand to a register.
23093 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23094 register immediate operand to a register.
23095 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23097 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23099 PR middle-end/81564
23100 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23102 2017-07-27 Richard Biener <rguenther@suse.de>
23104 PR tree-optimization/81573
23105 PR tree-optimization/81494
23106 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23107 multi defuse cycle case.
23109 2017-07-27 Richard Biener <rguenther@suse.de>
23111 PR tree-optimization/81571
23112 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23115 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23117 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23118 earlier and only if MASK_FPU is set. Adjust formatting.
23120 2017-07-27 Martin Liska <mliska@suse.cz>
23122 * opt-functions.awk: Add validation of value of Init.
23123 * optc-gen.awk: Pass new argument.
23125 2017-07-27 Martin Liska <mliska@suse.cz>
23127 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23128 Fix wrong condition.
23130 2017-07-27 Martin Liska <mliska@suse.cz>
23132 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23133 BBs and edges seen by autoFDO.
23135 2017-07-27 Richard Biener <rguenther@suse.de>
23137 PR tree-optimization/81502
23138 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23139 with incompatible but same sized type.
23140 (execute_update_addresses_taken): Likewise.
23142 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23144 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23145 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23147 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23150 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23151 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23152 Change s_operand to memory_operand.
23154 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23156 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23157 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23158 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23159 Emit instructions rather than returning an expression. Handle TFmode
23160 and KFmode by casting to TImode.
23161 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23162 (rs6000_emit_le_vsx_store): Likewise.
23163 * config/rs6000/vsx.md (VSX_TI): New iterator.
23164 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23165 (*vsx_le_undo_permute_<mode>): Likewise.
23166 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23167 emit the split sequence.
23168 (*vsx_le_perm_store_<mode>): Likewise.
23170 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23172 PR tree-optimization/81555
23173 PR tree-optimization/81556
23174 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23175 if true, force CHANGED for the recursive invocation.
23176 (reassociate_bb): Remember original length of ops array, pass
23177 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23179 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23180 attributes for noipa attribute. For naked attribute use
23181 lookup_attribute first before lookup_attribute_spec.
23182 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23184 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23185 for functions with noipa attribute.
23186 (cgraph_externally_visible_p): Return true for functions with noipa
23188 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23189 for functions with noipa attribute.
23190 * doc/extend.texi: Document noipa function attribute.
23191 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23192 also for functions with noipa attribute.
23193 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23195 2017-07-26 Andrew Pinski <apinski@cavium.com>
23197 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23198 vec_unalign_load_cost and vec_unalign_store_cost.
23200 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23202 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23203 -mvsx-small-integer option.
23204 (ISA_3_0_MASKS_IEEE): Likewise.
23205 (OTHER_VSX_VECTOR_MASKS): Likewise.
23206 (POWERPC_MASKS): Likewise.
23207 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23208 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23209 code, only testing for DImode being allowed in non-VSX floating
23211 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23212 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23214 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23215 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23216 TARGET_P8_VECTOR test.
23217 (rs6000_secondary_reload_simple_move): Likewise.
23218 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23219 since TARGET_P9_VECTOR was already tested.
23220 (rs6000_opt_masks): Remove -mvsx-small-integer.
23221 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23222 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23224 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23225 test for TARGET_VEXTRACTUB was used, and that uses
23227 (p9 extract splitter): Likewise.
23228 (vsx_extract_<mode>_di_p9): Likewise.
23229 (vsx_extract_<mode>_store_p9): Likewise.
23230 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23231 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23232 the elimination of TARGET_VSX_SMALL_INTEGER.
23233 (vsx_extract_<mode>_p8): Likewise.
23234 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23235 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23236 (vsx_set_<mode>_p9): Likewise.
23237 (vsx_set_v4sf_p9): Likewise.
23238 (vsx_set_v4sf_p9_zero): Likewise.
23239 (vsx_insert_extract_v4sf_p9): Likewise.
23240 (vsx_insert_extract_v4sf_p9_2): Likewise.
23241 * config/rs6000/rs6000.md (sign extend splitter): Change
23242 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23243 (floatsi<mode>2_lfiwax_mem): Likewise.
23244 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23245 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23246 since a test for TARGET_P9_VECTOR was used.
23247 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23248 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23249 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23250 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23251 TARGET_P8_VECTOR test.
23252 (fix_trunc<mode>si2_stfiwx): Likewise.
23253 (fix_trunc<mode>si2_internal): Likewise.
23254 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23255 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23257 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23258 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23259 TARGET_P8_VECTOR test.
23260 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23261 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23262 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23264 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23265 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23266 since a test for TARGET_P9_VECTOR was used.
23267 (splitter for loading small constants): Likewise.
23269 2017-07-26 Andrew Pinski <apinski@cavium.com>
23271 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23274 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23277 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23278 (fp_valid_at): Likewise.
23280 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23282 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23283 (qdf24xx_addrcost_table): Likewise.
23284 (cortexa57_tunings): Update to use generic_branch_cost.
23285 (cortexa72_tunings): Likewise.
23286 (cortexa73_tunings): Likewise.
23287 (qdf24xx_tunings): Likewise.
23289 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23291 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23292 (thunderx2t99_branch_cost): Likewise.
23293 (cortexa35_tunings): Update to use generic_branch_cost.
23294 (cortexa53_tunings): Likewise.
23295 (cortexa57_tunings): Likewise.
23296 (cortexa72_tunings): Likewise.
23297 (cortexa73_tunings): Likewise.
23298 (thunderx2t99_tunings): Likewise.
23300 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23302 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23303 (sparc_option_override): Honour MASK_FSMULD.
23304 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23305 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23306 * config/sparc/sparc.opt (mfsmuld): New option.
23307 * doc/invoke.texi (mfsmuld): Document option.
23309 2017-07-26 Marek Polacek <polacek@redhat.com>
23311 PR middle-end/70992
23312 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23314 2017-07-26 Richard Biener <rguenther@suse.de>
23316 * gimple-match-head.c (do_valueize): Return OP if valueize
23318 (get_def): New helper to get at the def stmt of a SSA name
23319 if valueize allows.
23320 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23321 do_valueize to get at the def stmt.
23322 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23324 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23326 PR middle-end/46932
23327 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23329 2017-07-26 Martin Liska <mliska@suse.cz>
23332 * function.c (expand_function_start): Make expansion of
23333 nonlocal_goto_save_area after parm_birth_insn.
23335 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23337 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23338 from all CPU target flags enable members.
23340 2017-07-26 Richard Biener <rguenther@suse.de>
23342 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23343 (decision_tree::gen): Make 'type' const.
23344 (write_predicate): Likewise.
23346 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23348 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23349 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23350 (rs6000_option_override_internal): Likewise.
23351 (rs6000_expand_vector_set): Likewise.
23352 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23353 (TARGET_UPPER_REGS_SF): Likewise.
23354 (TARGET_UPPER_REGS_DI): Likewise.
23355 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23356 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23357 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23358 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23359 (Splitters for DI constants in Altivec registers): Likewise.
23360 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23361 (vsx_set_v4sf_p9): Likewise.
23362 (vsx_set_v4sf_p9_zero): Likewise.
23363 (vsx_insert_extract_v4sf_p9): Likewise.
23364 (vsx_insert_extract_v4sf_p9_2): Likewise.
23366 2017-07-25 Carl Love <cel@us.ibm.com>
23368 * doc/extend.texi: Update the built-in documentation file for the
23369 existing built-in functions
23370 vector signed char vec_cnttz (vector signed char);
23371 vector unsigned char vec_cnttz (vector unsigned char);
23372 vector signed short vec_cnttz (vector signed short);
23373 vector unsigned short vec_cnttz (vector unsigned short);
23374 vector signed int vec_cnttz (vector signed int);
23375 vector unsigned int vec_cnttz (vector unsigned int);
23376 vector signed long long vec_cnttz (vector signed long long);
23377 vector unsigned long long vec_cnttz (vector unsigned long long);
23379 2017-07-25 Andrew Pinski <apinski@cavium.com>
23381 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23382 accesses where the use is for the first operand of a BIT_INSERT.
23384 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23387 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23388 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23390 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23392 * config/i386/gstabs.h: Delete.
23393 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23395 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23397 * config/i386/i386.c (ix86_decompose_address): Do not check for
23398 register RTX when looking at index_reg or base_reg.
23399 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23401 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23403 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23404 to update EH info here.
23406 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23408 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23410 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23412 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23414 2017-07-25 Torsten Duwe <duwe@suse.de>
23416 * common.opt: Introduce -fpatchable-function-entry
23417 command line option, and its variables function_entry_patch_area_size
23418 and function_entry_patch_area_start.
23419 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23420 including a two-value parser.
23421 * target.def (print_patchable_function_entry): New target hook.
23422 * targhooks.h (default_print_patchable_function_entry): New function.
23423 * targhooks.c (default_print_patchable_function_entry): Likewise.
23424 * toplev.c (process_options): Switch off IPA-RA if
23425 patchable function entries are being generated.
23426 * varasm.c (assemble_start_function): Look at the
23427 patchable-function-entry command line switch and current
23428 function attributes and maybe generate NOP instructions by
23429 calling the print_patchable_function_entry hook.
23430 * doc/extend.texi: Document patchable_function_entry attribute.
23431 * doc/invoke.texi: Document -fpatchable_function_entry
23432 command line option.
23433 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23435 * doc/tm.texi: Re-generate.
23437 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23440 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23441 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23443 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23445 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23447 2017-07-25 Richard Biener <rguenther@suse.de>
23449 PR tree-optimization/81455
23450 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23451 not walk in cycles when looking for guards.
23453 2017-07-25 Richard Biener <rguenther@suse.de>
23455 PR tree-optimization/81529
23456 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23457 when optimizing backedge uses.
23459 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23461 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23463 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23464 to dl_section_ref. On AIX, append an expression to subtract
23465 the size of the section length to dl_section_ref.
23467 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23469 * configure.ac: If any of the config.* scripts fail, exit 1.
23470 * configure: Regenerate.
23472 2017-07-25 Richard Biener <rguenther@suse.de>
23474 PR middle-end/81546
23475 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23476 of immediate uses, be more verbose on errors.
23478 2017-07-25 Richard Biener <rguenther@suse.de>
23480 PR tree-optimization/81510
23481 * tree-vect-loop.c (vect_is_simple_reduction): When the
23482 reduction stmt is not inside the loop bail out.
23484 2017-07-25 Richard Biener <rguenther@suse.de>
23486 PR tree-optimization/81303
23487 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23488 profitability check against LOOP_VINFO_NITERSM1.
23490 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23492 * domwalk.c (cmp_bb_postorder): Simplify.
23493 (sort_bbs_postorder): New function. Use it...
23494 (dom_walker::walk): ...here to optimize common cases.
23496 2017-07-25 Martin Liska <mliska@suse.cz>
23499 * ipa-visibility.c (function_and_variable_visibility): Make the
23500 redirection just on target that supports aliasing.
23501 Fix GNU coding style.
23503 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23506 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23508 * config/arm/rtems.h: Likewise.
23509 * config/bfin/rtems.h: Likewise.
23510 * config/i386/rtemself.h: Likewise.
23511 * config/lm32/rtems.h: Likewise.
23512 * config/m32c/rtems.h: Likewise.
23513 * config/m68k/rtemself.h: Likewise.
23514 * config/microblaze/rtems.h: Likewise.
23515 * config/mips/rtems.h: Likewise.
23516 * config/moxie/rtems.h: Likewise.
23517 * config/nios2/rtems.h: Likewise.
23518 * config/powerpcspe/rtems.h: Likewise.
23519 * config/rs6000/rtems.h: Likewise.
23520 * config/rtems.h: Likewise.
23521 * config/sh/rtems.h: Likewise.
23522 * config/sh/rtemself.h: Likewise.
23523 * config/sparc/rtemself.h: Likewise.
23525 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23528 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23529 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23530 * tree-ssa-structalias.c (alias_get_name): Same.
23532 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23535 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23536 instructions if no du chain is found.
23538 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23540 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23542 2017-07-25 Richard Biener <rguenther@suse.de>
23544 PR middle-end/81505
23545 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23548 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23550 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23551 upper-regs options.
23552 (ISA_2_7_MASKS_SERVER): Likewise.
23553 (ISA_3_0_MASKS_IEEE): Likewise.
23554 (OTHER_P8_VECTOR_MASKS): Likewise.
23555 (OTHER_VSX_VECTOR_MASKS): Likewise.
23556 (POWERPC_MASKS): Likewise.
23557 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23558 duplicate list of options.
23559 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23560 explicit -mupper-regs options.
23561 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23562 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23563 alias for -mupper-regs-df.
23564 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23565 (rs6000_init_hard_regno_mode_ok): Likewise.
23566 (rs6000_option_override_internal): Likewise.
23567 (rs6000_opt_masks): Likewise.
23568 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23569 options in terms of whether -mvsx or -mpower8-vector was used.
23570 (TARGET_UPPER_REGS_DI): Likewise.
23571 (TARGET_UPPER_REGS_SF): Likewise.
23572 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23573 -mupper-regs-* options.
23575 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23577 * passes.c (emergency_dump_function): Print some empty lines and a
23578 header before the RTL dump.
23580 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23582 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23584 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23587 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23588 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23590 2017-07-24 Carl Love <cel@us.ibm.com>
23592 * config/rs6000/rs6000-c.c: Add support for built-in functions
23593 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23594 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23595 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23596 vec_extract_fp_from_shortl): Add defines for the two builtins.
23597 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23598 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23600 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23601 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23602 * doc/extend.texi: Update the built-in documentation file for the
23603 new built-in function.
23605 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23608 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23610 * doc/generic.texi: Likewise.
23611 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23612 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23614 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23616 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23617 (aarch64_mls_elt_merge<mode>): Likewise.
23619 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23621 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23622 having __cxa_atexit.
23624 2017-07-23 Michael Collison <michael.collison@arm.com>
23626 * config/arm/arm.c (arm_option_override): Deprecate
23627 use of -mstructure-size-boundary.
23628 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23629 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23631 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23634 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23635 Reduce cost estimate for direct moves.
23637 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23640 * config/i386/i386.c (ix86_option_override_internal): Disable
23641 BMI, BMI2 and TBM instructions for -m16.
23643 2017-07-21 Carl Love <cel@us.ibm.com>
23645 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23646 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23647 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23648 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23649 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23650 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23651 VMULOSW): New enum "unspec" values.
23652 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23653 altivec_vmulosw): New patterns.
23654 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23655 VMULOSW): Add definitions.
23657 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23659 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23660 (qdf24xx): Likewise.
23661 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23662 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23663 (AARCH64_FL_V8_1): Renumber.
23664 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23665 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23666 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23667 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23668 rdma to feature modifiers list.
23670 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23672 PR middle-end/56727
23673 * ipa-visibility (function_and_variable_visibility): Convert
23674 recursive PLT call to direct call if appropriate.
23676 2017-07-21 Andrew Pinski <apinski@cavium.com>
23678 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23679 operand 1 to see if the types precision matches.
23680 * fold-const.c (operand_equal_p): Likewise.
23682 2017-07-21 Richard Biener <rguenther@suse.de>
23684 PR tree-optimization/81303
23685 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23686 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23687 (vect_peeling_hash_get_lowest_cost): Adjust.
23688 (vect_enhance_data_refs_alignment): Likewise. Use
23689 vect_get_peeling_costs_all_drs to compute the penalty for no
23690 peeling to match up costs.
23692 2017-07-21 Richard Biener <rguenther@suse.de>
23694 PR tree-optimization/81500
23695 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23696 we didn't identify a reduction path.
23698 2017-07-21 Tom de Vries <tom@codesourcery.com>
23699 Cesar Philippidis <cesar@codesourcery.com>
23701 PR gcov-profile/81442
23702 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23705 2017-07-21 Tom de Vries <tom@codesourcery.com>
23708 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23710 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23711 nvptx_override_options_after_change.
23713 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23715 * dwarf2out.c (output_file_names): Avoid double testing for
23716 dwarf_version >= 5.
23718 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23720 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23722 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23724 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23726 (try_crossjump_to_edge): Do not punt on partitioned functions.
23728 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23730 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23731 Put all BBs reachable only via paths crossing cold region to cold
23733 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23735 2016-07-21 Richard Biener <rguenther@suse.de>
23737 PR tree-optimization/81303
23738 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23739 into account prologue and epilogue iterations when raising
23740 min_profitable_iters to sth at least covering one vector iteration.
23742 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23744 * config/arm/arm.c (arm_test_cpu_arch_dat):
23747 2017-07-20 Nathan Sidwell <nathan@acm.org>
23749 Remove TYPE_METHODS.
23750 * tree.h (TYPE_METHODS): Delete.
23751 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23752 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23753 (dbxout_type_methods): Scan TYPE_FIELDS.
23754 (dbxout_type): Don't check TYPE_METHODS here.
23755 * function.c (use_register_for_decl): Always ignore register for
23756 class types when not optimizing.
23757 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23758 * tree.c (free_lang_data_in_type): Stitch out member functions and
23759 templates from TYPE_FIELDS.
23760 (build_distinct_type_copy, verify_type_variant,
23761 verify_type): Member fns are on TYPE_FIELDS.
23762 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23763 * tree-pretty-print.c (dump_generic_node): Likewise.
23765 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23768 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23769 V2TImode and V4TImode.
23770 (ix86_expand_vector_extract): Likewise.
23771 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23772 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23773 (ssescalarmode): Handle V4TImode and V2TImode.
23774 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23775 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23776 (VEXTRACTI128_MODE): New mode iterator.
23777 (splitter for *vec_extractv?ti first element): New.
23778 (VEC_INIT_MODE): New mode iterator.
23779 (vec_init<mode>): Consolidate 3 expanders into one using
23780 VEC_INIT_MODE mode iterator.
23782 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23784 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23785 non_spilled_static_chain_regno_p.
23787 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23789 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23791 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23793 * bb-reorder.c (connect_traces): Allow copying of blocks within
23796 2017-07-20 Richard Biener <rguenther@suse.de>
23798 * gimple.h (gimple_phi_result): Add gphi * overload.
23799 (gimple_phi_result_ptr): Likewise.
23800 (gimple_phi_arg): Likewise. Adjust index assert to only
23801 allow actual argument accesses rather than all slots available
23803 (gimple_phi_arg_def): Add gphi * overload.
23804 * tree-phinodes.c (make_phi_node): Initialize only actual
23806 (resize_phi_node): Clear memory not covered by old node,
23807 do not initialize excess argument slots.
23808 (reserve_phi_args_for_new_edge): Initialize new argument slot
23811 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23813 PR tree-optimization/81388
23815 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23817 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23818 by removing computation of may_be_zero.
23820 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23821 Tom de Vries <tom@codesourcery.com>
23823 PR middle-end/81030
23824 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23825 when gimple level profile disagrees with what RTL expander did.
23827 2017-07-20 Richard Biener <rguenther@suse.de>
23829 PR tree-optimization/61171
23830 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23831 (vect_analyze_stmt): Add slp instance parameter.
23832 (vectorizable_reduction): Likewise.
23833 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23834 (vect_is_simple_reduction): Deal with chains not detected
23835 as SLP reduction chain, specifically not properly associated
23836 chains containing a mix of plus/minus.
23837 (get_reduction_op): Remove.
23838 (get_initial_defs_for_reduction): Simplify, pass in whether
23839 this is a reduction chain, pass in the SLP node for the PHIs.
23840 (vect_create_epilog_for_reduction): Get the SLP instance as
23842 (vectorizable_reduction): Get the SLP instance as arg.
23843 During analysis remember the SLP node with the PHIs in the
23844 instance. Simplify getting at the vectorized reduction PHIs.
23845 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23846 through SLP instance.
23847 (vect_slp_analyze_operations): Likewise.
23848 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23849 (vect_transform_stmt): Likewise.
23851 2017-07-20 Tom de Vries <tom@codesourcery.com>
23853 PR tree-optimization/81489
23854 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23855 read of phi arg location to before loop that modifies phi.
23857 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23859 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23862 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23864 PR middle-end/81331
23865 * except.c (execute): Fix ordering issue.
23867 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23869 PR rtl-optimization/81423
23870 * combine.c (make_compound_operation_int): Don't try to optimize
23871 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23873 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23875 PR rtl-optimization/81423
23876 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23877 with a constant that is -1 in the truncated to mode.
23879 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23881 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23882 (determine_unlikely_bbs): ... here.
23883 * predict.h (propagate_unlikely_bbs_forward): Declare.
23884 * cfgexpand.c (pass_expand::execute): Use it.
23885 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23887 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23888 propagate_unlikely_bbs_forward.
23890 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23892 PR middle-end/81331
23893 * except.c (maybe_add_nop_after_section_switch): New function.
23896 2017-07-19 Tom de Vries <tom@codesourcery.com>
23898 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23900 2017-07-19 Tom de Vries <tom@codesourcery.com>
23902 * gimple.h (gimple_phi_arg): Make assert more strict.
23904 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23906 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23907 * config/rs6000/mmintrin.h: New file.
23908 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23910 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23912 PR tree-optimization/81346
23913 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23915 2017-07-19 Tom de Vries <tom@codesourcery.com>
23917 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23919 2017-07-19 Tom de Vries <tom@codesourcery.com>
23921 * config/nvptx/nvptx-modes.def: Add V2DImode.
23922 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23923 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23924 (nvptx_output_mov_insn): Handle lack of mov.b128.
23925 (nvptx_print_operand): Handle 'H' and 'L' codes.
23926 (nvptx_vector_mode_supported): Allow V2DImode.
23927 (nvptx_preferred_simd_mode): New function.
23928 (nvptx_data_alignment): New function.
23929 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23930 nvptx_preferred_simd_mode.
23931 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23933 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23935 2017-07-19 Tom de Vries <tom@codesourcery.com>
23937 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23938 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23939 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23940 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23941 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23942 (mov<VECIM>_insn): New define_insn.
23943 (define_expand "mov<VECIM>): New define_expand.
23945 2017-07-19 Tom de Vries <tom@codesourcery.com>
23947 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23949 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23951 PR tree-optimization/81346
23952 * fold-const.h (fold_div_compare, range_check_type): Declare.
23953 * fold-const.c (range_check_type): New function.
23954 (build_range_check): Use range_check_type.
23955 (fold_div_compare): No longer static, rewritten into
23956 a match.pd helper function.
23957 (fold_comparison): Don't call fold_div_compare here.
23958 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23959 as helper function.
23961 2017-07-19 Nathan Sidwell <nathan@acm.org>
23963 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23964 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23965 * tree.c (find_decls_types_r, verify_type): Use
23966 TYPE_{MIN,MAX}_VALUE_RAW.
23967 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23968 (hash_tree): Likewise.
23969 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23971 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23974 2017-07-18 Tom de Vries <tom@codesourcery.com>
23976 PR middle-end/81464
23977 * omp-expand.c (expand_omp_for_static_chunk): Handle
23978 equal-argument loop exit phi.
23980 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23983 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23984 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23985 operand 2 predicate.
23986 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23987 operand 2 predicate.
23988 (ror,rol -> rorx splitters): Use const_int_operand as
23989 operand 2 predicate.
23991 2017-06-18 Richard Biener <rguenther@suse.de>
23993 PR tree-optimization/81410
23994 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23995 the gap in the ! slp_perm SLP case after each group.
23997 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23999 PR middle-end/81463
24000 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24003 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24005 PR middle-end/81462
24006 * predict.c (set_even_probabilities): Cleanup; do not affect
24007 probabilities that are already known.
24008 (combine_predictions_for_bb): Call even when count is set.
24010 2017-07-18 Nathan Sidwell <nathan@acm.org>
24012 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24015 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24018 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24019 optimization for loop niter analysis.
24021 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24024 * config/avr/avr.c (avr_optimize_casesi): Don't use
24025 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24027 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24029 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24030 body_cost_vec from _vect_peel_extended_info.
24031 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24032 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24035 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24037 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24039 2017-07-18 Richard Biener <rguenther@suse.de>
24041 PR tree-optimization/80620
24042 PR tree-optimization/81403
24043 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24044 info when re-using a VN table entry.
24046 2017-07-18 Richard Biener <rguenther@suse.de>
24048 PR tree-optimization/81418
24049 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24050 vectype_in. Verify that with lane-reducing reduction operations
24051 we have a single def-use cycle.
24053 2017-07-17 Carl Love <cel@us.ibm.com>
24055 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24057 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24058 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24059 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24060 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24061 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24062 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24063 VMULOSW): New enum "unspec" values.
24064 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24065 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24066 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24067 altivec_vmulosw): New patterns.
24068 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24069 VMULOSW): Add definitions.
24071 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24073 * config/alpha/alpha.c: Include predict.h.
24075 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24077 * tree-vrp.c (compare_assert_loc): Fix comparison function
24078 to return predictable results.
24080 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24082 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24084 (subdi3): Likewise.
24085 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24086 * doc/invoke.texi (mexpand-adddi): Update text.
24088 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24090 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24091 that also clobbers the CC register. The old expand code is moved
24093 (*arc_clzsi2): ... here.
24094 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24095 the CC register. The old expand code is moved to ...
24096 (arc_ctzsi2): ... here.
24098 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24100 * config/arc/arc.opt (mindexed-loads): Use initial value
24101 TARGET_INDEXED_LOADS_DEFAULT.
24102 (mauto-modify-reg): Use initial value
24103 TARGET_AUTO_MODIFY_REG_DEFAULT.
24104 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24105 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24106 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24107 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24109 2017-07-17 Martin Liska <mliska@suse.cz>
24112 * opts.c (finish_options): Do not allow -fgnu-tm
24113 w/ -fsanitize={kernel-,}address. Say sorry.
24115 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24118 * tree-loop-distribution.c (classify_partition): Only assert on
24119 numer of iterations.
24120 (merge_dep_scc_partitions): Delete prameter. Update function call.
24121 (distribute_loop): Remove code handling loop with unknown niters.
24122 (pass_loop_distribution::execute): Skip loop with unknown niters.
24124 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24127 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24128 function sort_partitions_by_post_order.
24130 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24132 PR tree-optimization/81374
24133 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24134 the max index of basic blocks, rather than number of basic blocks.
24136 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24138 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24140 (arc_legitimate_pic_operand_p): Likewise.
24141 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24143 (arc_needs_pcl_p): Likewise.
24144 (arc_legitimate_pc_offset_p): Likewise.
24145 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24146 function is also used in constrains.md.
24147 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24148 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24149 PLUS. Only return true/false in known cases, otherwise assert.
24150 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24151 is already called in arc_legitimate_constant_p.
24152 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24154 (LEGITIMATE_PIC_OPERAND_P): Use
24155 arc_raw_symbolic_reference_mentioned_p function.
24156 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24161 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24162 Andrew Burgess <andrew.burgess@embecosm.com>
24164 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24165 (arc_return_address_register): New function.
24166 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24167 (arc_handle_fndecl_attribute): Add naked attribute.
24168 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24169 (TARGET_WARN_FUNC_RETURN): Likewise.
24170 (arc_allocate_stack_slots_for_args): New function.
24171 (arc_warn_func_return): Likewise.
24172 (machine_function): Change type fn_type.
24173 (arc_compute_function_type): Consider new naked function type,
24174 change function return type.
24175 (arc_must_save_register): Adapt to handle new
24176 arc_compute_function_type's return type.
24177 (arc_expand_prologue): Likewise.
24178 (arc_expand_epilogue): Likewise.
24179 (arc_return_address_regs): Delete.
24180 (arc_return_address_register): New function.
24181 (arc_epilogue_uses): Use above function.
24182 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24183 (arc_function_type): Change encoding, add naked type.
24184 (ARC_INTERRUPT_P): Change to handle the new encoding.
24185 (ARC_FAST_INTERRUPT_P): Likewise.
24186 (ARC_NORMAL_P): Define.
24187 (ARC_NAKED_P): Likewise.
24188 (arc_compute_function_type): Delete prototype.
24189 * config/arc/arc.md (in_ret_delay_slot): Use
24190 arc_return_address_register function.
24191 (simple_return): Likewise.
24192 (p_return_i): Likewise.
24194 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24196 PR tree-optimization/81428
24197 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24198 can't be built for those types.
24200 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24202 Remove stuff dead since r239246.
24204 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24205 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24206 (avr_inform_devices): Remove dead stuff.
24208 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24210 * config/arm/arm_neon.h: Fix softp typo.
24212 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24214 PR tree-optimization/81365
24215 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24216 aggregate moves onto bb predecessor edges, make sure there are no
24217 loads that could alias the lhs in between the start of bb and the
24220 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24223 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24224 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24225 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24227 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24229 PR tree-optimization/81396
24230 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24231 (init_symbolic_number): Initialize it to 1.
24232 (perform_symbolic_merge): Add n_ops from both operands into the new
24234 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24235 without base_addr as useless if they need more than one operation.
24236 (bswap_replace): Handle !bswap case for NULL base_addr.
24238 2017-07-17 Tom de Vries <tom@codesourcery.com>
24241 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24244 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24246 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24247 conditional builtin define __FIX_LEON3FT_B2BST.
24249 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24251 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24252 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24255 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24257 PR rtl-optimization/81424
24258 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24259 to remove potential trapping from operands if -fnon-call-exceptions.
24261 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24263 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24264 profile_proability for scalling.
24265 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24267 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24269 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24271 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24273 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24274 fixpoint arithmetics.
24276 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24278 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24279 fixpoint arithmetics.
24281 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24283 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24284 fixpoint arithmetics.
24286 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24288 * profile-count.h (profile_probability::from_reg_br_prob_note,
24289 profile_probability::to_reg_br_prob_note): New functions.
24290 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24291 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24292 * predict.c (probability_reliable_p): Update.
24293 (edge_probability_reliable_p): Update.
24294 (br_prob_note_reliable_p): Update.
24295 (invert_br_probabilities): Update.
24296 (add_reg_br_prob_note): New function.
24297 (combine_predictions_for_insn): Update.
24298 * asan.c (asan_clear_shadow): Update.
24299 * cfgbuild.c (compute_outgoing_frequencies): Update.
24300 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24301 (update_br_prob_note): Update.
24302 (rtl_verify_edges): Update.
24303 (purge_dead_edges): Update.
24304 (fixup_reorder_chain): Update.
24305 * emit-rtl.c (try_split): Update.
24306 * ifcvt.c (cond_exec_process_insns): Update.
24307 (cond_exec_process_if_block): Update.
24308 (dead_or_predicable): Update.
24309 * internal-fn.c (expand_addsub_overflow): Update.
24310 (expand_neg_overflow): Update.
24311 (expand_mul_overflow): Update.
24312 * loop-doloop.c (doloop_modify): Update.
24313 * loop-unroll.c (compare_and_jump_seq): Update.
24314 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24315 * predict.h: Update.
24316 * reorg.c (mostly_true_jump): Update.
24318 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24319 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24320 * config/arc/arc.c: (emit_unlikely_jump): Update.
24321 * config/arm/arm.c: (emit_unlikely_jump): Update.
24322 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24323 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24324 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24325 (ix86_print_operand): Update.
24326 (ix86_split_fp_branch): Update.
24327 (predict_jump): Update.
24328 * config/ia64/ia64.c (ia64_print_operand): Update.
24329 * config/mmix/mmix.c (mmix_print_operand): Update.
24330 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24331 (rs6000_expand_split_stack_prologue): Update.
24332 * config/rs6000/rs6000.c: Update.
24333 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24334 (s390_expand_vec_movstr): Update.
24335 (s390_expand_cs_tdsi): Update.
24336 (s390_expand_split_stack_prologue): Update.
24337 * config/sh/sh.c (sh_print_operand): Update.
24338 (expand_cbranchsi4): Update.
24339 (expand_cbranchdi4): Update.
24340 * config/sparc/sparc.c (output_v9branch): Update.
24341 * config/spu/spu.c (get_branch_target): Update.
24342 (ea_load_store_inline): Update.
24343 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24344 * config/tilepro/tilepro.c: Update.
24346 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24348 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24349 (gimplify_save_expr): Likewise.
24351 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24353 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24355 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24357 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24359 * timevar.def (TV_IPA_FNSUMMARY): Define.
24361 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24363 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24364 to back store errata sensitive sequence from being generated.
24365 (sqrtdf2_fix): Likewise.
24367 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24369 * tree-ssa-threadupdate.c (compute_path_counts,
24370 update_joiner_offpath_counts): Use profile_probability.
24372 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24375 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24377 * config/arm/arm-c.c (arm_cpu_builtins): Define
24378 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24380 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24382 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24383 array entries to represent __ieee128 versions of the
24384 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24385 scalar_extract_sig, and scalar_insert_exp built-in functions.
24386 (altivec_resolve_overloaded_builtin): Add special case handling
24387 for the __builtin_scalar_insert_exp function, as represented by
24388 the P9V_BUILTIN_VEC_VSIEDP constant.
24389 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24390 exponent support for __ieee128 argument.
24391 (VSESQP): Add scalar extract signature support for __ieee128
24393 (VSTDCNQP): Add scalar test negative support for __ieee128
24395 (VSIEQP): Add scalar insert exponent support for __int128 argument
24396 with __ieee128 result.
24397 (VSIEQPF): Add scalar insert exponent support for __ieee128
24398 argument with __ieee128 result.
24399 (VSTDCQP): Add scalar test data class support for __ieee128
24401 (VSTDCNQP): Add overload support for scalar test negative with
24402 __ieee128 argument.
24403 (VSTDCQP): Add overload support for scalar test data class
24404 __ieee128 argument.
24405 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24406 UNSPEC_VSX_SXSIGDP.
24407 (UNSPEC_VSX_SIEXPQP): New constant.
24408 (xsxexpqp): New insn for VSX scalar extract exponent quad
24410 (xsxsigqp): New insn for VSX scalar extract significand quad
24412 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24413 precision with floating point argument.
24414 (xststdcqp): New expand for VSX scalar test data class quad
24416 (xststdcnegqp): New expand for VSX scalar test negative quad
24418 (xststdcqp): New insn to match expansions for VSX scalar test data
24419 class quad precision and VSX scalar test negative quad precision.
24420 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24421 special case operand checking to enforce that second operand of
24422 VSX scalar test data class with quad precision argument is a 7-bit
24424 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24425 prototypes and descriptions of __ieee128 versions of
24426 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24427 scalar_test_data_class, and scalar_test_neg built-in functions.
24429 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24431 PR tree-optimization/81162
24432 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24433 replace a negate with an add.
24435 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24437 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24439 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24441 * config/arm/arm-c.c (arm_cpu_builtins): Define
24442 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24444 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24446 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24447 (armv8-r): Set ARM Cortex-R52 as default CPU.
24448 * config/arm/arm-tables.opt: Regenerate.
24449 * config/arm/arm-tune.md: Regenerate.
24450 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24452 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24453 extension for -mcpu=cortex-r52.
24455 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24457 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24458 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24459 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24460 (fp-armv8): Define it as FP_ARMv8 only.
24461 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24462 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24464 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24465 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24466 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24467 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24468 than TARGET_FPU_ARMV8.
24469 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24470 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24471 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24473 * config/arm/neon.md (neon_vrint): Likewise.
24474 (neon_vcvt): Likewise.
24475 (neon_<fmaxmin_op><mode>): Likewise.
24476 (<fmaxmin><mode>3): Likewise.
24477 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24478 * config/arm/predicates.md (arm_cond_move_operator): Check against
24479 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24481 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24483 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24484 to top of function.
24486 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24488 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24489 loop in comment with memset.
24491 2017-07-14 Martin Liska <mliska@suse.cz>
24493 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24494 * dwarf2out.c (is_java): Remove the function.
24495 (output_pubname): Remove usage of the function.
24496 (lower_bound_default): Remove usage of DW_LANG_Java.
24497 (gen_compile_unit_die): Likewise.
24498 * gcc.c: Remove compiler defaults for .java and .zip files.
24499 * gimple-expr.c (remove_suffix): Change as there's no longer
24500 extension than 4-letter one.
24501 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24502 (gimplify_save_expr): Likewise.
24503 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24504 as it's possible even for other languages than Java.
24505 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24506 * lto-opts.c (lto_write_options): Remove reference to Java.
24507 * opts.c (strip_off_ending): Update file extension handling.
24508 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24509 * tree-eh.c (lower_resx): Likewise.
24510 * tree.c (free_lang_data_in_type): Remove dead code.
24511 (find_decls_types_r): Likewise.
24512 (build_common_builtin_nodes): Remove Java from a comment.
24513 (verify_type): Remove dead code.
24514 * varasm.c (assemble_external): Remove Java from a comment.
24516 2017-07-14 Martin Liska <mliska@suse.cz>
24518 * opts.c (finish_options): Add quotes.
24519 (common_handle_option): Likewise.
24521 2017-07-14 Martin Liska <mliska@suse.cz>
24523 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24524 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24525 Remove N_SO_PASCAL.
24526 * dwarf2out.c (lower_bound_default): Do not handle
24528 (gen_compile_unit_die): Likewise.
24529 * gcc.c: Remove default extension binding for GNU Pascal.
24530 * stmt.c: Remove Pascal language from a comment.
24531 * xcoffout.c: Likewise.
24533 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24536 * diagnostic-show-locus.c (fixit_cmp): New function.
24537 (layout::layout): Sort m_fixit_hints.
24538 (column_range::column_range): Assert that the values are valid.
24539 (struct char_span): New struct.
24540 (correction::overwrite): New method.
24541 (struct source_line): New struct.
24542 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24543 calls in terms of classes source_line and char_span, and
24544 correction::overwrite.
24545 (selftest::test_overlapped_fixit_printing_2): New function.
24546 (selftest::diagnostic_show_locus_c_tests): Call it.
24548 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24550 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24551 early if there is no lhs.
24553 2017-07-13 Martin Liska <mliska@suse.cz>
24555 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24556 (gen_reference_type_die): Likewise.
24557 * stor-layout.c: Remove Pascal-related comment.
24559 2017-07-13 Martin Liska <mliska@suse.cz>
24561 * opts.c (finish_options): Add quotes to error messages.
24562 (parse_sanitizer_options): Likewise.
24564 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24566 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24568 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24570 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24572 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24574 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24576 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24578 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24581 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24582 provides the hardware capability bits, define the macro
24583 __BUILTIN_CPU_SUPPORTS__.
24584 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24585 if GLIBC does not provide the hardware capability bits. Add a
24586 gcc_unreachable call if the built-in cpu function is neither
24587 __builtin_cpu_is nor __builtin_cpu_supports.
24588 (rs6000_get_function_versions_dispatcher): Change the warning
24589 that an old GLIBC is used which does not export the capability
24590 bits to be an error.
24591 * doc/extend.texi (target_clones attribute): Document the
24592 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24593 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24594 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24595 the macros defined by GCC if the newer GLIBC is available.
24597 2017-07-12 Jeff Law <law@redhat.com>
24599 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24600 remaining includes slightly.
24601 * config/riscv/riscv-builtins.c: Include profile-count.h.
24603 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24606 * config/avr/avr.c (avr_set_current_function): In diagnostic
24607 messages: Quote keywords and (parts of) identifiers.
24608 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24611 2017-07-12 Carl Love <cel@us.ibm.com>
24613 * config/rs6000/rs6000-c.c: Add support for built-in functions
24614 vector bool char vec_revb (vector bool char);
24615 vector bool short vec_revb (vector short char);
24616 vector bool int vec_revb (vector bool int);
24617 vector bool long long vec_revb (vector bool long long);
24618 * doc/extend.texi: Update the built-in documentation file for the
24619 new built-in functions.
24621 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24623 * config/s390/s390.md: Remove movcc splitter.
24625 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24627 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24628 load/store on condition.
24630 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24633 * config/avr/avr.c (avr_encode_section_info)
24634 [progmem && !TREE_READONLY]: Error if progmem object needs
24637 2017-07-11 Michael Collison <michael.collison@arm.com>
24639 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24642 2017-07-11 Carl Love <cel@us.ibm.com>
24644 * config/rs6000/rs6000-c.c: Add support for builtins
24645 vector unsigned int vec_parity_lsbb (vector signed int);
24646 vector unsigned int vec_parity_lsbb (vector unsigned int);
24647 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24648 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24649 vector unsigned long long vec_parity_lsbb (vector signed long long);
24650 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24651 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24652 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24653 * doc/extend.texi: Update the built-in documentation file for the
24654 new built-in functions.
24656 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24658 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24659 (layout::m_primary_loc): New field.
24660 (layout::layout): Initialize new field. Move location filtering
24661 logic from here to...
24662 (layout::maybe_add_location_range): ...this new method. Add
24663 support for filtering to just the lines already specified by other
24665 (layout::will_show_line_p): New method.
24666 (gcc_rich_location::add_location_if_nearby): New method.
24667 (selftest::test_add_location_if_nearby): New test function.
24668 (selftest::diagnostic_show_locus_c_tests): Call it.
24669 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24672 2017-07-11 Tom de Vries <tom@codesourcery.com>
24674 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24675 (bb_first_real_insn): New function.
24676 (nvptx_single): Add extra initialization of broadcasted condition
24679 2017-07-11 Nathan Sidwell <nathan@acm.org>
24681 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24683 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24685 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24686 Binutils doc as TEXI will mess them up.
24687 * doc/invoke.texi (AVR Options): Same here.
24689 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24691 * config/sparc/sparc.opt (mfix-ut700): New option.
24692 (mfix-gr712rc): Likewise.
24693 (sparc_fix_b2bst): New variable.
24694 * doc/invoke.texi (SPARC options): Document them.
24695 (ARM options): Fix warnings.
24696 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24697 instructions to prevent sequences that can trigger the store-store
24698 errata for certain LEON3FT processors.
24699 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24700 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24701 * config/sparc/sparc.md (fix_b2bst): New attribute.
24702 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24704 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24707 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24709 (*rsqrtsf2_sse): Ditto.
24711 (div<mode>3): Macroize insn from divdf3 and divsf3
24712 using MODEF mode iterator.
24714 2017-07-10 Martin Sebor <msebor@redhat.com>
24716 PR tree-optimization/80397
24717 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24718 instead of testing for equality to INTEGER_TYPE.
24720 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24722 * config.gcc: Remove uclibc from arc target spec.
24724 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24726 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24728 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24731 * lto-wrapper.c (remove_option): New function.
24732 (merge_and_complain): Merge PIC/PIE options more realistically.
24734 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24736 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24740 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24741 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24742 * config.in: Regenerate.
24743 * configure: Regenerate.
24744 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24745 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24746 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24747 (TARGET_GASISR_PROLOGUES): ...target mask.
24748 * common/config/avr/avr-common.c
24749 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24750 Set -mgas-isr-prologues.
24751 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24752 INSERT_PASS_BEFORE for it.
24753 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24754 * config/avr/avr.c (avr_option_override)
24755 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24756 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24757 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24758 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24759 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24761 (make_avr_pass_pre_proep): New function.
24762 (emit_push_sfr) <treg>: Add argument to function and use it
24763 instead of TMP_REG.
24764 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24765 and set machine->gasisr.yes.
24766 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24767 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24768 __gcc_isr.n_pushed to .L__stack_usage.
24769 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24770 (avr_asm_final_postscan_insn): ...this new static function.
24771 * config/avr/avr.h (machine_function)
24772 <is_no_gccisr, use_L__stack_usage>: New fields.
24773 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24774 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24775 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24776 (gasisr, *gasisr): New expander and insn.
24777 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24778 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24779 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24781 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24783 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24786 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24788 Move jump-tables out of .text again.
24791 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24792 (ASM_OUTPUT_ADDR_VEC): New function.
24793 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24794 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24795 INSN_ADDRESSes as asm comment.
24796 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24797 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24798 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24799 * config/avr/avr.md (*tablejump): Adjust comment.
24800 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24801 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24803 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24804 (avr_output_addr_vec): New proto.
24805 (avr_log_t) <insn_addresses>: New field.
24807 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24810 * config/i386/i386.c (ix86_function_arg_advance): Set
24811 outgoing_args_on_stack to true if there are outgoing arguments
24813 (ix86_function_arg): Likewise.
24814 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24815 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24816 * config/i386/i386.h (machine_function): Add
24817 outgoing_args_on_stack.
24819 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24821 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24822 supporting pthreds.
24823 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24825 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24827 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24828 (REAL_H): Remove $(MACHMODE_H).
24829 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24831 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24832 $(MACHMODE_H) and double-int.h.
24833 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24835 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24838 2017-07-07 Andrew Pinski <apinski@cavium.com>
24840 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24841 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24843 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24845 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24846 Add warning if GCC was not configured to link against a GLIBC that
24847 exports the hardware capability bits.
24848 (make_resolver_func): Make resolver function private and not a
24849 COMDAT function. Create the name with clone_function_name instead
24850 of make_unique_name.
24853 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24854 correct operand in doing the split.
24856 2017-07-07 Carl Love <cel@us.ibm.com>
24858 * config/rs6000/rs6000-c: Add support for built-in function
24859 vector unsigned short vec_pack_to_short_fp32 (vector float,
24861 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24862 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24863 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24864 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24865 (convert_4f32_8i16): Add define_expand.
24866 * doc/extend.texi: Update the built-in documentation file for the
24867 new built-in function.
24869 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24871 * config/sparc/m8.md: New file.
24872 * config/sparc/sparc.md: Include m8.md.
24874 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24876 * config/sparc/sparc.opt: New option -mvis4b.
24877 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24878 (sparc_option_override): Handle VIS4B.
24879 (enum sparc_builtins): Define
24880 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24881 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24882 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24883 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24884 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24885 (check_constant_argument): New function.
24886 (sparc_vis_init_builtins): Define builtins
24887 __builtin_vis_dictunpack{8,16,32},
24888 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24889 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24890 __builtin_vis_fpcmpde{8,16,32}shl and
24891 __builtin_vis_fpcmpur{8,16,32}shl.
24892 (sparc_expand_builtin): Check that the constant operands to
24893 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24894 constant and in range.
24895 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24897 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24898 (SPARC_IMM5_P): Likewise.
24899 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24900 (enabled): Handle vis4b.
24901 (UNSPEC_DICTUNPACK): New unspec.
24902 (UNSPEC_FPCMPSHL): Likewise.
24903 (UNSPEC_FPUCMPSHL): Likewise.
24904 (UNSPEC_FPCMPDESHL): Likewise.
24905 (UNSPEC_FPCMPURSHL): Likewise.
24906 (cpu_feature): New CPU feature `vis4b'.
24907 (dictunpack{8,16,32}): New insns.
24908 (FPCSMODE): New mode iterator.
24909 (fpcscond): New code iterator.
24910 (fpcsucond): Likewise.
24911 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24912 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24913 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24914 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24915 * config/sparc/constraints.md: Define constraints `q' for unsigned
24916 2-bit integer constants and `t' for unsigned 5-bit integer
24918 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24920 (imm5_operand_dictunpack16): Likewise.
24921 (imm5_operand_dictunpack32): Likewise.
24922 (imm2_operand): Likewise.
24923 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24924 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24925 ditunpack* and fpcmp*shl builtins.
24927 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24929 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24930 * config.in: Add HAVE_AS_SPARC6 define.
24931 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24933 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24935 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24936 (CPP_CPU_SPEC): Handle m8.
24937 (ASM_CPU_SPEC): Likewise.
24938 * config/sparc/sparc-opts.h (enum processor_type): Add
24940 * config/sparc/sparc.c (m8_costs): New struct.
24941 (sparc_option_override): Handle TARGET_CPU_m8.
24942 (sparc32_initialize_trampoline): Likewise.
24943 (sparc64_initialize_trampoline): Likewise.
24944 (sparc_issue_rate): Likewise.
24945 (sparc_register_move_cost): Likewise.
24946 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24947 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24948 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24949 (CPP_CPU_SPEC): Handle M8.
24950 (ASM_CPU_SPEC): Likewise.
24951 (AS_M8_FLAG): Define.
24952 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24953 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24954 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24956 * configure: Regenerate.
24957 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24960 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24962 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24964 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24965 ("*movdi_insn_sp32"): Do not set v3pipe.
24966 ("*movsi_insn"): Likewise.
24967 ("*movdi_insn_sp64"): Likewise.
24968 ("*movsf_insn"): Likewise.
24969 ("*movdf_insn_sp32"): Likewise.
24970 ("*movdf_insn_sp64"): Likewise.
24971 ("*zero_extendsidi2_insn_sp64"): Likewise.
24972 ("*sign_extendsidi2_insn"): Likewise.
24973 ("*mov<VM32:mode>_insn"): Likewise.
24974 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24975 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24976 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24977 ("<vlop:code><VL:mode>3"): Likewise.
24978 ("*not_<vlop:code><VL:mode>3"): Likewise.
24979 ("*nand<VL:mode>_vis"): Likewise.
24980 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24981 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24982 ("one_cmpl<VL:mode>2"): Likewise.
24983 ("faligndata<VM64:mode>_vis"): Likewise.
24984 ("alignaddrsi_vis"): Likewise.
24985 ("alignaddrdi_vis"): Likweise.
24986 ("alignaddrlsi_vis"): Likewise.
24987 ("alignaddrldi_vis"): Likewise.
24988 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24989 ("bmaskdi_vis"): Likewise.
24990 ("bmasksi_vis"): Likewise.
24991 ("bshuffle<VM64:mode>_vis"): Likewise.
24992 ("cmask8<P:mode>_vis"): Likewise.
24993 ("cmask16<P:mode>_vis"): Likewise.
24994 ("cmask32<P:mode>_vis"): Likewise.
24995 ("pdistn<P:mode>_vis"): Likewise.
24996 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24998 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25000 * config/sparc/sparc.md ("subtype"): New insn attribute.
25001 ("*wrgsr_sp64"): Set insn subtype.
25002 ("*rdgsr_sp64"): Likewise.
25003 ("alignaddrsi_vis"): Likewise.
25004 ("alignaddrdi_vis"): Likewise.
25005 ("alignaddrlsi_vis"): Likewise.
25006 ("alignaddrldi_vis"): Likewise.
25007 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25008 ("fexpand_vis"): Likewise.
25009 ("fpmerge_vis"): Likewise.
25010 ("faligndata<VM64:mode>_vis"): Likewise.
25011 ("bshuffle<VM64:mode>_vis"): Likewise.
25012 ("cmask8<P:mode>_vis"): Likewise.
25013 ("cmask16<P:mode>_vis"): Likewise.
25014 ("cmask32<P:mode>_vis"): Likewise.
25015 ("fchksm16_vis"): Likewise.
25016 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25017 ("fmean16_vis"): Likewise.
25018 ("fp<plusminus_insn>64_vis"): Likewise.
25019 ("<plusminus_insn>v8qi3"): Likewise.
25020 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25021 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25022 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25023 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25024 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25025 ("*movqi_insn"): Likewise.
25026 ("*movhi_insn"): Likewise.
25027 ("*movsi_insn"): Likewise.
25028 ("movsi_pic_gotdata_op"): Likewise.
25029 ("*movdi_insn_sp32"): Likewise.
25030 ("*movdi_insn_sp64"): Likewise.
25031 ("movdi_pic_gotdata_op"): Likewise.
25032 ("*movsf_insn"): Likewise.
25033 ("*movdf_insn_sp32"): Likewise.
25034 ("*movdf_insn_sp64"): Likewise.
25035 ("*zero_extendhisi2_insn"): Likewise.
25036 ("*zero_extendqihi2_insn"): Likewise.
25037 ("*zero_extendqisi2_insn"): Likewise.
25038 ("*zero_extendqidi2_insn"): Likewise.
25039 ("*zero_extendhidi2_insn"): Likewise.
25040 ("*zero_extendsidi2_insn_sp64"): Likewise.
25041 ("ldfsr"): Likewise.
25042 ("prefetch_64"): Likewise.
25043 ("prefetch_32"): Likewise.
25044 ("tie_ld32"): Likewise.
25045 ("tie_ld64"): Likewise.
25046 ("*tldo_ldub_sp32"): Likewise.
25047 ("*tldo_ldub1_sp32"): Likewise.
25048 ("*tldo_ldub2_sp32"): Likewise.
25049 ("*tldo_ldub_sp64"): Likewise.
25050 ("*tldo_ldub1_sp64"): Likewise.
25051 ("*tldo_ldub2_sp64"): Likewise.
25052 ("*tldo_ldub3_sp64"): Likewise.
25053 ("*tldo_lduh_sp32"): Likewise.
25054 ("*tldo_lduh1_sp32"): Likewise.
25055 ("*tldo_lduh_sp64"): Likewise.
25056 ("*tldo_lduh1_sp64"): Likewise.
25057 ("*tldo_lduh2_sp64"): Likewise.
25058 ("*tldo_lduw_sp32"): Likewise.
25059 ("*tldo_lduw_sp64"): Likewise.
25060 ("*tldo_lduw1_sp64"): Likewise.
25061 ("*tldo_ldx_sp64"): Likewise.
25062 ("*mov<VM32:mode>_insn"): Likewise.
25063 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25064 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25066 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25068 * config/sparc/sparc.md ("type"): New insn type viscmp.
25069 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25071 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25072 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25073 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25074 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25076 ("n7_vis_logical_11cycle"): Likewise.
25077 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25078 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25079 * config/sparc/niagara.md ("niag_vis"): Likewise.
25080 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25081 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25083 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25085 * config/sparc/sparc.md: New instruction type `bmask'.
25086 (bmaskdi_vis): Use the `bmask' type.
25087 (bmasksi_vis): Likewise.
25088 * config/sparc/ultra3.md (us3_array): Likewise.
25089 * config/sparc/niagara7.md (n7_array): Likewise.
25090 * config/sparc/niagara4.md (n4_array): Likewise.
25091 * config/sparc/niagara2.md (niag2_vis): Likewise.
25092 (niag3_vis): Likewise.
25093 * config/sparc/niagara.md (niag_vis): Likewise.
25095 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25097 * ipa-comdats.c: Remove optimize check from gate.
25098 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25099 for functions not optimized.
25100 (ipa_fn_summary_read): Skip optimize check.
25101 (ipa_fn_summary_write): Likewise.
25102 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25104 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25106 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25107 (check_callers): Give up on caller which is not optimized.
25108 (inline_small_functions): Likewise.
25109 (ipa_inline): Do not give up when not optimizing.
25110 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25111 away unoptimizes cdtors.
25112 (whole_program_function_and_variable_visibility): Do
25113 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25114 * ipa.c (process_references): Do not check optimize.
25115 (symbol_table::remove_unreachable_nodes): Update optimize check.
25116 (set_writeonly_bit): Update optimize check.
25117 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25118 (pass_ipa_single_use::gate): Remove.
25120 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25122 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25123 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25124 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25125 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25126 permute_load, permute_store, adjust_extract, adjust_splat,
25127 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25128 replace_swap_with_copy, dump_swap_insn_table,
25129 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25130 recombine_lvx_pattern, recombine_stvx_pattern,
25131 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25132 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25133 to file rs6000-p8swap.c.
25134 * config/rs6000/rs6000-p8swap.c: New file.
25135 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25136 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25137 and rs6000*-*-* targets.
25139 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25141 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25143 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25145 * lto-wrapper.c (merge_and_complain): Do not merge
25146 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25147 fsigned_zeros, ftrapping_math, fwrapv.
25148 (append_compiler_options): Do not track these options.
25149 (append_linker_options): Likewie
25151 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25153 * cgraphunit.c (cgraph_node::finalize_function): When
25154 !flag_toplevel_reorde set no_reorder flag.
25155 (varpool_node::finalize_decl): Likewise.
25156 (symbol_table::compile): Drop no toplevel reorder path.
25158 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25160 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25161 edges; zero probability is not better than uninitialized.
25163 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25165 * asan.h (asan_sanitize_allocas_p): Declare.
25166 * asan.c (asan_sanitize_allocas_p): New function.
25167 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25168 (handle_builtin_alloca): Likewise.
25169 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25170 if !asan_sanitize_allocas_p.
25171 * params.def (asan-instrument-allocas): Add new option.
25172 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25173 * opts.c (common_handle_option): Disable allocas sanitization for
25176 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25178 * asan.c: Include gimple-fold.h.
25179 (get_last_alloca_addr): New function.
25180 (handle_builtin_stackrestore): Likewise.
25181 (handle_builtin_alloca): Likewise.
25182 (asan_emit_allocas_unpoison): Likewise.
25183 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25184 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25185 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25186 (instrument_builtin_call): Pass gimple iterator to
25187 get_mem_refs_of_builtin_call.
25188 (last_alloca_addr): New global.
25189 * asan.h (asan_emit_allocas_unpoison): Declare.
25190 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25191 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25192 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25193 if function calls alloca.
25194 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25195 * gimple-fold.h (replace_call_with_value): Declare.
25196 * internal-fn.c: Include asan.h.
25197 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25198 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25200 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25202 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25203 (C_SELFTEST_FLAGS): New.
25204 (CPP_SELFTEST_FLAGS): New.
25205 (SELFTEST_DEPS): New, from deps of s-selftest.
25206 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25207 (CPP_SELFTEST_DEPS): New.
25208 (selftest): Add dependency on s-selftest-c++.
25209 (s-selftest): Rename to...
25210 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25211 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25212 than SELFTEST_FLAGS.
25213 (selftest-gdb): Rename to...
25214 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25216 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25217 (selftest-valgrind): Rename to...
25218 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25220 (selftest-valgrind): Reintroduce as an alias for
25221 selftest-c-valgrind.
25222 (s-selftest-c++): New.
25223 (selftest-c++-gdb): New.
25224 (selftest-c++-valgrind): New.
25226 2017-07-06 Olivier Hainque <hainque@adacore.com>
25228 * gcc.c (process_command): When deciding if undefined variables
25229 should be ignored when processing specs, accept "gcc -v" as well.
25231 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25233 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25234 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25236 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25238 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25239 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25240 * config/arm/arm-tables.opt: Regenerate.
25241 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25243 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25245 2017-07-06 Carl Love <cel@us.ibm.com>
25247 * ChangeLog: Clean up from mid air collision
25249 2017-07-06 Carl Love <cel@us.ibm.com>
25251 * config/rs6000/rs6000-c.c: Add support for built-in functions
25252 vector signed int vec_subc (vector signed int, vector signed int);
25253 vector signed __int128 vec_subc (vector signed __int128,
25254 vector signed __int128);
25255 vector unsigned __int128 vec_subc (vector unsigned __int128,
25256 vector unsigned __int128);
25257 vector signed int vec_sube (vector signed int, vector signed int,
25258 vector signed int);
25259 vector unsigned int vec_sube (vector unsigned int,
25260 vector unsigned int,
25261 vector unsigned int);
25262 vector signed __int128 vec_sube (vector signed __int128,
25263 vector signed __int128,
25264 vector signed__int128);
25265 vector unsigned __int128 vec_sube (vector unsigned __int128,
25266 vector unsigned __int128,
25267 vector unsigned __int128);
25268 vector signed int vec_subec (vector signed int, vector signed int,
25269 vector signed int);
25270 vector unsigned int vec_subec (vector unsigned int,
25271 vector unsigned int,
25272 vector unsigned int);
25273 vector signed __int128 vec_subec (vector signed __int128,
25274 vector signed __int128,
25275 vector signed__int128);
25276 vector unsigned __int128 vec_subec (vector unsigned __int128,
25277 vector unsigned __int128,
25278 vector unsigned __int128);
25279 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25280 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25281 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25282 BU_ALTIVEC_OVERLOAD_X definitions.
25283 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25284 * doc/extend.texi: Update the built-in documentation file for the new
25285 built-in functions.
25287 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25290 * diagnostic-show-locus.c (layout::layout): Use start and finish
25291 spelling location for the start and finish of each range.
25292 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25293 Add unused aspect param.
25294 * input.c (expand_location_1): Add "aspect" param, and use it
25295 to access the correct part of the location.
25296 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25298 (expand_location_to_spelling_point): Likewise.
25299 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25300 param, and pass it to expand_location_1.
25302 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25304 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25305 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25306 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25307 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25308 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25309 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25310 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25311 _mm_maskz_getmant_ss): New intrinsics.
25312 (__builtin_ia32_getexpss128_mask): Changed to ...
25313 __builtin_ia32_getexpss128_round ... this.
25314 (__builtin_ia32_getexpsd128_mask): Changed to ...
25315 __builtin_ia32_getexpsd128_round ... this.
25316 * config/i386/i386-builtin-types.def
25317 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25318 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25319 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25320 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25321 __builtin_ia32_getmantss_mask_round): New builtins.
25322 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25323 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25324 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25325 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25326 * config/i386/sse.md
25327 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25328 avx512f_sgetexp<mode><mask_scalar_name>
25329 <round_saeonly_scalar_name> ... this.
25330 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25331 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25332 vgetexp<ssescalarmodesuffix>
25333 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25334 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25335 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25336 avx512f_vgetmant<mode><mask_scalar_name>
25337 <round_saeonly_scalar_name> ... this.
25338 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25339 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25340 vgetmant<ssescalarmodesuffix>
25341 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25342 %0<mask_scalar_operand4>, %1, %2
25343 <round_saeonly_scalar_mask_op4>, %3} ... this.
25344 * config/i386/subst.md (mask_scalar_operand4,
25345 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25346 round_saeonly_scalar_nimm_predicate): New subst attributes.
25348 2017-07-06 Julia Koval <julia.koval@intel.com>
25350 * config/i386/i386.c (ix86_erase_embedded_rounding):
25351 Remove code for old rounding pattern.
25353 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25355 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25357 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25359 * doc/sourcebuild.texi (Test Directives, Variants of
25360 dg-require-support): Add documentation for dg-require-stack-check.
25362 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25364 * config/i386/subst.md (mask_scalar, round_scalar,
25365 round_saeonly_scalar): New meta-templates.
25366 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25367 round_scalar_mask_operand3, round_scalar_mask_op3,
25368 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25369 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25370 round_saeonly_scalar_constraint,
25371 round_saeonly_scalar_prefix): New subst attribute.
25372 * config/i386/sse.md
25373 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25374 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25375 <round_scalar_name> ... this.
25376 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25377 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25378 <round_scalar_name> ... this.
25379 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25380 <sse>_vm<code><mode>3<mask_scalar_name>
25381 <round_saeonly_scalar_name> ... this.
25382 (v<plusminus_mnemonic><ssescalarmodesuffix>
25383 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25384 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25385 v<plusminus_mnemonic><ssescalarmodesuffix>
25386 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25387 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25388 (v<multdiv_mnemonic><ssescalarmodesuffix>
25389 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25390 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25391 v<multdiv_mnemonic><ssescalarmodesuffix>
25392 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25393 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25394 (v<maxmin_float><ssescalarmodesuffix>
25395 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25396 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25397 v<maxmin_float><ssescalarmodesuffix>
25398 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25399 %0<mask_scalar_operand3>, %1, %<iptr>2
25400 <round_saeonly_scalar_mask_op3>} ... this.
25402 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25404 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25405 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25407 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25408 Alan Hayward <alan.hayward@arm.com>
25409 David Sherwood <david.sherwood@arm.com>
25411 * combine.c (simplify_if_then_else): Remove "enum" before
25413 * compare-elim.c (can_eliminate_compare): Likewise.
25414 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25416 (aarch64_lookup_simd_builtin_type): Likewise.
25417 (aarch64_simd_builtin_type): Likewise.
25418 (aarch64_init_simd_builtin_types): Likewise.
25419 (aarch64_simd_expand_args): Likewise.
25420 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25422 (aarch64_reverse_mask): Likewise.
25423 (aarch64_simd_emit_reg_reg_move): Likewise.
25424 (aarch64_gen_adjusted_ldpstp): Likewise.
25425 (aarch64_ccmp_mode_to_code): Likewise.
25426 (aarch64_operands_ok_for_ldpstp): Likewise.
25427 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25428 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25430 (aarch64_min_divisions_for_recip_mul): Likewise.
25431 (aarch64_reassociation_width): Likewise.
25432 (aarch64_get_condition_code_1): Likewise.
25433 (aarch64_simd_emit_reg_reg_move): Likewise.
25434 (aarch64_simd_attr_length_rglist): Likewise.
25435 (aarch64_reverse_mask): Likewise.
25436 (aarch64_operands_ok_for_ldpstp): Likewise.
25437 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25438 (aarch64_gen_adjusted_ldpstp): Likewise.
25439 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25441 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25442 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25443 (arm_lookup_simd_builtin_type): Likewise.
25444 (arm_simd_builtin_type): Likewise.
25445 (arm_init_simd_builtin_types): Likewise.
25446 (arm_expand_builtin_args): Likewise.
25447 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25448 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25449 (ft32_setup_incoming_varargs): Likewise.
25450 (ft32_function_arg): Likewise.
25451 (ft32_function_arg_advance): Likewise.
25452 (ft32_pass_by_reference): Likewise.
25453 (ft32_arg_partial_bytes): Likewise.
25454 (ft32_valid_pointer_mode): Likewise.
25455 (ft32_addr_space_pointer_mode): Likewise.
25456 (ft32_addr_space_legitimate_address_p): Likewise.
25457 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25459 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25460 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25461 (iamcu_alignment): Likewise.
25462 (canonicalize_vector_int_perm): Likewise.
25463 (ix86_noce_conversion_profitable_p): Likewise.
25464 (ix86_mpx_bound_mode): Likewise.
25465 (ix86_operands_ok_for_move_multiple): Likewise.
25466 * config/microblaze/microblaze-protos.h
25467 (microblaze_expand_conditional_branch_reg): Likewise.
25468 * config/microblaze/microblaze.c
25469 (microblaze_expand_conditional_branch_reg): Likewise.
25470 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25472 (rs6000_reassociation_width): Likewise.
25473 (rs6000_invalid_binary_op): Likewise.
25474 (fusion_p9_p): Likewise.
25475 (emit_fusion_p9_load): Likewise.
25476 (emit_fusion_p9_store): Likewise.
25477 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25479 (riscv_hard_regno_mode_ok_p): Likewise.
25480 (riscv_address_insns): Likewise.
25481 (riscv_split_symbol): Likewise.
25482 (riscv_legitimize_move): Likewise.
25483 (riscv_function_value): Likewise.
25484 (riscv_hard_regno_nregs): Likewise.
25485 (riscv_expand_builtin): Likewise.
25486 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25487 (riscv_build_integer): Likewise.
25488 (riscv_split_integer): Likewise.
25489 (riscv_legitimate_constant_p): Likewise.
25490 (riscv_cannot_force_const_mem): Likewise.
25491 (riscv_regno_mode_ok_for_base_p): Likewise.
25492 (riscv_valid_base_register_p): Likewise.
25493 (riscv_valid_offset_p): Likewise.
25494 (riscv_valid_lo_sum_p): Likewise.
25495 (riscv_classify_address): Likewise.
25496 (riscv_legitimate_address_p): Likewise.
25497 (riscv_address_insns): Likewise.
25498 (riscv_load_store_insns): Likewise.
25499 (riscv_force_binary): Likewise.
25500 (riscv_split_symbol): Likewise.
25501 (riscv_force_address): Likewise.
25502 (riscv_legitimize_address): Likewise.
25503 (riscv_move_integer): Likewise.
25504 (riscv_legitimize_const_move): Likewise.
25505 (riscv_legitimize_move): Likewise.
25506 (riscv_address_cost): Likewise.
25507 (riscv_subword): Likewise.
25508 (riscv_output_move): Likewise.
25509 (riscv_canonicalize_int_order_test): Likewise.
25510 (riscv_emit_int_order_test): Likewise.
25511 (riscv_function_arg_boundary): Likewise.
25512 (riscv_pass_mode_in_fpr_p): Likewise.
25513 (riscv_pass_fpr_single): Likewise.
25514 (riscv_pass_fpr_pair): Likewise.
25515 (riscv_get_arg_info): Likewise.
25516 (riscv_function_arg): Likewise.
25517 (riscv_function_arg_advance): Likewise.
25518 (riscv_arg_partial_bytes): Likewise.
25519 (riscv_function_value): Likewise.
25520 (riscv_pass_by_reference): Likewise.
25521 (riscv_setup_incoming_varargs): Likewise.
25522 (riscv_print_operand): Likewise.
25523 (riscv_elf_select_rtx_section): Likewise.
25524 (riscv_save_restore_reg): Likewise.
25525 (riscv_for_each_saved_reg): Likewise.
25526 (riscv_register_move_cost): Likewise.
25527 (riscv_hard_regno_mode_ok_p): Likewise.
25528 (riscv_hard_regno_nregs): Likewise.
25529 (riscv_class_max_nregs): Likewise.
25530 (riscv_memory_move_cost): Likewise.
25531 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25532 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25533 (rl78_addr_space_address_mode): Likewise.
25534 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25536 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25537 (rs6000_reassociation_width): Likewise.
25538 (rs6000_invalid_binary_op): Likewise.
25539 (fusion_p9_p): Likewise.
25540 (emit_fusion_p9_load): Likewise.
25541 (emit_fusion_p9_store): Likewise.
25542 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25543 (ok_for_simple_move_operands): Likewise.
25544 (ok_for_simple_move_strict_operands): Likewise.
25545 (ok_for_simple_arith_logic_operands): Likewise.
25546 (visium_legitimize_reload_address): Likewise.
25547 (visium_select_cc_mode): Likewise.
25548 (output_cbranch): Likewise.
25549 (visium_split_double_move): Likewise.
25550 (visium_expand_copysign): Likewise.
25551 (visium_expand_int_cstore): Likewise.
25552 (visium_expand_fp_cstore): Likewise.
25553 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25554 (visium_function_arg): Likewise.
25555 (visium_function_arg_advance): Likewise.
25556 (visium_libcall_value): Likewise.
25557 (visium_setup_incoming_varargs): Likewise.
25558 (visium_legitimate_constant_p): Likewise.
25559 (visium_legitimate_address_p): Likewise.
25560 (visium_legitimize_address): Likewise.
25561 (visium_secondary_reload): Likewise.
25562 (visium_register_move_cost): Likewise.
25563 (visium_memory_move_cost): Likewise.
25564 (prepare_move_operands): Likewise.
25565 (ok_for_simple_move_operands): Likewise.
25566 (ok_for_simple_move_strict_operands): Likewise.
25567 (ok_for_simple_arith_logic_operands): Likewise.
25568 (visium_function_value_1): Likewise.
25569 (rtx_ok_for_offset_p): Likewise.
25570 (visium_legitimize_reload_address): Likewise.
25571 (visium_split_double_move): Likewise.
25572 (visium_expand_copysign): Likewise.
25573 (visium_expand_int_cstore): Likewise.
25574 (visium_expand_fp_cstore): Likewise.
25575 (visium_split_cstore): Likewise.
25576 (visium_select_cc_mode): Likewise.
25577 (visium_split_cbranch): Likewise.
25578 (output_cbranch): Likewise.
25579 (visium_print_operand_address): Likewise.
25580 * expmed.c (flip_storage_order): Likewise.
25581 * expmed.h (emit_cstore): Likewise.
25582 (flip_storage_order): Likewise.
25583 * genrecog.c (validate_pattern): Likewise.
25584 * hsa-gen.c (gen_hsa_addr): Likewise.
25585 * internal-fn.c (expand_arith_overflow): Likewise.
25586 * ira-color.c (allocno_copy_cost_saving): Likewise.
25587 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25588 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25589 (process_invariant_for_inheritance): Likewise.
25590 * lra-eliminations.c (move_plus_up): Likewise.
25591 * omp-low.c (lower_oacc_reductions): Likewise.
25592 * simplify-rtx.c (simplify_subreg): Likewise.
25593 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25594 (TARGET_CHKP_BOUND_MODE): Likewise..
25595 * targhooks.c (default_chkp_bound_mode): Likewise.
25596 (default_setup_incoming_vararg_bounds): Likewise.
25597 * targhooks.h (default_chkp_bound_mode): Likewise.
25598 (default_setup_incoming_vararg_bounds): Likewise.
25599 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25600 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25601 (have_whole_vector_shift): Likewise.
25602 * tree-vect-stmts.c (vectorizable_load): Likewise.
25603 * doc/tm.texi: Regenerate.
25605 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25607 Graceful degrade if Binutils PR21472 is not available.
25610 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25611 .rodata in flash test fails.
25612 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25613 * confgure: Regenerate.
25614 * config.in: Regenerate.
25615 * config/avr/avr.c (avr_asm_named_section)
25616 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25617 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25618 (avr_asm_init_sections): Same.
25620 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25622 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25623 (fma<VH:mode>4_intrinsic): Likewise.
25624 (*fmsub<VCVTF:mode>4): Likewise.
25625 (*fmsub<VH:mode>4_intrinsic): Likewise.
25627 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25630 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25631 Don't depend on "optimize > 0".
25632 (out_movhi_r_mr, out_movqi_mr_r): Same.
25633 (out_movhi_mr_r, out_movqi_r_mr): Same.
25634 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25635 io_address_operand on "optimize > 0".
25637 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25639 * tree-loop-distribution.c: Add general explanantion on the pass.
25640 (generate_loops_for_partition): Mark distributed loop.
25641 (pg_add_dependence_edges): New parameter. Handle alias data
25642 dependence specially and record it in the parameter if asked.
25643 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25644 (init_partition_graph_vertices, add_partition_graph_edge): New.
25645 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25646 (free_partition_graph_vdata, build_partition_graph): New.
25647 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25648 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25649 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25650 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25651 (version_for_distribution_p, finalize_partitions): New.
25652 (distribute_loop): Handle alias data dependence specially. Factor
25653 out loop fusion code as functions and call these functions.
25655 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25657 * tree-loop-distribution.c (classify_partition): New parameter and
25658 better handle reduction statement.
25659 (rdg_build_partitions): Revise comment.
25660 (distribute_loop): Compute statements in all partitions and pass it
25661 to classify_partition.
25663 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25665 * tree-loop-distribution.c (enum partition_type): New.
25666 (struct partition): New field type.
25667 (partition_merge_into): Add parameter. Update partition type.
25668 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25669 (build_rdg_partition_for_vertex): Compute partition type.
25670 (rdg_build_partitions): Dump partition type.
25671 (distribute_loop): Update calls to partition_merge_into.
25673 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25675 * tree-loop-distribution.c (struct ddr_hasher): New.
25676 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25678 (classify_partition): Call get_data_dependence.
25679 (pg_add_dependence_edges): Ditto.
25680 (distribute_loop): Release data dependence hash table.
25682 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25684 * tree-loop-distribution.c (ref_base_address): Delete.
25685 (similar_memory_accesses): Rename ...
25686 (share_memory_accesses): ... to this. Check if partitions access
25687 the same memory reference.
25688 (distribute_loop): Call share_memory_accesses.
25690 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25692 * tree-loop-distribution.c (struct partition): New field recording
25693 its data reference.
25694 (partition_alloc, partition_free): Init and release data refs.
25695 (partition_merge_into): Merge data refs.
25696 (build_rdg_partition_for_vertex): Collect data refs for partition.
25697 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25699 (distribute_loop): Remve data refs from vertice data of partition
25702 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25704 * tree-loop-distribution.c (params.h): Include header file.
25705 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25706 (datarefs_vec): New global var.
25707 (create_rdg_vertices): Use datarefs_vec directly.
25708 (free_rdg): Don't free data references.
25709 (build_rdg): Update use. Don't free data references.
25710 (distribute_loop): Compute global variable for data references.
25711 Bail out if there are too many data references.
25713 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25715 * tree-loop-distribution.c (loop_nest): New global var.
25716 (build_rdg): Use loop directly, rather than loop nest.
25717 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25719 (distribute_loop): Compute global variable loop nest. Update use.
25721 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25723 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25724 (partition_merge_into): New parameter. Dump reason for fusion.
25725 (distribute_loop): Update use of partition_merge_into.
25727 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25729 * tree-loop-distribution.c (bb_top_order_index): New.
25730 (bb_top_order_index_size, bb_top_order_cmp): New.
25731 (stmts_from_loop): Use topological order.
25732 (pass_loop_distribution::execute): Compute and release topological
25733 order for basic blocks.
25735 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25737 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25740 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25742 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25743 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25744 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25745 * internal-fn.def (LOOP_DIST_ALIAS): New.
25746 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25747 (fold_loop_internal_call): ... this.
25748 (vect_loop_dist_alias_call): New function.
25749 (set_uid_loop_bbs): Call fold_loop_internal_call.
25750 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25753 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25756 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25757 Require dead FLAGS_REG at the beginning of a peephole.
25759 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25762 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25763 arguments in the call to __builtin_ia32_sbb_u32.
25764 (_subborrow_u64): Swap _X and _Y arguments in the call to
25765 __builtin_ia32_sbb_u64.
25767 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25770 * tree-vrp.c (compare_assert_loc): Turn into a function template
25771 with stable template parameter. Only test if a->e is NULL,
25772 !a->e == !b->e has been verified already. Use e == NULL or
25773 e != NULL instead of e or ! e tests. If stable is true, don't use
25774 iterative_hash_expr, on the other side allow a or b or both NULL
25775 and sort the NULLs last.
25776 (process_assert_insertions): Sort using compare_assert_loc<false>
25777 instead of compare_assert_loc, later sort using
25778 compare_assert_loc<true> before calling process_assert_insertions_for
25779 in a loop. Use break instead of continue once seen NULL pointer.
25781 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25783 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25784 Cortex-R7 and Cortex-R8 processors.
25786 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25788 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25789 uninitialized while src is not.
25791 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25793 * common/config/arm/arm-common.c: Adjust include path for
25795 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25796 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25797 (arm-cpu-data.h): Likewise.
25798 (arm-cpu-cdata.h): Likewise.
25799 * config/arm/arm-cpu.h: Delete.
25800 * config/arm/arm-cpu-cdata.h: Delete.
25801 * config/arm/arm-cpu-data.h: Delete.
25803 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25805 * config/arm/arm-cpus.in (cortex-a55): New.
25806 (cortex-a75): Likewise.
25807 (cortex-a75.cortex-a55): Likewise.
25808 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25810 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25811 * config/arm/arm-cpu-cdata.h: Regenerate.
25812 * config/arm/arm-cpu-data.h: Regenerate.
25813 * config/arm/arm-cpu.h: Regenerate.
25814 * config/arm/arm-tables.opt: Regenerate.
25815 * config/arm/arm-tune.md: Regenerate.
25817 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25819 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25821 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25823 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25826 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25828 PR tree-optimization/81292
25829 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25830 full_string_p, also call adjust_related_strinfos if the adjustment
25831 is simple, otherwise invalidate related strinfos.
25833 2017-07-04 Martin Liska <mliska@suse.cz>
25836 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25837 newly created variable as DECL_IGNORED_P.
25839 2017-07-04 Martin Liska <mliska@suse.cz>
25842 * ipa-inline.c (inline_small_functions):
25843 Use xstrdup_for_dump.
25845 2017-07-04 Tom de Vries <tom@codesourcery.com>
25847 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25849 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25852 * config/darwin.c (darwin_function_switched_text_sections):
25855 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25857 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25859 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25861 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25863 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25865 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25866 min_profitable_iters, and th as inclusive lower bounds.
25867 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25868 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25869 for min_profitable_iters and min_profitable_estimate.
25870 (vect_transform_loop): Treat th as an inclusive lower bound.
25871 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25873 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25876 * config/darwin.c (darwin_function_switched_text_sections):
25877 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25878 in two pieces, and suppress the use of buf.
25880 2017-07-03 Nathan Sidwell <nathan@acm.org>
25882 * hash-table.h (hash_table_mod1): Fix indentation.
25884 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25886 PR middle-end/81290
25887 * predict.c (force_edge_cold): Be more careful about propagation
25889 * profile-count.h (profile_probability::guessed,
25890 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25892 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25894 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25896 * doc/invoke.texi (rcpc architecture extension): Document it.
25898 2017-07-03 Richard Biener <rguenther@suse.de>
25900 PR tree-optimization/60510
25901 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25902 the scalar reduction PHI and use it.
25903 (vectorizable_reduction): Properly guard the single_defuse_cycle
25904 path for non-SLP reduction chains where we cannot use it.
25905 Rework reduc_def/index and vector type deduction. Rework
25906 vector operand gathering during reduction op code-gen.
25907 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25908 chains dissolve the chain and leave it to non-SLP reduction
25911 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25913 * tree-data-ref.h (dr_alignment): Declare.
25914 * tree-data-ref.c (dr_alignment): New function.
25915 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25916 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25918 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25920 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25922 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25923 and base_misalignment fields.
25924 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25925 * tree-data-ref.c: Include builtins.h.
25926 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25927 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25928 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25929 * tree-vect-data-refs.c: Include tree-cfg.h.
25930 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25931 fields instead of calculating an alignment here.
25932 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25933 innermost_loop_behavior fields.
25935 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25937 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25939 (DR_STEP_ALIGNMENT): New macro.
25940 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25941 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25942 (create_data_ref): Print it.
25943 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25944 to tell whether the step preserves vector (mis)alignment.
25945 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25946 Move the check for an integer step and generalise to all INTEGER_CST.
25947 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25948 Print the outer step alignment.
25950 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25952 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25953 with offset_alignment.
25954 (DR_ALIGNED_TO): Delete.
25955 (DR_OFFSET_ALIGNMENT): New macro.
25956 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25957 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25958 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25959 (create_data_ref): Likewise.
25960 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25961 (vect_analyze_data_refs): Likewise.
25962 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25963 creating dummy innermost behavior.
25965 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25967 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25968 with a "innermost_loop_behavior *" and refeence tree.
25969 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25970 (create_data_ref): Update call accordingly.
25971 * tree-predcom.c (find_looparound_phi): Likewise.
25973 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25975 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25976 fields with dr_wrt_vec_loop.
25977 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25978 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25979 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25980 (vect_dr_behavior): New function.
25981 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25982 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25983 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25984 track whether the step preserves the misalignment.
25985 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25986 Use vect_dr_behavior.
25987 (vect_setup_realignment): Update call accordingly.
25988 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25989 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25990 call to vect_create_addr_base_for_vector_ref.
25991 (vect_create_cond_for_align_checks): Likewise.
25992 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25993 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25994 (vect_recog_mask_conversion_pattern): Likewise.
25995 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25996 (new_stmt_vec_info): Remove redundant zeroing.
25998 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26000 * common/config/arm/arm-common.c (arm_be8_option): New function.
26001 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26002 (ISA_ARMv6): Add isa_bit_be8.
26003 * config/arm/arm.h (arm_be8_option): Add prototype.
26004 (BE8_SPEC_FUNCTION): New define.
26005 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26006 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26007 (mlittle-endian): Similarly.
26008 (mbe8, mbe32): New options.
26009 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26010 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26012 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26014 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26016 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26018 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26019 (cleanup_tree_cfg_bb): Use it.
26020 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26022 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26024 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26027 * loop-doloop.c (add_test): Update profile.
26029 2017-07-03 Martin Liska <mliska@suse.cz>
26032 * sanopt.c (rewrite_usage_of_param): New function.
26033 (sanitize_rewrite_addressable_params): Likewise.
26034 (pass_sanopt::execute): Call rewrite_usage_of_param.
26036 2017-07-03 Richard Biener <rguenther@suse.de>
26038 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26039 back to using VIEW_CONVERT_EXPR.
26041 2017-07-03 Martin Liska <mliska@suse.cz>
26044 * doc/extend.texi: Document when a resolver function is
26045 generated for target_clones.
26047 2017-07-03 Martin Liska <mliska@suse.cz>
26049 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26050 and shadow memory of auto variables which are subject of
26051 use-after-scope sanitization.
26052 (asan_expand_mark_ifn): Add do set only when is_poison.
26054 2016-07-03 Richard Biener <rguenther@suse.de>
26056 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26058 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26059 (vectorizable_reduction): Perform reduction PHI creation when
26060 visiting a reduction PHI and adjust and simplify code generation
26061 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26062 (vect_transform_loop): Visit reduction PHIs.
26063 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26064 defs into the SLP tree.
26065 (vect_build_slp_tree): Reduction defs terminate the recursion.
26066 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26068 (vect_get_vec_defs_for_stmt_copy): Export.
26069 (vect_get_vec_defs): Likewise.
26070 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26072 (vect_get_vec_defs_for_stmt_copy): Declare.
26073 (vect_get_vec_defs): Likewise.
26075 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26077 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26078 parameter with a "loop" parameter and use it instead of the
26079 loop containing DR_STMT. Don't check simple_iv when doing
26080 BB analysis. Describe the two analysis modes in the comment.
26082 2017-07-03 Tom de Vries <tom@codesourcery.com>
26084 PR tree-optimization/69468
26085 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26086 (find_same_succ_bb): Handle ignore_edge_flags.
26088 2017-07-03 Tom de Vries <tom@codesourcery.com>
26090 PR tree-optimization/81192
26091 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26093 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26095 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26097 2017-07-03 Tom de Vries <tom@codesourcery.com>
26099 PR tree-optimization/81192
26100 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26101 BB_SAME_SUCC (bb) == NULL.
26103 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26105 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26108 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26110 * dumpfile.c: Include profile-count.h
26111 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26113 (insert_cond_bb): Update profile.
26114 * tree-cfg.h (insert_cond_bb): Update prototype.
26115 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26116 * tree-dump.c: Do not include tree-cfg.
26118 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26120 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26122 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26124 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26127 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26129 * tree-complex.c (expand_complex_div_wide): update profile.
26131 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26132 Alan Hayward <alan.hayward@arm.com>
26133 David Sherwood <david.sherwood@arm.com>
26135 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26136 (CORETYPES_H): New define.
26137 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26138 (insn-modes-inline.h, s-modes-inline-h): New rules.
26139 (generated_files): Add insn-modes-inline.h.
26140 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26141 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26142 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26143 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26144 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26145 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26146 (build/gencodes.o, build/genconditions.o): Likewise.
26147 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26148 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26149 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26150 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26151 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26152 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26153 * coretypes.h: Include everything up to real.h for generators.
26154 Include insn-modes.h first. Include wide-int-print.h after
26155 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26156 * machmode.h: Don't include insn-modes.h here.
26157 * function-tests.c: Remove includes of signop.h, machmode.h,
26158 double-int.h and wide-int.h.
26160 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26162 * optc-save-gen.awk: Likewise.
26163 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26164 * godump.c: Remove include of wide-int-print.h.
26165 * pretty-print.h: Likewise.
26166 * wide-int-print.cc: Likewise.
26167 * wide-int.cc: Likewise.
26168 * hash-map-tests.c: Remove include of signop.h.
26169 * hash-set-tests.c: Likewise.
26170 * rtl-tests.c: Likewise.
26171 * mkconfig.sh: Remove include of machmode.h.
26172 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26174 (emit_insn_modes_inline_h): ...this new function. Emit the code
26175 into an insn-modes-inline.h header file, adding appropriate
26176 include guards and end comments.
26177 (emit_insn_modes_c_header): Remove include of machmode.h.
26178 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26180 (main): Handle -i flag and call emit_insn_modes_inline_h when
26183 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26185 * tree-ssa-strlen.c (strinfo): Rename the length field to
26186 nonzero_chars. Add a full_string_p field.
26187 (compare_nonzero_chars, zero_length_string_p): New functions.
26188 (get_addr_stridx): Add an offset_out parameter.
26189 Use compare_nonzero_chars.
26190 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26191 (new_strinfo): Update after above changes to strinfo.
26192 (set_endptr_and_length): Set full_string_p.
26193 (get_string_length): Update after above changes to strinfo.
26194 (unshare_strinfo): Update call to new_strinfo.
26195 (maybe_invalidate): Likewise.
26196 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26197 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26198 as a uhwi instead of an shwi. Update after above changes to
26199 strinfo and new_strinfo.
26200 (zero_length_string): Assert that chainsi contains full strings.
26201 Use zero_length_string_p. Update call to new_strinfo.
26202 (adjust_related_strinfos): Update after above changes to strinfo.
26203 Copy full_string_p from origsi.
26204 (adjust_last_stmt): Use zero_length_string_p.
26205 (handle_builtin_strlen): Update after above changes to strinfo and
26206 new_strinfo. Install the lhs as the string length if the previous
26207 entry didn't describe a full string.
26208 (handle_builtin_strchr): Update after above changes to strinfo
26210 (handle_builtin_strcpy): Likewise.
26211 (handle_builtin_strcat): Likewise.
26212 (handle_builtin_malloc): Likewise.
26213 (handle_pointer_plus): Likewise.
26214 (handle_builtin_memcpy): Likewise. Track nonzero characters
26215 that aren't necessarily followed by a nul terminator.
26216 (handle_char_store): Likewise.
26218 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26220 PR tree-optimization/80769
26221 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26222 for malloc and calloc. Document the new invariant that all related
26223 strinfos have delayed lengths or none do.
26224 (verify_related_strinfos): Move earlier in file.
26225 (set_endptr_and_length): New function, split out from...
26226 (get_string_length): ...here. Also set the lengths of related
26228 (zero_length_string): Assert that chainsi has known (rather than
26230 (adjust_related_strinfos): Likewise.
26232 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26234 PR tree-optimization/81136
26235 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26236 assert that two references with the same misalignment have the same
26237 compile-time misalignment if those compile-time misalignments
26240 2017-07-01 Andi Kleen <ak@linux.intel.com>
26242 * print-tree.c (print_node): Print all attributes.
26244 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26246 * cfg.c (scale_bbs_frequencies): New function.
26247 * cfg.h (scale_bbs_frequencies): Declare it.
26248 * cfgloopanal.c (single_likely_exit): Cleanup.
26249 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26251 (scale_loop_profile): Likewise.
26252 (loop_version): Likewise.
26253 (create_empty_loop_on_edge): Update.
26254 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26255 scale_loop_frequencies, scale_loop_profile, loopify,
26256 loop_version): Update prototypes.
26257 * modulo-sched.c (sms_schedule): Update.
26258 * predict.c (unlikely_executed_edge_p): Also check probability.
26259 (probably_never_executed_edge_p): Fix typo.
26260 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26261 * tree-parloops.c (gen_parallel_loop): Update.
26262 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26263 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26264 * tree-ssa-loop-split.c (split_loop): Update.
26265 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26266 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26267 (vect_loop_versioning): Update.
26268 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26270 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26272 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26274 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26276 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26277 to keep profile consistent.
26279 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26281 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26282 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26283 * profile-count.h (max_safe_multiplier): Make unsigned.
26284 (profile_count::guessed_zero): New.
26286 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26288 * bb-reorder.c (fix_up_crossing_landing_pad,
26289 fix_crossing_conditional_branches): Use make_single_succ_edge
26290 to keep profile consistent.
26292 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26294 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26297 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26300 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26301 the right scopes, make sure cond_jump isn't preserved between multiple
26302 iterations. Search for fallthru edge whenever there are 3+ edges and
26303 use find_fallthru_edge for it.
26305 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26307 Patch by Alexander Monakov <amonakov@ispras.ru>
26308 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26309 probabilities consistently.
26311 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26313 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26314 emit_cmp_and_jump_insns.
26316 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26319 * tree-inline.c (expand_call_inline): Combine profile statuses.
26321 2017-06-30 Andrew Pinski <apinski@cavium.com>
26323 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26324 fold_stmt returned true.
26326 2017-06-30 Nathan Sidwell <nathan@acm.org>
26328 * ggc.h (empty_string): Delete.
26329 * cfgexpand.c (expand_asm_stmt): Use plain "".
26330 * optabs.c (expand_asm_memory_barrier): Likewise.
26331 * stringpool.c (empty_string): Delete.
26332 (digit_vector, digit_string): Delete.
26333 (ggc_alloc_string): Use plain "", don't optimize single digit
26334 strings. Use ggc_alloc_atomic.
26336 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26338 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26339 comparison set and one other set, use the cost of the non-comparison
26342 2017-06-30 Nathan Sidwell <nathan@acm.org>
26344 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26347 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26349 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26350 loops. Remove now unneeded calls to gimple_switch_set_label() that
26351 just set removed labels to NULL_TREE.
26353 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26355 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26356 (set_range_info): ...here. Only call set_range_info_raw if domain
26358 (set_nonzero_bits): Call set_range_info_raw.
26359 * tree-ssanames.h (set_range_info_raw): New.
26361 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26364 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26365 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26366 of nonimmediate_operand and <store_mask_constraint> instead of m
26367 for the input operand. For V8FI iterator, always split if input
26368 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26369 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26370 <store_mask_predicate> instead of register_operand and
26371 <store_mask_constraint> instead of v for the input operand. Make
26372 sure both operands aren't MEMs for if not <mask_applied>.
26374 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26376 * lto-wrapper.c (copy_file) Close both file descriptors before
26379 2017-06-30 Martin Liska <mliska@suse.cz>
26382 * multiple_target.c (create_dispatcher_calls): Make ifunc
26383 also for function that don't have calls or are not referenced.
26385 2017-06-30 Richard Biener <rguenther@suse.de>
26387 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26388 analyze the first scalar stmt. Move vector type computation
26389 for the BB case here from ...
26390 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26391 live operation processing in the SLP case properly.
26393 2017-06-30 Richard Biener <rguenther@suse.de>
26395 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26397 2017-06-30 Martin Liska <mliska@suse.cz>
26400 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26401 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26403 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26405 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26408 2017-06-30 Richard Biener <rguenther@suse.de>
26410 PR tree-optimization/81249
26411 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26412 condition reduction result to original scalar type.
26414 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26416 * profile-count.h (enum profile_quality): Fix typos and whitespace
26419 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26421 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26422 type for branch probabilities.
26424 2017-06-29 Julian Brown <julian@codesourcery.com>
26425 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26427 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26428 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26429 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26430 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26432 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26434 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26435 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26436 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26437 CC usage from generic code to here.
26438 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26439 CC usage into the target macros.
26441 2017-06-29 Maya Rashish <coypu@sdf.org>
26443 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26446 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26448 * arm/arm-builtins.c: Include profile-count.h
26449 * except.c (sjlj_emit_function_enter): Use
26450 profile_probability::unlikely.
26452 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26454 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26455 and tocrel_offset be pointer args rather than implicitly using
26457 (legitimate_constant_pool_address_p, rs6000_emit_move,
26458 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26459 tocrel_offset and use in toc_relative_expr_p call.
26460 (print_operand, print_operand_address): Use static tocrel_base_oac
26461 and tocrel_offset_oac.
26462 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26465 2017-06-29 Maya Rashish <coypu@sdf.org>
26467 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26469 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26471 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26472 objects, take into account only the alignment of 'op0' and 'mode1' if
26475 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26477 * ccmp.c (ccmp_tree_comparison_p): New function.
26478 (ccmp_candidate_p): Update to use above function.
26479 (get_compare_parts): New function.
26480 (expand_ccmp_next): Update to use new functions.
26481 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26483 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26484 take mode as argument.
26485 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26486 * expr.c (expand_expr_real_1): Pass mode as argument.
26488 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26490 * combine.c (combine_instructions): Print insns to dump_file, together
26493 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26495 * asan.c (asan_emit_stack_protection): Update.
26496 (create_cond_insert_point): Update.
26497 * auto-profile.c (afdo_propagate_circuit): Update.
26498 * basic-block.h (struct edge_def): Turn probability to
26499 profile_probability.
26500 (EDGE_FREQUENCY): Update.
26501 * bb-reorder.c (find_traces_1_round): Update.
26502 (better_edge_p): Update.
26503 (sanitize_hot_paths): Update.
26504 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26505 (make_single_succ_edge): Update.
26506 (check_bb_profile): Update.
26507 (dump_edge_info): Update.
26508 (update_bb_profile_for_threading): Update.
26509 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26510 probabilitycount to 0.
26511 * cfgbuild.c (compute_outgoing_frequencies): Update.
26512 * cfgcleanup.c (try_forward_edges): Update.
26513 (outgoing_edges_match): Update.
26514 (try_crossjump_to_edge): Update.
26515 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26516 (expand_gimple_tailcall): Update.
26517 (construct_init_block): Use make_single_succ_edge.
26518 (construct_exit_block): Use make_single_succ_edge.
26519 * cfghooks.c (verify_flow_info): Update.
26520 (redirect_edge_succ_nodup): Update.
26521 (split_edge): Update.
26522 (account_profile_record): Update.
26523 * cfgloopanal.c (single_likely_exit): Update.
26524 * cfgloopmanip.c (scale_loop_profile): Update.
26525 (set_zero_probability): Remove.
26526 (duplicate_loop_to_header_edge): Update.
26527 * cfgloopmanip.h (loop_version): Update prototype.
26528 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26529 (force_nonfallthru_and_redirect): Update.
26530 (update_br_prob_note): Update.
26531 (rtl_verify_edges): Update.
26532 (purge_dead_edges): Update.
26533 (rtl_lv_add_condition_to_bb): Update.
26534 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26535 * cgraphunit.c (init_lowered_empty_function): Update.
26536 (cgraph_node::expand_thunk): Update.
26537 * cilk-common.c: Include profile-count.h
26538 * dojump.c (inv): Remove.
26539 (jumpifnot): Update.
26540 (jumpifnot_1): Update.
26541 (do_jump_1): Update.
26543 (do_jump_by_parts_greater_rtx): Update.
26544 (do_compare_rtx_and_jump): Update.
26545 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26546 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26547 * dwarf2cfi.c: Include profile-count.h
26548 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26549 (sjlj_emit_dispatch_table): Likewise.
26550 * explow.c: Include profile-count.h
26551 * expmed.c (emit_store_flag_force): Update.
26552 (do_cmp_and_jump): Update.
26553 * expr.c (compare_by_pieces_d::generate): Update.
26554 (compare_by_pieces_d::finish_mode): Update.
26555 (emit_block_move_via_loop): Update.
26556 (store_expr_with_bounds): Update.
26557 (store_constructor): Update.
26558 (expand_expr_real_2): Update.
26559 (expand_expr_real_1): Update.
26560 * expr.h (try_casesi, try_tablejump): Update prototypes.
26561 * gimple-pretty-print.c (dump_probability): Update.
26562 (dump_profile): New.
26563 (dump_gimple_label): Update.
26564 (dump_gimple_bb_header): Update.
26565 * graph.c (draw_cfg_node_succ_edges): Update.
26566 * hsa-gen.c (convert_switch_statements): Update.
26567 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26568 (find_if_case_1): Update.
26569 (find_if_case_2): Update.
26570 * internal-fn.c (expand_arith_overflow_result_store): Update.
26571 (expand_addsub_overflow): Update.
26572 (expand_neg_overflow): Update.
26573 (expand_mul_overflow): Update.
26574 (expand_vector_ubsan_overflow): Update.
26575 * ipa-cp.c (good_cloning_opportunity_p): Update.
26576 * ipa-split.c (split_function): Use make_single_succ_edge.
26577 * ipa-utils.c (ipa_merge_profiles): Update.
26578 * loop-doloop.c (add_test): Update.
26579 (doloop_modify): Update.
26580 * loop-unroll.c (compare_and_jump_seq): Update.
26581 (unroll_loop_runtime_iterations): Update.
26582 * lra-constraints.c (lra_inheritance): Update.
26583 * lto-streamer-in.c (input_cfg): Update.
26584 * lto-streamer-out.c (output_cfg): Update.
26585 * mcf.c (adjust_cfg_counts): Update.
26586 * modulo-sched.c (sms_schedule): Update.
26587 * omp-expand.c (expand_omp_for_init_counts): Update.
26588 (extract_omp_for_update_vars): Update.
26589 (expand_omp_ordered_sink): Update.
26590 (expand_omp_for_ordered_loops): Update.
26591 (expand_omp_for_generic): Update.
26592 (expand_omp_for_static_nochunk): Update.
26593 (expand_omp_for_static_chunk): Update.
26594 (expand_cilk_for): Update.
26595 (expand_omp_simd): Update.
26596 (expand_omp_taskloop_for_outer): Update.
26597 (expand_omp_taskloop_for_inner): Update.
26598 * omp-simd-clone.c (simd_clone_adjust): Update.
26599 * optabs.c (expand_doubleword_shift): Update.
26600 (expand_abs): Update.
26601 (emit_cmp_and_jump_insn_1): Update.
26602 (expand_compare_and_swap_loop): Update.
26603 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26604 * predict.c (predictable_edge_p): Update.
26605 (edge_probability_reliable_p): Update.
26606 (set_even_probabilities): Update.
26607 (combine_predictions_for_insn): Update.
26608 (combine_predictions_for_bb): Update.
26609 (propagate_freq): Update.
26610 (estimate_bb_frequencies): Update.
26611 (force_edge_cold): Update.
26612 * profile-count.c (profile_count::dump): Add missing space into dump.
26613 (profile_count::debug): Add newline.
26614 (profile_count::differs_from_p): Explicitly convert to unsigned.
26615 (profile_count::stream_in): Update.
26616 (profile_probability::dump): New member function.
26617 (profile_probability::debug): New member function.
26618 (profile_probability::differs_from_p): New member function.
26619 (profile_probability::differs_lot_from_p): New member function.
26620 (profile_probability::stream_in): New member function.
26621 (profile_probability::stream_out): New member function.
26622 * profile-count.h (profile_count_quality): Rename to ...
26623 (profile_quality): ... this one.
26624 (profile_probability): New.
26625 (profile_count): Update.
26626 * profile.c (compute_branch_probabilities): Update.
26627 * recog.c (peep2_attempt): Update.
26628 * sched-ebb.c (schedule_ebbs): Update.
26629 * sched-rgn.c (find_single_block_region): Update.
26630 (compute_dom_prob_ps): Update.
26631 (schedule_region): Update.
26632 * sel-sched-ir.c (compute_succs_info): Update.
26633 * stmt.c (struct case_node): Update.
26634 (do_jump_if_equal): Update.
26635 (get_outgoing_edge_probs): Update.
26636 (conditional_probability): Update.
26637 (emit_case_dispatch_table): Update.
26638 (expand_case): Update.
26639 (expand_sjlj_dispatch_table): Update.
26640 (emit_case_nodes): Update.
26641 * targhooks.c: Update.
26642 * tracer.c (better_p): Update.
26643 (find_best_successor): Update.
26644 * trans-mem.c (expand_transaction): Update.
26645 * tree-call-cdce.c: Update.
26646 * tree-cfg.c (gimple_split_edge): Upate.
26647 (move_sese_region_to_fn): Upate.
26648 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26649 * tree-eh.c (lower_resx): Upate.
26650 (cleanup_empty_eh_move_lp): Upate.
26651 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26652 * tree-inline.c (copy_edges_for_bb): Update.
26653 (copy_cfg_body): Update.
26654 * tree-parloops.c (gen_parallel_loop): Update.
26655 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26656 (gimple_gen_time_profiler): Update.
26657 * tree-ssa-dce.c (remove_dead_stmt): Update.
26658 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26659 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26660 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26661 (unloop_loops): Update.
26662 (try_peel_loop): Update.
26663 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26664 * tree-ssa-loop-split.c (connect_loops): Update.
26665 (split_loop): Update.
26666 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26667 (hoist_guard): Update.
26668 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26669 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26670 (value_replacement): Update.
26671 * tree-ssa-reassoc.c (branch_fixup): Update.
26672 * tree-ssa-tail-merge.c (replace_block_by): Update.
26673 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26674 (create_edge_and_update_destination_phis): Update.
26675 (compute_path_counts): Update.
26676 (recompute_probabilities): Update.
26677 (update_joiner_offpath_counts): Update.
26678 (freqs_to_counts_path): Update.
26679 (duplicate_thread_path): Update.
26680 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26681 (struct switch_conv_info): Update.
26682 (gen_inbound_check): Update.
26683 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26684 (vect_do_peeling): Update.
26685 (vect_loop_versioning): Update.
26686 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26687 (optimize_mask_stores): Update.
26688 * ubsan.c (ubsan_expand_null_ifn): Update.
26689 * value-prof.c (gimple_divmod_fixed_value): Update.
26690 (gimple_divmod_fixed_value_transform): Update.
26691 (gimple_mod_pow2): Update.
26692 (gimple_mod_pow2_value_transform): Update.
26693 (gimple_mod_subtract): Update.
26694 (gimple_mod_subtract_transform): Update.
26695 (gimple_ic): Update.
26696 (gimple_stringop_fixed_value): Update.
26697 (gimple_stringops_transform): Update.
26698 * value-prof.h: Update.
26700 2017-06-29 Carl Love <cel@us.ibm.com>
26702 * config/rs6000/rs6000-c.c: Add support for built-in functions
26703 vector signed int vec_signed (vector float);
26704 vector signed long long vec_signed (vector double);
26705 vector signed int vec_signed2 (vector double, vector double);
26706 vector signed int vec_signede (vector double);
26707 vector signed int vec_signedo (vector double);
26708 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26709 instruction generator.
26710 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26711 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26712 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26714 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26715 vunsignede_v2df): Add define_expands.
26716 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26717 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26718 VEC_UNSIGNEDO): Add definitions.
26719 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26720 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26721 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26722 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26723 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26724 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26725 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26726 vec_unsignede, vec_unsignedo): Add builtin defines.
26727 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26729 * doc/extend.texi: Update the built-in documentation file for the
26730 new built-in functions.
26732 2017-06-29 Richard Biener <rguenther@suse.de>
26734 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26735 reduction chains to LOOP_VINFO_REDUCTIONS.
26736 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26737 SLP reductions after processing reduction chains.
26739 2017-06-29 Nathan Sidwell <nathan@acm.org>
26741 * builtins.c (fold_builtin_FUNCTION): Use
26742 lang_hooks.decl_printable_name.
26744 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26746 PR middle-end/81194
26747 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26748 with only one label.
26749 * stmt.c (expand_case): Assert NCASES is greater than one.
26751 2017-06-29 Richard Biener <rguenther@suse.de>
26753 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26755 (group_case_labels): Likewise.
26756 (find_taken_edge): Push sanity checking on val to workers...
26757 (find_taken_edge_cond_expr): ... here
26758 (find_taken_edge_switch_expr): ... and here, handle cases
26759 with just a default label.
26760 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26761 (group_case_labels): Likewise.
26762 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26763 group_case_labels does anything cleanup the CFG again.
26765 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26767 PR tree-optimization/81196
26768 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26769 exit condition comparing two IVs.
26771 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26773 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26774 profile to the dummy entry at the end of the list of architectures.
26775 * config/arm/arm-cpu-cdata.h: Regenerated.
26777 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26778 Michael Collison <michael.collison@arm.com>
26781 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26783 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26784 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26785 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26786 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26787 of shift when the shift amount is masked with constant equal to
26788 the size of the mode.
26789 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26792 2017-06-29 Martin Liska <mliska@suse.cz>
26794 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26796 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26799 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26802 (edge_growth_cache_entry::edge_growth_cache_entry): New
26804 (reset_edge_growth_cache): Update to use constructor.
26806 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26808 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26809 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26810 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26812 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26814 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26815 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26817 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26819 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26820 (*-linux-uclibc*): Add t-uclibc tmake_file.
26821 * config/t-musl: New.
26822 * config/t-uclibc: New.
26824 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26826 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26828 (gen_comm_data): Emit architectural setting of arch_prof.
26829 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26831 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26832 (armv8-m.base, armv8-m.main): Likewise.
26833 * arm-protos.h (arm_build_target): Add profile field.
26834 (arch_option): Likewise.
26835 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26837 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26838 arm_active_target.profile.
26840 2017-06-28 Richard Biener <rguenther@suse.de>
26842 PR middle-end/81227
26843 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26844 TYPE_OVERFLOW_WRAPS.
26845 * match.pd (negate_expr_p): Likewise.
26846 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26847 fold_build2, not fold_binary.
26849 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26851 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26852 Convert memory address to Pmode.
26853 (aarch64_print_operand): Assert MEM operands are always Pmode.
26855 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26858 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26859 Remove redundant if.
26860 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26861 * config/arm/aarch-common-protos.h
26862 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26863 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26865 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26868 * multiple_target.c (create_dispatcher_calls): Set the default
26869 clone to be static, not public.
26871 2017-06-28 Richard Biener <rguenther@suse.de>
26873 * tree-vect-loop.c (vectorizable_reduction): Move special
26874 cond reduction IV var creation ...
26875 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26876 parameter. Use STMT_VINFO_VECTYPE.
26877 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26880 2017-06-28 Martin Liska <mliska@suse.cz>
26883 * ipa-visibility.c (non_local_p): Handle visibility.
26885 2017-06-28 Martin Liska <mliska@suse.cz>
26888 * common.opt: Add IntegerRange to various options.
26889 * opt-functions.awk (integer_range_info): New function.
26890 * optc-gen.awk: Add integer_range_info to cl_options struct.
26891 * opts-common.c (decode_cmdline_option): Handle
26892 CL_ERR_INT_RANGE_ARG.
26893 (cmdline_handle_error): Likewise.
26894 * opts.c (print_filtered_help): Show valid interval in
26895 when --help is provided.
26896 * opts.h (struct cl_option): Add range_min and range_max fields.
26897 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26899 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26901 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26902 (x * C EQ/NE y * C): New transformation.
26904 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26906 * genmultilib (combination_space): Accept '+' in option names.
26908 2017-06-28 Martin Liska <mliska@suse.cz>
26911 * asan.c (instrument_derefs): Bail out inner references
26912 that are hard register variables.
26914 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26917 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26918 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26920 2017-06-28 Richard Biener <rguenther@suse.de>
26922 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26923 (vect_get_slp_defs): Adjust.
26924 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26926 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26928 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26929 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26930 (vectorizable_reduction): Adjust.
26931 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26933 (vect_get_slp_defs): Likewise.
26934 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26935 (vectorizable_bswap): Adjust.
26936 (vectorizable_call): Likewise.
26937 (vectorizable_conversion): Likewise.
26938 (vectorizable_assignment): Likewise.
26939 (vectorizable_shift): Likewise.
26940 (vectorizable_operation): Likewise.
26941 (vectorizable_store): Likewise.
26942 (vectorizable_condition): Likewise.
26943 (vectorizable_comparison): Likewise.
26945 2017-06-28 Michael Collison <michael.collison@arm.com>
26948 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26950 (arm_gen_movmemqi): Removed unused variable 'i'.
26951 Convert 'for' loop into 'while' loop.
26952 (arm_expand_prologue): Remove last unnecessary set of insn.
26953 (thumb_pop): Remove unused variable 'pushed_words'.
26954 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26956 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26958 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26959 * config/s390/s390-protos.h: Add prototype of
26960 s390_rel_address_ok_p.
26961 * config/s390/s390.c (s390_got_symbol): New function.
26962 (s390_rel_address_ok_p): New function.
26963 (legitimize_pic_address): Use s390_rel_address_ok_p.
26964 (s390_load_got): Use s390_got_symbol.
26965 (s390_option_override): Issue error if
26966 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26967 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26969 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26971 2017-06-27 Andrew Pinski <apinski@cavium.com>
26973 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26974 (X * copysign (1.0, X)): New pattern.
26975 (X * copysign (1.0, -X)): New pattern.
26976 (copysign (-1.0, CST)): New pattern.
26978 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26980 * genmultilib (combination_space): Remove variable.
26981 Validate reuse rules against regular expression for any sequence
26982 of multilib options in any order.
26984 2017-06-27 Michael Collison <michael.collison@arm.com>
26986 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26987 call aarch64_split_simd_combine.
26988 * (aarch64_combine_internal<mode>): Delete pattern.
26989 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26990 Allow register and subreg operands.
26992 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26994 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26995 specific need, just fallback on defaults.
26996 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26998 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26999 Olivier Hainque <hainque@adacore.com>
27001 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27003 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27004 targets. Pick a default if no particular attempt applied.
27005 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27008 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27010 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27011 (x86_64-wrs-vxworks7): Likewise.
27013 2017-06-27 Marek Polacek <polacek@redhat.com>
27016 * ubsan.c (instrument_null): Check get_base_address's result for null.
27018 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27020 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27022 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27024 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27025 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27026 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27027 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27028 New function types.
27029 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27030 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27031 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27032 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27033 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27034 BUILT_IN_FEUPDATEENV): New builtins.
27035 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27036 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27037 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27038 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27040 (builtin_structptr_types): Adjust size.
27041 * tree.c (builtin_structptr_types): Add four entries.
27043 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27044 Olivier Hainque <hainque@adacore.com>
27046 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27047 (TLS_SYM): New local macro, forcing reference to __tls__ on
27048 link command lines for VxWorks 7 RTPs, triggering initialization
27050 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27051 OS features TLS support, true for RTPs on VxWorks 7.
27052 * config/vxworks.c (vxworks_override_options): Setup emutls
27055 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27057 * predict.c (test_prediction_value_range): Use -1U instead of -1
27058 to avoid narrowing conversion warning.
27059 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27060 to avoid narrowing conversion warning.
27061 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27063 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27065 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27067 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27068 64bit configurations.
27069 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27070 (SIZE_TYPE): Likewise.
27071 * config/vxworks.c (vxworks_emutls_var_fields): Use
27072 long_unsigned_type_node instead of unsigned_type_node as the offset
27073 field type, which is "pointer" mode in emutls.c.
27075 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27078 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27080 PR middle-end/81207
27081 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27082 gimple_vuse copying separately from gimple_vdef copying.
27084 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27086 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27089 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27090 Olivier Hainque <hainque@adacore.com>
27092 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27093 for all vxworks7 targets.
27094 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27095 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27096 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27097 variations for VX6/VX7 and 32/64bits later on in ...
27098 (VXWORKS_LIB_SPEC): Leverage new macros.
27099 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27100 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27102 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27104 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27105 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27107 2017-06-26 Carl Love <cel@us.ibm.com>
27109 * config/rs6000/rs6000-c.c: Add support for built-in functions
27110 vector bool char vec_reve (vector bool char);
27111 vector signed char vec_reve (vector signed char);
27112 vector unsigned char vec_reve (vector unsigned char);
27113 vector bool int vec_reve (vector bool int);
27114 vector signed int vec_reve (vector signed int);
27115 vector unsigned int vec_reve (vector unsigned int);
27116 vector bool long long vec_reve (vector bool long long);
27117 vector signed long long vec_reve (vector signed long long);
27118 vector unsigned long long vec_reve (vector unsigned long long);
27119 vector bool short vec_reve (vector bool short);
27120 vector signed short vec_reve (vector signed short);
27121 vector double vec_reve (vector double);
27122 vector float vec_reve (vector float);
27123 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27124 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27125 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27126 (altivec_vreve): New pattern.
27127 * config/rs6000/altivec.h (vec_reve): New define.
27128 * doc/extend.texi (vec_rev): Update the built-in documentation file
27129 for the new built-in functions.
27131 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27133 PR tree-optimization/71815
27134 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27136 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27138 (slsr_process_phi): Likewise.
27139 (replace_uncond_cands_and_profitable_phis): Don't replace a
27140 multiply candidate with a stride of 1 (copy or cast).
27141 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27143 (lowest_cost_path): Likewise.
27144 (total_savings): Likewise.
27146 2017-06-26 Richard Biener <rguenther@suse.de>
27149 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27150 Use def_builtin_pure for all gather builtins.
27152 2017-06-26 Richard Biener <rguenther@suse.de>
27154 PR tree-optimization/81203
27155 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27156 non-dominating BBs.
27158 2017-06-26 Marek Polacek <polacek@redhat.com>
27161 * doc/invoke.texi: Document -Wmultistatement-macros.
27163 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27165 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27166 arm_neon_ok_no_float_abi effective target.
27168 2017-06-26 Richard Biener <rguenther@suse.de>
27170 PR tree-optimization/80928
27171 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27172 (copy_bbs): Set BB_DUPLICATED flag early.
27173 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27175 (execute_on_shrinking_pred): Likewise.
27176 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27177 BB_DUPLICATED blocks.
27178 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27179 iterate over all PHIs considering removal of *gsi.
27181 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27183 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27186 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27188 * config/rs6000/rs6000-string.c: (expand_block_clear,
27189 do_load_for_compare, select_block_compare_mode,
27190 compute_current_alignment, expand_block_compare,
27191 expand_strncmp_align_check, expand_strn_compare,
27192 expand_block_move, rs6000_output_load_multiple)
27193 Move functions related to string/block move/compare
27194 to a separate file.
27195 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27196 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27197 for this function which is now used in two files.
27198 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27199 * config.gcc: Add rs6000-string.o to extra_objs for
27200 targets powerpc*-*-* and rs6000*-*-*.
27202 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27205 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27206 32-bit, since indexed is not valid for DImode.
27207 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27208 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27209 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27210 (define_peephole2 for Altivec d-form store): Likewise.
27213 * multiple_target.c (create_dispatcher_calls): Only create the
27214 dispatcher call if the function is the default clone of a
27215 versioned function.
27217 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27219 PR middle-end/80902
27220 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27221 a call, force the call to not be a tail call.
27223 2017-06-23 Jeff Law <law@redhat.com>
27225 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27228 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27230 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27231 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27232 handling for early expansion of vector shifts (sl,sr,sra,rl).
27233 (builtin_function_type): Add vector shift right instructions
27234 to the unsigned argument list.
27236 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27238 rtl-optimizatoin/79286
27239 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27240 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27241 trap. PIC register plus a const unspec without offset can never trap.
27243 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27245 * tree.h (builtin_structptr_type): New type.
27246 (builtin_structptr_types): Declare new array.
27247 * tree.c (builtin_structptr_types): New array.
27248 (free_lang_data, build_common_tree_nodes): Use it.
27250 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27253 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27256 2017-06-22 Matt Turner <mattst88@gmail.com>
27258 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27259 Lake models to skylake case. Assume skylake for unknown
27260 models with clflushopt.
27262 2017-06-22 Jeff Law <law@redhat.com>
27264 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27265 frame sizes that do not satisfy aarch64_uimm12_shift.
27267 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27269 * profile-count.h (apply_probability,
27270 apply_scale, probability_in): Fix checks for zero.
27272 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27274 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27275 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27277 2016-06-22 Richard Biener <rguenther@suse.de>
27279 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27280 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27281 REDUC_MAX_EXPR support.
27282 (vectorizable_reduction): Likewise.
27283 (vect_create_epilog_for_reduction): Likewise.
27285 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27287 * match.pd (A / (1 << B) -> A >> B): New.
27288 * generic-match-head.c: Include optabs-tree.h.
27289 * gimple-match-head.c: Likewise.
27290 * optabs-tree.h (target_supports_op_p): New.
27291 * optabs-tree.c (target_supports_op_p): New.
27293 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27295 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27296 $gcc_cv_ld --help output.
27297 (gcc_cv_ld_demangle): Likewise.
27298 (gcc_cv_ld_eh_frame_hdr): Likewise.
27299 (gcc_cv_ld_pie): Likewise.
27300 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27301 (gcc_cv_ld_buildid): Likewise.
27302 (gcc_cv_ld_sysroot): Likewise.
27303 (ld_bndplt_support): Likewise.
27304 (ld_pushpopstate_support): Likewise.
27305 * configure: Regenerate.
27306 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27308 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27311 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27312 operands indexes to avoid gap between operands and match_dups.
27314 2017-06-21 Andrew Pinski <apinski@cavium.com>
27316 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27317 Increment Arith_shift and Arith_shift_reg by 1.
27318 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27320 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27321 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27322 (aarch64_strip_extend): Add new argument and test for it.
27323 (aarch64_cheap_mult_shift_p): New function.
27324 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27325 add a cost if it is true.
27326 Update calls to aarch64_strip_extend.
27327 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27329 2017-06-21 Andrew Pinski <apinski@cavium.com>
27331 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27333 (thunderxt88): Likewise.
27334 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27335 (thunderx_prefetch_tune): New variable.
27336 (thunderx2t99_prefetch_tune): Update for the correct values.
27337 (thunderxt88_tunings): New variable.
27338 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27339 generic_prefetch_tune.
27340 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27342 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27344 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27345 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27346 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27347 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27348 (aarch64_atomic_cas<mode>, GPI): Likewise.
27350 2017-06-21 Martin Liska <mliska@suse.cz>
27352 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27353 statements on cold and hot labels.
27354 * predict.c (tree_estimate_probability_bb): Remove the
27355 prediction from this place.
27357 2017-06-21 Martin Liska <mliska@suse.cz>
27359 PR tree-optimization/79489
27360 * gimplify.c (maybe_add_early_return_predict_stmt): New
27362 (gimplify_return_expr): Call the function.
27363 * predict.c (tree_estimate_probability_bb): Remove handling
27365 * predict.def: Update comment about early return predictor.
27366 * gimple-predict.h (is_gimple_predict): New function.
27367 * predict.def: Change default value of early return to 66.
27368 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27370 * passes.def: Put pass_strip_predict_hints to the beginning of
27373 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27375 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27376 FUNCTION_DECL declarations.
27377 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27379 (dwaf2out_decl): Likewise.
27380 * godump.c (go_early_global_decl): Skip call to the real debug hook
27381 for FUNCTION_DECL declarations.
27382 * passes.c (rest_of_decl_compilation): Skip call to the
27383 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27384 -fdump-go-spec is passed.
27386 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27388 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27389 Reorder for compactness.
27390 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27391 (def_builtin_pure, def_builtin_pure2): New functions.
27392 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27394 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27396 * match.pd (nop_convert): New predicate.
27397 ((A +- CST1) +- CST2): Allow some NOP conversions.
27399 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27402 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27403 with ctors/dtors if GOVD_SHARED is set.
27405 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27407 * config/aarch64/aarch64.md (movti_aarch64):
27408 Emit mov rather than orr.
27409 (movtf_aarch64): Likewise.
27410 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27411 Emit mov rather than orr.
27413 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27415 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27416 Swap alternatives, make integer dup more expensive.
27418 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27420 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27421 Return true for non-tls symbols.
27423 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27425 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27426 (cortex-a75): Likewise.
27427 (cortex-a75.cortex-a55): Likewise.
27428 * config/aarch64/aarch64-tune.md: Regenerate.
27429 * doc/invoke.texi (-mtune): Document new values for -mtune.
27431 2017-06-21 Tom de Vries <tom@codesourcery.com>
27433 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27434 stack_size feature.
27435 (Effective-Target Keywords, Other attributes): Suggest using
27436 dg-add-options stack_size feature to get stack limit in stack_size
27437 effective target documentation.
27439 2017-06-21 Julian Brown <julian@codesourcery.com>
27440 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27442 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27443 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27444 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27446 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27447 attribute type list for neon_multiply.
27448 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27449 attribute type list for neon_multiply.
27450 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27451 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27452 attribute type list for neon_multiply.
27453 * config/arm/types.md (crypto_pmull): Add.
27454 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27455 attribute type list.
27457 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27459 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27462 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27464 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27465 to make sure not to dereference a NULL cost_classes_ptr pointer.
27467 2017-06-20 Carl Love <cel@us.ibm.com>
27469 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27470 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27471 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27472 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27473 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27474 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27475 VMULOSW): New enum "unspec" values.
27476 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27477 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27478 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27479 altivec_vmulosw): New patterns.
27480 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27481 VMULOSW): Add definitions.
27483 2017-06-20 Julia Koval <julia.koval@intel.com>
27485 * config/i386/i386.c: Fix rounding expand for new pattern.
27486 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27488 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27490 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27491 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27493 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27495 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27498 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27500 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27501 architecture, then by implementer ID.
27502 * config/aarch64/aarch64-tune.md: Regenerate.
27504 2017-06-20 Richard Biener <rguenther@suse.de>
27506 PR middle-end/81097
27507 * fold-const.c (split_tree): Fold to type before negating.
27509 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27511 * diagnostic-show-locus.c
27512 (selftest::test_fixit_deletion_affecting_newline): New function.
27513 (selftest::diagnostic_show_locus_c_tests): Call it.
27515 2017-06-20 Andreas Schwab <schwab@suse.de>
27518 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27521 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27523 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27524 __ARM_FEATURE_COPROC according to support.
27526 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27528 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27529 Rewritten to avoid overflow for > 32-bit pointers.
27532 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27533 by removing enum keyword.
27534 (ubsan_type_descriptor): Likewise. Formatting fix.
27537 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27538 splitter): Require TARGET_SSE2 in the condition.
27540 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27543 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27544 for doing vector set of SFmode on ISA 3.0.
27545 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27546 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27548 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27549 SFmode value into a V4SF variable that was extracted from another
27550 V4SF variable without converting the element to double precision
27551 and back to single precision vector format.
27552 (vsx_insert_extract_v4sf_p9_2): Likewise.
27554 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27556 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27557 in UWHI to avoid undefined overflow.
27560 * ubsan.h (enum ubsan_encode_value_phase): New.
27561 (ubsan_encode_value): Change second argument to
27562 enum ubsan_encode_value_phase with default value of
27563 UBSAN_ENCODE_VALUE_GENERIC.
27564 * ubsan.c (ubsan_encode_value): Change second argument to
27565 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27566 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27567 create_tmp_var_raw instead of create_tmp_var and use a
27569 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27570 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27571 ubsan_encode_value callers.
27574 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27575 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27576 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27578 2017-06-19 Richard Biener <rguenther@suse.de>
27580 PR middle-end/81118
27581 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27582 estimates if we changed anything.
27584 2017-06-19 Richard Biener <rguenther@suse.de>
27586 PR tree-optimization/80887
27587 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27588 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27589 simplified lookups, then reset mprts_hook.
27590 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27592 (try_to_simplify): Likewise.
27594 2017-06-19 Martin Liska <mliska@suse.cz>
27597 * gimplify.c (gimplify_switch_expr):
27598 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27600 2017-06-19 Martin Liska <mliska@suse.cz>
27602 * doc/install.texi: Document that PGO runs in 4 stages.
27604 2017-06-19 Martin Liska <mliska@suse.cz>
27607 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27608 to dispatcher function name.
27609 * multiple_target.c (replace_function_decl): New function.
27610 (create_dispatcher_calls): Redirect both edges and references.
27612 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27614 * profile-count.c (profile_count::dump): Dump quality.
27615 (profile_count::differs_from_p): Update for unsigned val.
27616 * profile-count.h (profile_count_quality): New enum.
27617 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27619 2017-06-19 Richard Biener <rguenther@suse.de>
27621 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27622 struct function as arg.
27623 (estimate_numbers_of_iterations): Export overload with loop arg.
27624 (free_numbers_of_iterations_estimates_loop): Use an overload of
27625 free_numbers_of_iterations_estimates instead.
27626 * tree-cfg.c (remove_bb): Adjust.
27627 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27628 * tree-parloops.c (gen_parallel_loop): Likewise.
27629 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27631 (tree_unroll_loops_completely): Likewise.
27632 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27633 Use an overload instead and export.
27634 (estimated_loop_iterations): Adjust.
27635 (max_loop_iterations): Likewise.
27636 (likely_max_loop_iterations): Likewise.
27637 (estimate_numbers_of_iterations): Take struct function as arg
27639 (loop_exits_before_overflow): Adjust.
27640 (free_numbers_of_iterations_estimates_loop): Use an overload.
27641 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27642 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27644 2017-06-19 Richard Biener <rguenther@suse.de>
27647 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27648 RANGE_EXPR conservatively.
27650 2017-06-16 Carl Love <cel@us.ibm.com>
27652 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27653 definitions for vec_float, vec_float2, vec_floato,
27654 vec_floate built-ins.
27655 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27656 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27658 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27659 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27660 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27661 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27662 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27663 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27664 vec_floato): Add builtin defines.
27665 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27666 Update the built-in documentation file for the new built-in
27669 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27671 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27672 (mthumb): Mark as the negative of -marm.
27674 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27676 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27678 (ARM Options, -mtune): Document that this accepts the same
27679 extension options as -mcpu.
27680 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27682 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27684 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27685 permitted extensions.
27687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27689 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27690 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27691 (armv8-m.main): Add option +nodsp.
27692 * config/arm/arm-cpu-cdata.h: Regenerated.
27694 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27696 * config/arm/t-fuchsia: New file.
27697 * config.gcc (arm*-*-fuchsia*): Use it.
27699 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27701 * config/arm/t-symbian: Rewrite for new option infrastructure.
27703 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27705 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27706 (MULTILIB_REQUIRED): Likewise.
27708 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27710 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27711 (MULTILIB_RESUE): Likewise.
27712 (MULTILIB_MATCHES): Likewise.
27713 (MULTLIB_REQUIRED): Likewise.
27715 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27717 * config/arm/t-rtems: Rewrite for new option framework.
27719 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27721 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27722 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27723 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27724 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27725 * config/arm/t-multilib: ... here.
27726 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27727 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27728 armv7-a and armv8*-a when A-profile libraries have not been built.
27729 * config/arm/t-rmprofile: Rewrite.
27731 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27733 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27734 with a backslash. Remove the backslash after substituting unescaped
27736 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27738 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27740 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27741 the driver to pass through the default setting of -mfloat-abi.
27742 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27744 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27745 (all_feat_combs): New rule.
27746 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27748 * config/arm/t-aprofile: Rewrite.
27750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27752 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27753 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27754 fpu is not specified by the user/command-line.
27755 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27756 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27757 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27758 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27759 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27760 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27764 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27765 * config/arm/t-arm-elf: Rewritten.
27767 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27769 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27770 have some floating-point instructions.
27771 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27772 (TARGET_MAYBE_HARD_FLOAT): New macro.
27773 * config/arm/arm-builtins.c (arm_init_builtins): Use
27774 TARGET_MAYBE_HARD_FLOAT.
27775 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27777 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27779 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27780 (configargs.h): Include it.
27781 (arm_print_hint_for_fpu_option): New function.
27782 (arm_parse_fpu_option): New function.
27783 (candidate_extension): New class.
27784 (arm_canon_for_multilib): New function.
27785 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27786 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27787 (ARCH_CANONICAL_SPECS): New macro.
27788 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27790 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27792 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27793 are set after handling multilib fragments. Set target_cpu_default2
27796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27798 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27800 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27801 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27803 (check_arch): Likewise.
27804 * config/arm/arm.c (arm_configure_build_target): Handle
27805 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27806 options in the default.
27808 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27810 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27811 when an option is an alias of another.
27812 * config/arm/parsecpu.awk (optalias): New parser token.
27813 (gen_comm_data): Mark non-alias options as such. Emit entries
27814 for extension aliases.
27815 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27816 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27817 (armv6kz, armv6zk, armv6t2): Likewise.
27818 (armv7): Make vfpv3-d16 an alias.
27819 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27821 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27822 Sort in canonical order.
27823 (armv8-a): Sort in canonical order.
27824 (armv8.1-a, armv8.2-a): Likewise.
27825 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27827 (cortex-a9): Sort in canonical order.
27828 * config/arm/arm.c (selftests.h): Include it.
27829 (arm_test_cpu_arch_data): New function.
27830 (arm_run_self_tests): New function.
27831 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27832 (targetm): Move declaration to the end of the file.
27833 * arm-cpu-cdata.h: Regenerated.
27835 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27837 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27838 call to target_mode_check describing the type of option passed.
27839 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27840 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27841 arm_parse_cpu_option_name to match parameters against list of
27843 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27844 arm_arch_core_flags data structure.
27845 * config/arm/arm-cpu_cdata.h: Regenerated.
27847 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27849 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27851 (arm_print_hint_for_cpu_option): Likewise.
27852 (arm_print_hint_for_arch_option): Likewise.
27853 (arm_parse_cpu_option_name): Likewise.
27854 (arm_parse_arch_option_name): Likewise.
27855 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27856 of entries in the all_fpus list.
27857 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27858 (arm_parse_cpu_option_name): Declare.
27859 (arm_parse_arch_option_name): Declare.
27860 (arm_parse_option_features): Declare.
27861 (arm_intialize_isa): Declare.
27862 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27864 (gen_comm_data): ... here. Make definitions non-static.
27865 * config/arm/arm-cpu-data.h: Regenerated.
27866 * config/arm/arm-cpu-cdata.h: Regenerated.
27868 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27870 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27871 (cpu_arch_extension): New structure.
27872 (cpu_arch_option, arch_option, cpu_option): New structures.
27873 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27874 architecture types.
27875 (gen_data): Generate new format data tables.
27876 * config/arm/arm.c (cpu_tune): New structure.
27877 (cpu_option, processors): Delete.
27878 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27879 (arm_print_hint_for_cpu_option): ... this and ...
27880 (arm_print_hint_for_arch_option): ... this.
27881 (arm_parse_arch_cpu_name): Delete. Replace with ...
27882 (arm_parse_cpu_option_name): ... this and ...
27883 (arm_parse_arch_option_name): ... this.
27884 (arm_unrecognized_feature): Change type of target parameter to
27886 (arm_parse_arch_cpu_features): Delete. Replace with ...
27887 (arm_parse_option_features): ... this.
27888 (arm_configure_build_target): Rework to use new configuration data
27890 (arm_print_tune_info): Rework for new configuration data tables.
27891 * config/arm/arm-cpu-data.h: Regenerated.
27892 * config/arm/arm-cpu.h: Regenerated.
27894 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27896 * Makefile.in (OBJS): Move sbitmap.o from here ...
27897 (OBJS-libcommon): ... to here.
27899 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27901 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27902 (ISA_ALL_CRYPTO): New macro.
27903 (ISA_ALL_SIMD): New macro
27904 (ISA_ALL_FP): New macro.
27905 * config/arm/arm.c (fpu_bitlist): Update initializer.
27906 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27908 (arm9e): Add fpu. Add option for nofp
27909 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27910 (arm926ej-s, arm1026ej-s): Likewise.
27911 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27912 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27913 neon-fp16, neon-vfpv4, nofp and nosimd.
27914 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27915 (cortex-a8): Add fpu. Add option for nofp.
27916 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27917 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27918 (cortex-r4f): Add fpu.
27919 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27920 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27922 (cortex-r8): Likewise.
27923 (cortex-m4): Add fpu. Add option for nofp.
27924 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27925 (cortex-a17.cortex-a7): Likewise.
27926 (cortex-a32): Add fpu. Add options for crypto and nofp.
27927 (cortex-a35, cortex-a53): Likewise.
27928 (cortex-a57): Add fpu. Add option for crypto.
27929 (cortex-a72, cortex-a73): Likewise.
27930 (exynos-m1): Likewise.
27931 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27932 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27933 (cortex-m33): Add fpu. Add option for nofp.
27934 * config/arm/arm-cpu-cdata.h: Regenerated
27935 * config/arm/arm-cpu-data.h: Regenerated.
27937 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27939 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27940 (armv5te, armv5tej): Likewise.
27941 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27942 (armv7): Add options fp and vfpv3-d16.
27943 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27944 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27946 (armv7ve): Likewise.
27947 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27948 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27949 (armv8-a): Add nocrypto option.
27950 (armv8.1-a, armv8.2-a): Likewise.
27951 (armv8-m.main): add options fp, fp.dp and nofp.
27953 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27955 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27957 (armv8-a+crc): Delete.
27958 (armv8.1-a): Add options simd, crypto and nofp.
27959 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27960 (armv8.2-a+fp16): Delete.
27961 (armv8-m.main): Add option dsp.
27962 (armv8-m.main+dsp): Delete.
27963 (cortex-a8): Add fpu. Add nofp option.
27964 (cortex-a9): Add fpu. Add nofp and nosimd options.
27965 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27966 link to main cpu and architecture data structures.
27967 (gen_comm_data): Only put isa attributes from the main architecture
27969 (option): New statement for architecture and CPU entries.
27970 * arm.c (struct cpu_option): New structure.
27971 (struct processors): Add entry for options.
27972 (arm_unrecognized_feature): New function.
27973 (arm_parse_arch_cpu_name): Ignore any characters after the first
27975 (arm_parse_arch_cpu_feature): New function.
27976 (arm_configure_build_target): Separate out any CPU and architecture
27977 features and parse separately. Don't error out if -mfpu=auto is
27978 used with only an architecture string.
27979 (arm_print_asm_arch_directives): New function.
27980 (arm_file_start): Call it.
27981 * config/arm/arm-cpu-cdata.h: Regenerated.
27982 * config/arm/arm-cpu-data.h: Likewise.
27983 * config/arm/arm-tables.opt: Likewise.
27985 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27987 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27988 assembler when it is not -mfpu=auto.
27990 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27992 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27993 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27994 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27995 (ASM_CPU_SPEC): Rewrite.
27996 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27997 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27998 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27999 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28000 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28001 copied string is NUL-terminated. Also strip any characters prefixed
28003 (arm_rewrite_selected_arch): New function.
28004 (arm_rewrite_march): New function.
28006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28008 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28009 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28010 (march, mcpu, mtune): Convert to string-based options.
28011 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28012 (arm_parse_arch_cpu_name): New function.
28013 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28014 identify selected architecture or CPU.
28015 (arm_option_save): New function.
28016 (TARGET_OPTION_SAVE): Redefine.
28017 (arm_option_restore): Restore string options.
28018 (arm_option_print): Print string options.
28020 2017-06-16 Martin Sebor <msebor@redhat.com>
28022 PR tree-optimization/80933
28023 PR tree-optimization/80934
28024 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28025 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28026 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28027 (gimple_fold_builtin): Call them.
28029 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28031 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28032 as unlikely; update profile.
28034 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28036 * predict.c (force_edge_cold): Handle declaring edges impossible
28039 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28041 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28043 (try_unroll_loop_completely): Fix reporting.
28045 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28047 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28049 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28052 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28053 if given a non-constant argument for an intrinsic which requires a
28056 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28058 * profile.c (compare_freqs): New function.
28059 (branch_prob): Sort edge list.
28060 (find_spanning_tree): Assume that the list is priority sorted.
28062 2017-06-16 Richard Biener <rguenther@suse.de>
28064 PR tree-optimization/81090
28065 * passes.def (pass_record_bounds): Remove.
28066 * tree-pass.h (make_pass_record_bounds): Likewise.
28067 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28068 make_pass_record_bounds): Likewise.
28069 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28070 not free niter estimates at the beginning but at the end.
28071 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28073 2017-06-16 Richard Biener <rguenther@suse.de>
28075 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28076 initializer to workaround ICE in host GCC 4.8.
28078 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28080 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28082 (clone_inlined_nodes): Update.
28084 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28086 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28087 prefetch settings, and enable prefetching by default at -O3.
28089 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28091 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28092 Set flag_prefetch_loop_arrays according to tuning data.
28094 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28096 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28097 New tune structure.
28098 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28099 [Unrelated to main purpose of the patch] Place the pointer field last
28100 to enable type checking errors when tune structure are wrongly merged.
28101 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28102 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28103 (thunderx2t99_prefetch_tune): New tune constants.
28104 (tune_params *_tunings): Update all tunings (no functional change).
28105 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28106 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28107 from tunings structures.
28109 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28112 * ubsan.c (instrument_null): Add T argument, use it instead
28113 of computing it based on IS_LHS.
28114 (instrument_object_size): Likewise.
28115 (pass_ubsan::execute): Adjust instrument_null and
28116 instrument_object_size callers to pass gimple_get_lhs or
28117 gimple_assign_rhs1 result to it. Use instrument_null instead of
28118 calling get_base_address and instrument_mem_ref. Handle
28119 aggregate call arguments for object-size sanitization.
28121 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28123 PR tree-optimization/81089
28124 * tree-vrp.c (is_masked_range_test): Validate operands of
28127 2017-06-15 Martin Sebor <msebor@redhat.com>
28130 * dumpfile.c (dump_register): Avoid calling memset to initialize
28131 a class with a default ctor.
28132 * gcc.c (struct compiler): Remove const qualification.
28133 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28134 * hash-table.h: Ditto.
28135 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28137 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28138 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28140 * params.h (struct param_info): Make struct members non-const.
28141 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28142 with default initialization.
28143 * vec.h (vec_copy_construct, vec_default_construct): New helper
28145 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28146 with vec_copy_construct.
28147 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28148 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28149 * doc/invoke.texi (-Wclass-memaccess): Document.
28151 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28153 * emit-rtl.h (is_leaf): Update comment about local
28154 register allocator.
28156 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28159 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28160 for a variable to have a section before checking if the section has a
28162 Set section to.persistent if persistent attribute is set.
28163 Warn if .persistent attribute is used on an automatic variable.
28165 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28167 PR rtl-optimization/80474
28168 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28170 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28172 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28175 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28177 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28178 (casesi): Emit bounds checking as RTL.
28179 (casesi_internal_mips16_<mode>): Remove bounds checking.
28181 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28183 * config/xtensa/xtensa.c (xtensa_option_override): Append
28184 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28185 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28186 xtensa_doloop_hooks): Define unconditionally.
28187 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28189 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28190 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28191 for it in xtensa_option_override.
28192 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28193 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28195 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28197 * doc/cppopts.texi: Document '-' special value to -MF.
28199 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28201 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28202 (cortex_a53_fconst): Likewise.
28203 (cortex_a53_fpmul): Likewise.
28204 (cortex_a53_f_load_64): Likewise.
28205 (cortex_a53_f_load_many): Likewise.
28206 (cortex_a53_advsimd_alu): Likewise.
28207 (cortex_a53_advsimd_alu_q): Likewise.
28208 (cortex_a53_advsimd_mul): Likewise.
28209 (cortex_a53_advsimd_mul_q): Likewise.
28210 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28211 Add missing fmul, r2f_cvt and fconst cases.
28213 2017-06-14 Richard Biener <rguenther@suse.de>
28215 PR middle-end/81088
28216 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28218 (fold_binary_loc): When associating do not treat pre-existing
28219 TREE_OVERFLOW on literal constants as a reason to allow
28220 TREE_OVERFLOW on associated literal constants.
28222 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28224 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28225 (MASK_FEATURES): New macro.
28226 * config/sparc/sparc.c (sparc_option_override): Remove the special
28227 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28229 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28231 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28232 a division of 0 if non-call exceptions are enabled.
28234 2017-06-14 Andrew Pinski <apinski@cavium.com>
28235 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28238 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28239 Improve vector initialization code gen for only variable case.
28241 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28243 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28245 2017-06-14 Richard Biener <rguenther@suse.de>
28247 PR tree-optimization/81083
28248 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28251 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28253 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28254 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28255 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28256 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28257 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28258 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28260 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28262 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28263 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28265 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28267 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28269 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28271 * config/rs6000/t-rtems: Don't handle SPE.
28273 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28275 * config/rs6000/t-linux: Don't handle SPE.
28277 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28279 * config/rs6000/eabispe.h: Delete file.
28281 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28283 * config/rs6000/t-spe: Delete file.
28285 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28287 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28288 (rs6000_legitimate_offset_address_p): Return false for anything in
28289 V2SImode or V2SFmode.
28291 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28293 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28294 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28295 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28297 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28298 (rs6000_legitimate_offset_address_p): Ditto.
28299 (rs6000_emit_move): Ditto.
28300 (rs6000_init_builtins): Remove V4HI_type_node.
28302 2017-06-13 Martin Liska <mliska@suse.cz>
28305 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28306 (gate_asan): Likewise.
28307 * asan.h (asan_no_sanitize_address_p): Remove the function.
28308 (sanitize_flags_p): New function.
28309 * builtins.def: Fix coding style.
28310 * common.opt: Use renamed enum value.
28311 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28312 * doc/extend.texi: Document no_sanitize attribute.
28313 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28314 to SANITIZE_UNDEFINED_NONDEFAULT.
28315 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28316 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28317 Use sanitize_flags_p.
28318 * gimplify.c (gimplify_function_tree): Likewise.
28319 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28320 * opts.c (parse_no_sanitize_attribute): New function.
28321 (common_handle_option): Use renamed enum value.
28322 * opts.h (parse_no_sanitize_attribute): Declare.
28323 * tree.c (sanitize_flags_p): New function.
28324 * tree.h: Declared here.
28325 * tsan.c: Use sanitize_flags_p.
28326 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28327 (instrument_mem_ref): Likewise.
28328 (instrument_bool_enum_load): Likewise.
28329 (do_ubsan_in_current_function): Remove the function.
28330 (pass_ubsan::execute): Use sanitize_flags_p.
28331 * ubsan.h: Remove do_ubsan_in_current_function
28332 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28333 (dump_function_to_file): Use it here.
28335 2017-06-13 Martin Jambor <mjambor@suse.cz>
28337 PR tree-optimization/80803
28338 PR tree-optimization/81063
28339 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28340 (propagate_subaccesses_across_link): Enqueue subtree whenever
28341 necessary instead of relying on the caller.
28343 2017-06-13 Martin Jambor <mjambor@suse.cz>
28345 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28346 that have a first_link.
28347 (sort_and_splice_var_accesses): Do not check first_link before
28349 (subtree_mark_written_and_enqueue): Likewise.
28350 (propagate_all_subaccesses): Likewise and do not stop at first
28351 parent with a first_link.
28353 2017-06-13 Martin Jambor <mjambor@suse.cz>
28355 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28358 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28360 * match.pd: New pattern.
28362 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28364 * tree-vrp.c (is_masked_range_test): New function.
28365 (register_edge_assert_for): Determine ranges for
28368 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28370 PR tree-optimization/67328
28371 * fold-const.c (maskable_range_p): New function.
28372 (build_range_check): Generate bittests if possible.
28374 2017-06-13 Martin Liska <mliska@suse.cz>
28376 * gimple-pretty-print.c (dump_probability): Add new argument.
28377 (dump_edge_probability): Dump both probability and count.
28378 (dump_gimple_label): Likewise.
28379 (dump_gimple_bb_header): Likewise.
28381 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28384 * config/avr/avr-devices.c: Fix indentation.
28385 * config/avr/gen-avr-mmcu-specs.c: Dito.
28387 2017-06-13 Richard Biener <rguenther@suse.de>
28389 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28390 instead get vector type from stmt_info.
28391 (vectorizable_reduction): Adjust. Remove dead code.
28393 2017-06-13 Richard Biener <rguenther@suse.de>
28395 PR middle-end/81065
28396 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28397 case of C * (x * C2 + C3).
28398 (fold_addr_of_array_ref_difference): Properly fold index difference.
28400 2017-06-12 David S. Miller <davem@davemloft.net>
28403 * config/sparc/sparc.md (return expander): Emit frame blockage if
28404 function uses alloca.
28406 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28408 * combine.c (make_field_assignment): Check len rather than the mode
28409 precision when calling force_to_mode.
28411 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28413 Support multilibs and devices that see flash in RAM address range.
28416 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28417 (avr_mcu_t) <flash_pm_offset>: New field.
28418 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28419 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28420 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28421 (AVR_TINY_PM_OFFSET): Remove macro.
28422 * config/avr/avr.opt (-mshort-calls): New option.
28423 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28424 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28425 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28426 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28427 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28428 instead of avr_arch->have_jmp_call.
28429 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28430 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28431 avr_arch->flash_pm_offset to define.
28432 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28433 new field flash_pm_offset. Add entry for avrxmega3.
28434 (avr_texinfo): Add entry for avrxmega3.
28435 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28436 attiny212, attiny214,
28437 attiny412, attiny414, attiny416, attiny417,
28438 attiny814, attiny816, attiny817,
28439 attiny1614, attiny1616, attiny1617,
28440 attiny3214, attiny3216, attiny3217.
28441 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28442 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28443 (avr_print_operand_address) [AVR_TINY]: Same.
28444 (avr_asm_init_sections) <readonly_data_section>: Only patch
28445 callback if avr_arch->flash_pm_offset = 0.
28446 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28447 for rodata if avr_arch->flash_pm_offset != 0.
28448 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28449 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28450 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28451 (m_options): Append opt_rcall.
28452 (m_dirnames): Append dir_rcall.
28453 * config/avr/t-multilib: Regenerate.
28455 * configure.ac [target=avr]: Check whether avrxmega3 default
28456 linker description file works as needed.
28457 * configure: Regenerate.
28458 * doc/avr-mmcu.texi: Regenerate.
28459 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28460 <__AVR_ARCH__>: Document avrxmega3 and 103.
28461 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28462 <__AVR_SHORT_CALLS__>: Document it.
28463 <__AVR_PM_BASE_ADDRESS__>: Document it.
28464 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28465 (AVR Variable Attributes) <progmem>: Document this is
28466 not needed for avrxmega3.
28467 (AVR Named Address Spaces) <__flash>: Dito.
28469 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28471 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28473 2017-06-12 Doug Rupp <rupp@adacore.com>
28475 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28476 Append vxworks-stdint.h to the tm_file list.
28477 * config/vxworks-stdint.h: New file.
28479 2017-06-12 Martin Liska <mliska@suse.cz>
28481 PR tree-optimization/81041
28482 * tree-profile.c (gimple_gen_ic_func_profiler):
28483 Create an extra BB in profile-generate
28484 (gimple_gen_time_profiler): Likewise.
28486 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28488 PR tree-optimization/81003
28489 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28490 (update_range_test): Use it instead of force_gimple_operand_gsi.
28492 2017-06-12 Richard Biener <rguenther@suse.de>
28494 PR tree-optimization/81053
28495 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28496 with backedge value not defined in loop. Simplify def stmt
28499 2017-06-11 Tom de Vries <tom@codesourcery.com>
28502 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28504 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28505 nvptx_cannot_force_const_mem.
28507 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28509 * opts.c (finish_options): Move test for flag_split_stack after
28510 it has been initialized.
28512 2017-06-11 Jason Merrill <jason@redhat.com>
28514 * tree.h (id_equal): New.
28515 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28516 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28517 instead of strcmp of IDENTIFIER_POINTER.
28519 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28521 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28522 (mark_all_inlined_calls_cdtor): Fix formating.
28523 (inline_transform): Rescale profile before inlining.
28525 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28527 * cgraph.h (cgraph_edge::clone): Update prototype.
28528 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28529 (cgraph_node::create_clone): Update.
28530 (cgraph_node::create_version_clone): Update.
28531 * tree-inline.c (copy_bb): Update.
28532 (expand_call_inline): Update.
28534 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28536 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28537 factored out from ...
28538 (rs6000_emit_prologue): ... here.
28540 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28542 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28543 factored out from ...
28544 (rs6000_emit_prologue): ... here.
28546 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28548 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28550 (handle_missing_profiles): Fix computation of tp_first_run.
28551 (counts_to_freqs): Do not touch freqs when count is 0.
28553 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28555 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28558 2017-06-10 Tom de Vries <tom@codesourcery.com>
28560 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28561 attributes): Document signal effective target.
28563 2017-06-10 Tom de Vries <tom@codesourcery.com>
28565 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28566 Document effective target stack_size.
28568 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28570 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28571 to the edit_context if they can be auto-applied.
28573 2017-06-9 Ian Lance Taylor <iant@golang.org>
28575 * opts.c (finish_options): If -fsplit-stack, disable implicit
28576 -forder-blocks-and-partition.
28577 * doc/invoke.texi (Optimize Options): Document that when using
28578 -fsplit-stack -forder-blocks-and-partition is not implicitly
28581 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28583 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28584 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28585 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28586 * builtins.def (abort, trap, unreachable): Declare cold.
28587 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28588 * tree-core.h (ECF_COLD): New.
28589 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28590 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28592 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28594 * predict.c (unlikely_executed_stmt_p): Cleanup.
28596 2017-06-09 Richard Biener <rguenther@suse.de>
28598 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28599 model if the ref is always written to.
28601 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28603 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28605 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28607 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28610 2017-06-09 Tom de Vries <tom@codesourcery.com>
28613 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28614 "target cannot support label values" when encountering LABEL_REF.
28616 2017-06-09 Martin Liska <mliska@suse.cz>
28618 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28619 (gimple_gen_ic_func_profiler): Emit direct comparison
28620 of __gcov_indirect_call_callee with NULL.
28621 (gimple_gen_time_profiler): Change probability from
28622 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28624 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28626 * profile.c (edge_gcov_counts): Turn to pointer.
28627 (compute_branch_probabilities, compute_branch_probabilities): Update.
28628 (branch_prob): Do not clear edge_gcov_count.
28629 * profile.h (edge_gcov_counts): Turn to pointer.
28630 (edge_gcov_count): Update.
28632 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28634 * gimple.h (gimple_check_failed): Mark cold.
28636 2017-06-09 Richard Biener <rguenther@suse.de>
28638 PR tree-optimization/66623
28639 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28640 refactor check_reduction into two parts, properly computing
28641 whether we have to check reduction validity for outer loop
28644 2017-06-09 Richard Biener <rguenther@suse.de>
28646 PR tree-optimization/79483
28647 * graphite-scop-detection.c (order): New global.
28648 (get_order): Compute bb to order mapping that satisfies code
28649 generation constraints.
28650 (cmp_pbbs): New helper.
28651 (build_scops): Start domwalk at entry block, sort generated
28654 2017-06-09 Richard Biener <rguenther@suse.de>
28656 PR middle-end/81007
28657 * ipa-polymorphic-call.c
28658 (ipa_polymorphic_call_context::restrict_to_inner_class):
28659 Skip FIELD_DECLs with error_mark_node type.
28660 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28663 2017-06-09 Martin Liska <mliska@suse.cz>
28665 * predict.c (struct branch_predictor): New struct.
28666 (test_prediction_value_range): New test.
28667 (predict_c_tests): New function.
28668 * selftest-run-tests.c (selftest::run_tests): Run the function.
28669 * selftest.h: Declare new tests.
28671 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28674 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28675 gen_add3_insn did not fail.
28676 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28677 r0, construct that number in a temporary reg and add that reg to r0.
28678 If asked to put the result in r0 as well, fail.
28680 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28682 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28683 for early expansion of vec_eqv.
28685 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28687 PR middle-end/81005
28688 * ubsan.c (instrument_null): Avoid pointless code temporary.
28689 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28691 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28695 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28698 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28701 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28703 * predict.c (unlikely_executed_edge_p): Move ahead.
28704 (probably_never_executed_edge_p): Use it.
28706 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28708 PR middle-end/79988
28709 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28710 gimple_call_builtin_p call.
28712 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28714 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28715 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28716 rtl_check_failed_type2, rtl_check_failed_code1,
28717 rtl_check_failed_code2, rtl_check_failed_code_mode,
28718 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28719 rtvec_check_failed_bounds, rtl_check_failed_flag,
28720 _fatal_insn_not_found, _fatal_insn): Likewise.
28721 * tree.h (tree_contains_struct_check_failed,
28722 tree_check_failed, tree_not_check_failed,
28723 tree_class_check_failed, tree_range_check_failed,
28724 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28725 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28726 tree_operand_check_failed, omp_clause_check_failed,
28727 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28730 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28732 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28733 flag_branch_probabilities.
28734 * ipa-inline.c (edge_badness): Likewise.
28735 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28736 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28737 * predict.c (maybe_hot_frequency_p): Likewise.
28738 (probably_never_executed): Likewise.
28739 * sched-ebb.c (schedule_ebbs): Likewise.
28740 * sched-rgn.c (find_single_block_region): Likewise.
28741 * tracer.c (tail_duplicate): Likewise.
28743 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28745 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28746 longer requires x_flag_profile_use.
28748 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28750 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28751 instead of flag_reorder_blocks_and_partition.
28752 * dbxout.c (dbxout_function_end): Likewise.
28753 * dwarf2out.c (gen_subprogram_die): Likewise.
28754 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28755 * hw-doloop.c (reorg_loops): Likewise.
28756 * varasm.c (assemble_start_function,
28757 assemble_end_function): Likewise.
28758 (decide_function_section): Do not check for
28759 flag_reorder_blocks_and_partition.
28761 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28763 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28765 (chkp_get_hard_register_fake_addr_expr): Ditto.
28766 (chkp_build_addr_expr): Add check for hard reg case.
28767 (chkp_parse_array_and_component_ref): Ditto.
28768 (chkp_find_bounds_1): Ditto.
28769 (chkp_process_stmt): Don't generate bounds store for
28772 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28774 * predict.c (maybe_hot_bb_p): Do not check profile status.
28775 (maybe_hot_edge_p): Likewise.
28776 (probably_never_executed): Check for zero counts even if profile
28778 (unlikely_executed_edge_p): New function.
28779 (unlikely_executed_stmt_p): New function.
28780 (unlikely_executed_bb_p): New function.
28781 (set_even_probabilities): Use unlikely predicates.
28782 (combine_predictions_for_bb): Likewise.
28783 (predict_paths_for_bb): Likewise.
28784 (predict_paths_leading_to_edge): Likewise.
28785 (determine_unlikely_bbs): New function.
28786 (estimate_bb_frequencies): Use it.
28787 (compute_function_frequency): Use zero counts even if profile is
28789 * profile-count.h: Fix typo.
28791 2017-08-08 Julia Koval <julia.koval@intel.com>
28793 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28794 _mm512_mask_cvtsepi16_storeu_epi8,
28795 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28796 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28797 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28798 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28799 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28800 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28801 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28802 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28803 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28804 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28805 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28806 __builtin_ia32_pmovuswb256mem_mask,
28807 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28808 __builtin_ia32_pmovwb512mem_mask): New builtins.
28810 2017-08-08 Julia Koval <julia.koval@intel.com>
28812 PR target/73350,80862
28813 * config/i386/subst.md (round): Fix round pattern.
28814 * config/i386/i386.c (ix86_erase_embedded_rounding):
28815 Fix erasing rounding for the fixed pattern.
28817 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28819 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28821 2017-06-08 Martin Liska <mliska@suse.cz>
28823 PR gcov-profile/80911
28824 * gcov.c (block_info::block_info): New constructor.
28826 2017-06-07 Carl Love <cel@us.ibm.com>
28828 * config/rs6000/rs6000-c: The return type of the following
28829 built-in functions was implemented as int not long long. Fix sign
28830 of return value for the unsigned version of vec_mulo and vec_mule.
28831 vector unsigned long long vec_bperm (vector unsigned long long,
28832 vector unsigned char)
28833 vector signed long long vec_mule (vector signed int,
28835 vector unsigned long long vec_mule (vector unsigned int,
28836 vector unsigned int)
28837 vector signed long long vec_mulo (vector signed int,
28839 vector unsigned long long vec_mulo (vector unsigned int,
28840 vector unsigned int)
28841 * doc/extend.texi: Fix the documentation for the built-in
28844 2017-06-07 Carl Love <cel@us.ibm.com>
28847 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28850 2017-06-07 Carl Love <cel@us.ibm.com>
28852 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28853 support, Generate doublehv for signed int/float for BE case only.
28855 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28857 * doc/invoke.texi (mcx16): Rewrite.
28859 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28861 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28862 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28863 *mov<mode>_softfloat, and an anonymous splitter): Use
28864 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28866 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28868 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28870 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28871 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28872 (rs6000_debug_reg_global): Adjust.
28873 (rs6000_init_hard_regno_mode_ok): Adjust.
28874 (rs6000_dbx_register_number): Adjust.
28875 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28876 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28877 Remove SPE_ACC and SPEFSCR.
28878 (REG_ALLOC_ORDER): Ditto.
28879 (FRAME_POINTER_REGNUM): Change to 111.
28880 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28881 (REG_CLASS_NAMES): Ditto.
28882 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28883 (REGISTER_NAMES): Ditto.
28884 (ADDITIONAL_REG_NAMES): Ditto.
28885 (rs6000_reg_names): Ditto.
28886 * config/rs6000/rs6000.md: Renumber some register number
28889 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28891 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28893 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28894 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28896 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28897 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28898 Delete the SPE high registers.
28899 (REG_ALLOC_ORDER): Ditto.
28900 (enum reg_class): Remove SPE_HIGH_REGS.
28901 (REG_CLASS_NAMES): Ditto.
28902 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28903 (REGISTER_NAMES): Ditto.
28904 (rs6000_reg_names): Ditto.
28905 * doc/tm.texi.in: Remove SPE as example.
28906 * doc/tm.texi: Regenerate.
28908 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28910 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28911 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28912 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28913 * config/rs6000/rs6000.md (type): Remove "brinc".
28915 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28917 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28918 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28919 * config/rs6000/linuxspe.h: Delete file.
28920 * config/rs6000/rs6000.md: Don't include spe.md.
28921 * config/rs6000/spe.h: Delete file.
28922 * config/rs6000/spe.md: Delete file.
28923 * config/rs6000/t-rs6000: Remove spe.md.
28925 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28927 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28928 (reg_or_none500mem_operand): Delete.
28929 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28930 instead of reg_or_none500mem_operand.
28932 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28934 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28935 handling of SPE flags.
28936 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28938 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28940 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28942 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28943 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28944 paired_divv2sf3): Similar.
28945 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28946 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28947 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28949 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28950 Rename the paired_* instruction patterns.
28951 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28953 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28954 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28955 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28956 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28957 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28958 PAIRED_VECTOR_MODE.
28959 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28960 (spe_func_has_64bit_regs_p): Delete.
28961 (spe_expand_predicate_builtin): Delete.
28962 (spe_expand_evsel_builtin): Delete.
28963 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28964 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28965 (invalid_e500_subreg): Delete.
28966 (rs6000_legitimize_address): Always force_reg op2 as well, for
28967 paired single memory accesses.
28968 (rs6000_member_type_forces_blk): Delete.
28969 (rs6000_spe_function_arg): Delete.
28970 (rs6000_expand_unop_builtin): Delete SPE handling.
28971 (rs6000_expand_binop_builtin): Ditto.
28972 (spe_expand_stv_builtin): Delete.
28973 (bdesc_2arg_spe): Delete.
28974 (spe_expand_builtin): Delete.
28975 (spe_expand_predicate_builtin): Delete.
28976 (spe_expand_evsel_builtin): Delete.
28977 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28978 (spe_init_builtins): Delete.
28979 (spe_func_has_64bit_regs_p): Delete.
28980 (savres_routine_name): Delete "info" parameter. Adjust callers.
28981 (rs6000_emit_stack_reset): Ditto.
28982 (rs6000_dwarf_register_span): Delete.
28983 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28984 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28985 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28987 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28989 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28990 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28991 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28992 mulv2sf3, divv2sf3): Delete expanders.
28994 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28996 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28998 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29000 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29001 * config/rs6000/rs6000.c: Ditto.
29003 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29005 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29006 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29007 comparison_operator.
29009 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29011 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29012 * config/rs6000/rs6000.opt: Ditto.
29013 * config/rs6000/t-rtems: Ditto.
29015 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29017 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29018 TARGET_E500_SINGLE by 0, simplify.
29019 * config/rs6000/rs6000.c: Ditto.
29020 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29021 (spe_build_register_parallel): Delete.
29022 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29023 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29024 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29025 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29026 (E500_CONVERT): Delete.
29027 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29029 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29031 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29032 * config/rs6000/dfp.md: Ditto.
29033 (negdd2, *negdd2_fpr): Merge.
29034 (absdd2, *absdd2_fpr): Merge.
29035 (negtd2, *negtd2_fpr): Merge.
29036 (abstd2, *abstd2_fpr): Merge.
29037 * config/rs6000/e500.h: Delete file.
29038 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29039 TARGET_FPRS by 1 and simplify.
29040 * config/rs6000/rs6000-c.c: Ditto.
29041 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29042 TARGET_DF_SPE by 0.
29043 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29045 * config/rs6000/rs6000.md: Ditto.
29046 (floatdidf2, *floatdidf2_fpr): Merge.
29047 (move_from_CR_gt_bit): Delete.
29048 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29049 (E500_CR_IOR_COMPARE): Delete.
29050 (All patterns that require !TARGET_FPRS): Delete.
29051 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29053 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29055 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29057 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29059 * graphds.c (add_edge): Intitialize edge's attached data.
29060 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29061 pointer parameter. Call pointed function on each edge during
29062 graph traversing. Skip traversing the edge when the function
29064 (graphds_dfs, graphds_scc): Ditto.
29065 (for_each_edge): New parameter. Pass the new parameter to callback
29067 * graphds.h (skip_edge_callback): New function pointer type.
29068 (graphds_dfs, graphds_scc): New function pointer parameter.
29069 (graphds_edge_callback, for_each_edge): New parameter.
29071 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29073 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29074 out code checking if runtime alias check is possible to below ...
29075 Call the new function.
29076 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29077 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29079 2017-06-07 Marek Polacek <polacek@redhat.com>
29082 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29083 TYPE_OVERFLOW_WRAPS check.
29085 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29087 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29088 if versioning is required.
29089 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29090 peeling with the check for versioning.
29092 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29094 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29095 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29096 Set true to new parameter if new ssa variable is defined.
29097 (vect_gen_vector_loop_niters): Refactor. Set range information
29098 for the new vector loop bound variable.
29099 (vect_do_peeling): Ditto.
29101 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29103 * tree-affine.c (ssa.h): Include header file.
29104 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29105 has wrapping overflow behavior.
29107 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29109 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29111 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29113 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29114 (tree_to_aff_combination): ... here.
29116 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29118 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29119 reg_pressure model function.
29120 (ivopts_global_cost_for_size): Delete.
29121 (determine_set_costs, iv_ca_recount_cost): Call new model function
29122 ivopts_estimate_reg_pressure.
29124 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29126 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29127 expensive than udiv. Remove floating point cases from mod.
29129 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29131 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29132 Increase idiv cost.
29134 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29136 * config/aarch64/aarch64.md
29137 (copysignsf3): Fix mask generation.
29139 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29141 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29143 (class dump_manager): Add register_dumps method.
29144 * dumpfile.c: Include langhooks.h.
29145 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29146 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29147 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29148 (dump_manager::dump_register): Start with 512 entries instead of 32.
29149 (dump_manager::register_dumps): New method.
29150 * toplev.c (general_init): Instead of invoking register_dumps
29151 langhook, invoke register_dumps method on the dump manager.
29152 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29155 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29157 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29158 Say that # requires an associated define_split to exist, and that
29159 the define_split must be suitable for use after register allocation.
29161 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29163 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29164 (compute_outgoing_frequencies): Also initialize zero counts.
29165 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29166 around loops; preserve more of profile when nothing changes.
29168 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29170 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29172 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29173 * config/arm/arm-cpu-cdata.h: Regenerate.
29174 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29175 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29176 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29177 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29178 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29180 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29182 * config/arm/t-rmprofile: Likewise.
29183 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29185 2017-06-06 David S. Miller <davem@davemloft.net>
29188 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29189 blockage if function uses alloca.
29191 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29193 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29194 New "uid" fields to hold pretty-print IDs of group and ref.
29195 Memory references are now identified as <group_id>:<ref_id>
29196 instead of using [random] addresses.
29197 (dump_mem_details): Simplify, no functional change.
29198 (dump_mem_ref): Simplify and make output more concise.
29199 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29200 (find_or_create_group): Initialize group uid.
29201 (record_ref): Initialize ref uid. Improve debug output.
29202 (prune_group_by_reuse, should_issue_prefetch_p,)
29203 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29204 (mark_nontemporal_store, determine_loop_nest_reuse):
29205 Improve debug output.
29207 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29209 * dbgcnt.def (prefetch): New debug counter.
29210 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29211 (schedule_prefetches): Stop issueing prefetches if debug counter
29214 2017-06-06 Tom de Vries <tom@codesourcery.com>
29216 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29217 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29219 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29221 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29222 Use aarch64_reg_or_zero predicate for operand 4.
29223 (aarch64_compare_and_swap<mode> define_insn_and_split):
29224 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29225 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29227 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29229 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29230 (arm_compute_save_core_reg_mask): This.
29231 (thumb1_compute_save_reg_mask): Rename into ...
29232 (thumb1_compute_save_core_reg_mask): This.
29233 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29234 (arm_compute_frame_layout): Likewise.
29236 2017-06-06 Richard Biener <rguenther@suse.de>
29238 PR tree-optimization/80974
29239 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29240 keep or clear leaders SSA info.
29242 2017-06-06 Tom de Vries <tom@codesourcery.com>
29244 * config/nvptx/nvptx.c (split_mode_p): New function.
29245 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29247 2017-06-06 Tom de Vries <tom@codesourcery.com>
29249 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29251 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29254 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29257 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29259 * shrink-wrap.c (handle_simple_exit): Update profile.
29260 (try_shrink_wrapping): Upate profile.
29262 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29264 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29265 (tree_guess_outgoing_edge_probabilities): New.
29266 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29267 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29269 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29271 * ipa-split.c (split_function): Initialize return bb profile.
29273 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29275 * profile.c (compute_branch_probabilities): Also initialize
29276 EXIT_BLOCK profile.
29278 2017-06-06 Richard Biener <rguenther@suse.de>
29280 PR tree-optimization/80928
29281 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29282 (vect_analyze_loop_operations): Properly guard analysis for
29284 (vect_transform_loop): Likewise.
29285 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29286 (vect_model_induction_cost): Do not cost for pure SLP.
29287 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29288 of induction in inner loop vectorization.
29289 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29290 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29291 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29293 (vect_analyze_slp_cost_1): Cost induction.
29294 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29295 (vect_get_slp_vect_defs): Likewise.
29296 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29297 (vect_transform_stmt): Handle SLP reductions.
29298 * tree-vectorizer.h (vectorizable_induction): Adjust.
29300 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29302 * config/rs6000/rs6000.c (make_resolver_func): Update
29303 init_lowered_empty_function call.
29305 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29307 * doc/invoke.texi: Document the -fprofile-abs-path option.
29308 * common.opt (fprofile-abs-path): New option.
29309 * gcov-io.h (gcov_write_filename): Declare.
29310 * gcov-io.c (gcov_write_filename): New function.
29311 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29312 * profile.c (output_location): Likewise.
29314 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29316 * shring-wrap.c: Revert accidental commit.
29318 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29320 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29322 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29324 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29326 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29327 profile in callgraph edge.
29328 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29329 (apply_scale): Likewise.
29330 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29331 Also scale profile when inlining function with zero profile.
29332 (initialize_cfun): Update exit block profile even when it is zero.
29333 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29334 when profile is read.
29336 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29338 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29339 (CLONE_*): New constants to define the processors we can generate
29340 code for with the target_clone attribute.
29341 (rs6000_clone_map): New array to identify which clone processors
29342 the current program is running on.
29343 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29344 target_clone attribute.
29345 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29346 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29347 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29348 (cpu_expand_builtin): Add support for target_clone attribute.
29349 (rs6000_valid_attribute_p): Allow "default" attribute.
29350 (get_decl_name): New debug function to simplify printing the
29351 current function name in debugging statements.
29352 (rs6000_clone_priority): New functions to support the target_clone
29353 attribute, and be able to generate code to switch between ISA 2.05
29354 through ISA 3.0 (power6 through power9).
29355 (rs6000_compare_version_priority): Likewise.
29356 (rs6000_get_function_versions_dispatcher): Likewise.
29357 (make_resolver_func): Likewise.
29358 (add_condition_to_bb): Likewise.
29359 (dispatch_function_versions): Likewise.
29360 (rs6000_generate_version_dispatcher_body): Likewise.
29361 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29362 (fusion_gpr_load_p): Fix a spacing issue.
29363 * doc/extend.texi (Common Function Attributes): Document that the
29364 PowerPC supports the target_clone attribute.
29366 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29368 * config/arm/arm.h: explain F symbol found in description of ARM
29369 register allocation in its legend.
29371 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29373 * config/mips/frame-header-opt.c: Include profile-count.h.
29374 * config/riscv/riscv.c: Include profile-count.h
29376 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29378 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29380 (sm_set_flag_if_changed): Add bbs field.
29381 (execute_sm_if_changed_flag_set): Pass BBS.
29382 (execute_sm): Update.
29384 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29386 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29389 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29391 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29392 (peephole2): New peephole2 to emit the above.
29393 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29395 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29397 * config/aarch64/aarch64.c (define_peephole2 above
29398 *sub_<shift>_<mode>): New peephole.
29400 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29402 * config/i386/i386.c (make_resolver_func): Update.
29403 * Makefile.in: Add profile-count.h and profile-count.o
29404 * auto-profile.c (afdo_indirect_call): Update to new API.
29405 (afdo_set_bb_count): Update.
29406 (afdo_propagate_edge): Update.
29407 (afdo_propagate_circuit): Update.
29408 (afdo_calculate_branch_prob): Update.
29409 (afdo_annotate_cfg): Update.
29410 * basic-block.h: Include profile-count.h
29411 (struct edge_def): Turn count to profile_count.
29412 (struct basic_block_def): Likewie.
29413 (REG_BR_PROB_BASE): Move to profile-count.h
29414 (RDIV): Move to profile-count.h
29415 * bb-reorder.c (max_entry_count): Turn to profile_count.
29416 (find_traces): Update.
29417 (rotate_loop):Update.
29418 (connect_traces):Update.
29419 (sanitize_hot_paths):Update.
29420 * bt-load.c (migrate_btr_defs): Update.
29421 * cfg.c (RDIV): Remove.
29422 (init_flow): Use alloc_block.
29423 (alloc_block): Uninitialize count.
29424 (unchecked_make_edge): Uninitialize count.
29425 (check_bb_profile): Update.
29426 (dump_edge_info): Update.
29427 (dump_bb_info): Update.
29428 (update_bb_profile_for_threading): Update.
29429 (scale_bbs_frequencies_int): Update.
29430 (scale_bbs_frequencies_gcov_type): Update.
29431 (scale_bbs_frequencies_profile_count): New.
29432 * cfg.h (update_bb_profile_for_threading): Update.
29433 (scale_bbs_frequencies_profile_count): Declare.
29434 * cfgbuild.c (compute_outgoing_frequencies): Update.
29435 (find_many_sub_basic_blocks): Update.
29436 * cfgcleanup.c (try_forward_edges): Update.
29437 (try_crossjump_to_edge): Update.
29438 * cfgexpand.c (expand_gimple_tailcall): Update.
29439 (construct_exit_block): Update.
29440 * cfghooks.c (verify_flow_info): Update.
29441 (dump_bb_for_graph): Update.
29442 (split_edge): Update.
29443 (make_forwarder_block): Update.
29444 (duplicate_block): Update.
29445 (account_profile_record): Update.
29446 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29447 (get_estimated_loop_iterations): Update.
29448 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29449 (single_likely_exit): Update.
29450 * cfgloopmanip.c (scale_loop_profile): Update.
29452 (set_zero_probability): Update.
29453 (lv_adjust_loop_entry_edge): Update.
29454 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29455 (purge_dead_edges): Update.
29456 (rtl_account_profile_record): Update.
29457 * cgraph.c (cgraph_node::create): Uninitialize count.
29458 (symbol_table::create_edge): Uninitialize count.
29459 (cgraph_update_edges_for_call_stmt_node): Update.
29460 (cgraph_edge::dump_edge_flags): Update.
29461 (cgraph_node::dump): Update.
29462 (cgraph_edge::maybe_hot_p): Update.
29463 * cgraph.h: Include profile-count.h
29464 (create_clone), create_edge, create_indirect_edge): Update.
29465 (cgraph_node): Turn count to profile_count.
29466 (cgraph_edge0: Likewise.
29467 (make_speculative, clone): Update.
29468 (create_edge): Update.
29469 (init_lowered_empty_function): Update.
29470 * cgraphclones.c (cgraph_edge::clone): Update.
29471 (duplicate_thunk_for_node): Update.
29472 (cgraph_node::create_clone): Update.
29473 * cgraphunit.c (cgraph_node::analyze): Update.
29474 (cgraph_node::expand_thunk): Update.
29475 * final.c (dump_basic_block_info): Update.
29476 * gimple-streamer-in.c (input_bb): Update.
29477 * gimple-streamer-out.c (output_bb): Update.
29478 * graphite.c (print_global_statistics): Update.
29479 (print_graphite_scop_statistics): Update.
29480 * hsa-brig.c: Include basic-block.h.
29481 * hsa-dump.c: Include basic-block.h.
29482 * hsa-gen.c (T sum_slice): Update.
29483 (convert_switch_statements):Update.
29484 * hsa-regalloc.c: Include basic-block.h.
29485 * ipa-chkp.c (chkp_produce_thunks): Update.
29486 * ipa-cp.c (struct caller_statistics): Update.
29487 (init_caller_stats): Update.
29488 (gather_caller_stats): Update.
29489 (ipcp_cloning_candidate_p): Update.
29490 (good_cloning_opportunity_p): Update.
29491 (get_info_about_necessary_edges): Update.
29492 (dump_profile_updates): Update.
29493 (update_profiling_info): Update.
29494 (update_specialized_profile): Update.
29495 (perhaps_add_new_callers): Update.
29496 (decide_about_value): Update.
29497 (ipa_cp_c_finalize): Update.
29498 * ipa-devirt.c (struct odr_type_warn_count): Update.
29499 (struct decl_warn_count): Update.
29500 (struct final_warning_record): Update.
29501 (possible_polymorphic_call_targets): Update.
29502 (ipa_devirt): Update.
29503 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29504 * ipa-icf.c (sem_function::merge): Update.
29505 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29506 * ipa-inline.c (compute_uninlined_call_time): Update.
29507 (compute_inlined_call_time): Update.
29508 (want_inline_small_function_p): Update.
29509 (want_inline_self_recursive_call_p): Update.
29510 (edge_badness): Update.
29511 (lookup_recursive_calls): Update.
29512 (recursive_inlining): Update.
29513 (inline_small_functions): Update.
29514 (dump_overall_stats): Update.
29515 (dump_inline_stats): Update.
29516 * ipa-profile.c (ipa_profile_generate_summary): Update.
29517 (ipa_propagate_frequency): Update.
29518 (ipa_profile): Update.
29519 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29520 * ipa-utils.c (ipa_merge_profiles): Update.
29521 * loop-doloop.c (doloop_modify): Update.
29522 * loop-unroll.c (report_unroll): Update.
29523 (unroll_loop_runtime_iterations): Update.
29524 * lto-cgraph.c (lto_output_edge): Update.
29525 (lto_output_node): Update.
29526 (input_node): Update.
29527 (input_edge): Update.
29528 (merge_profile_summaries): Update.
29529 * lto-streamer-in.c (input_cfg): Update.
29530 * lto-streamer-out.c (output_cfg): Update.
29531 * mcf.c (create_fixup_graph): Update.
29532 (adjust_cfg_counts): Update.
29533 (sum_edge_counts): Update.
29534 * modulo-sched.c (sms_schedule): Update.
29535 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29536 * predict.c (maybe_hot_count_p): Update.
29537 (probably_never_executed): Update.
29538 (dump_prediction): Update.
29539 (combine_predictions_for_bb): Update.
29540 (propagate_freq): Update.
29541 (handle_missing_profiles): Update.
29542 (counts_to_freqs): Update.
29543 (rebuild_frequencies): Update.
29544 (force_edge_cold): Update.
29545 * predict.h: Include profile-count.h
29546 (maybe_hot_count_p, counts_to_freqs): UPdate.
29547 * print-rtl-function.c: Do not include cfg.h
29548 * print-rtl.c: Include basic-block.h
29549 * profile-count.c: New file.
29550 * profile-count.h: New file.
29551 * profile.c (is_edge_inconsistent): Update.
29552 (correct_negative_edge_counts): Update.
29553 (is_inconsistent): Update.
29554 (set_bb_counts): Update.
29555 (read_profile_edge_counts): Update.
29556 (compute_frequency_overlap): Update.
29557 (compute_branch_probabilities): Update; Initialize and deinitialize
29559 (branch_prob): Update.
29560 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29561 (edge_gcov_count): New.
29562 (bb_gcov_count): New.
29563 * shrink-wrap.c (try_shrink_wrapping): Update.
29564 * tracer.c (better_p): Update.
29565 * trans-mem.c (expand_transaction): Update.
29566 (ipa_tm_insert_irr_call): Update.
29567 (ipa_tm_insert_gettmclone_call): Update.
29568 * tree-call-cdce.c: Update.
29569 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29570 (gimple_duplicate_sese_tail): Update.
29571 (gimple_account_profile_record): Update.
29572 (execute_fixup_cfg): Update.
29573 * tree-inline.c (copy_bb): Update.
29574 (copy_edges_for_bb): Update.
29575 (initialize_cfun): Update.
29576 (freqs_to_counts): Update.
29577 (copy_cfg_body): Update.
29578 (expand_call_inline): Update.
29579 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29580 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29581 (try_unroll_loop_completely): Update.
29582 (try_peel_loop): Update.
29583 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29584 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29585 * tree-ssa-loop-split.c (connect_loops): Update.
29586 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29587 * tree-ssa-reassoc.c (branch_fixup): Update.
29588 * tree-ssa-tail-merge.c (replace_block_by): Update.
29589 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29590 (compute_path_counts): Update.
29591 (update_profile): Update.
29592 (recompute_probabilities): Update.
29593 (update_joiner_offpath_counts): Update.
29594 (estimated_freqs_path): Update.
29595 (freqs_to_counts_path): Update.
29596 (clear_counts_path): Update.
29597 (ssa_fix_duplicate_block_edges): Update.
29598 (duplicate_thread_path): Update.
29599 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29600 (struct switch_conv_info): Update.
29601 * tree-tailcall.c (decrease_profile): Update.
29602 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29603 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29604 * value-prof.c (check_counter): Update.
29605 (gimple_divmod_fixed_value): Update.
29606 (gimple_mod_pow2): Update.
29607 (gimple_mod_subtract): Update.
29608 (gimple_ic_transform): Update.
29609 (gimple_stringop_fixed_value): Update.
29610 * value-prof.h (gimple_ic): Update.
29612 2017-06-02 Carl Love <cel@us.ibm.com>
29614 * config/rs6000/rs6000-c: Add support for built-in functions
29615 vector double vec_doublee (vector signed int);
29616 vector double vec_doublee (vector unsigned int);
29617 vector double vec_doublee (vector float);
29618 vector double vec_doubleh (vector signed int);
29619 vector double vec_doubleh (vector unsigned int);
29620 vector double vec_doubleh (vector float);
29621 vector double vec_doublel (vector signed int);
29622 vector double vec_doublel (vector unsigned int);
29623 vector double vec_doublel (vector float);
29624 vector double vec_doubleo (vector signed int);
29625 vector double vec_doubleo (vector unsigned int);
29626 vector double vec_doubleo (vector float);.
29627 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29628 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29630 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29631 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29632 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29634 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29635 vec_doublel, vec_doubleh.
29636 * doc/extend.texi: Update the built-in documentation file for the
29637 new built-in functions.
29639 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29642 * ipa-inline-analysis.c (free_growth_caches): Set
29643 edge_removal_hook_holder to NULL after removing it.
29645 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29647 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29648 comparision with zero.
29650 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29651 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29652 for early expansion of vec_min and vec_max builtins.
29653 (builtin_function_type): Add min/max unsigned variants to those
29654 identified as having unsigned arguments.
29656 2017-06-02 Olivier Hainque <hainque@adacore.com>
29658 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29660 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29662 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29663 Use VALL_F16 iterator rather than VALL.
29665 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29667 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29668 Emit CBNZ inside loop when doing a strong exchange and comparing
29669 against zero. Generate the CC flags after the loop.
29671 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29673 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29674 (dl_section_ref): New.
29675 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29676 On AIX, append an expression to subtract the size of the
29677 section length to dl_section_ref.
29679 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29681 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29682 for early expansion of vector absolute builtins.
29684 2017-06-02 Richard Biener <rguenther@suse.de>
29686 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29687 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29689 2017-06-02 Richard Biener <rguenther@suse.de>
29691 PR tree-optimization/80948
29692 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29693 stmt order as well.
29695 2017-06-02 Richard Biener <rguenther@suse.de>
29697 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29699 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29700 for inductions as relevant.
29702 2017-06-02 Richard Biener <rguenther@suse.de>
29704 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29705 (vectorizable_induction): ... this. Remove dead code.
29707 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29709 * builtins. (expand_builtin_alloca): Remove second parameter and
29710 infer its value from the first parameter instead.
29711 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29713 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29715 PR rtl-optimization/80903
29716 * loop-doloop.c (add_test): Unshare sequence.
29718 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29720 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29722 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29724 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29726 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29727 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29728 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29729 (xlogue_layout::get_stub_rtx): Make static.
29730 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29731 (xlogue_layout::compute_stub_managed_regs): Rename to...
29732 (xlogue_layout::count_stub_managed_regs): ...this.
29733 (xlogue_layout::is_stub_managed_reg): New function.
29734 (xlogue_layout::m_stub_names): Rename to...
29735 (xlogue_layout::s_stub_names): ...this, make static.
29736 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29737 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29738 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29739 xlogue_layout::s_stub_names): Instantiate statics.
29740 (stub_managed_regs): Remove.
29741 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29742 (disable_call_ms2sysv_xlogues): Rename to...
29743 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29744 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29746 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29747 change after reload_completed.
29748 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29750 (ix86_expand_prologue): Likewise.
29751 (ix86_expand_epilogue): Likewise.
29752 (ix86_expand_split_stack_prologue): Likewise.
29753 (ix86_compute_frame_layout): Remove frame parameter ...
29754 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29755 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29757 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29758 (ix86_frame): Move from here ...
29759 * config/i386/i386.h (ix86_frame): ... to here.
29760 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29761 complete ix86_frame data structure instead. Remove some_ld_name.
29763 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29765 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29766 symbols that hold a DECL_VALUE_EXPR.
29768 2017-06-01 Martin Jambor <mjambor@suse.cz>
29770 PR tree-optimization/80898
29771 * tree-sra.c (process_subtree_disqualification): Removed.
29772 (disqualify_candidate): Do not acll
29773 process_subtree_disqualification.
29774 (subtree_mark_written_and_enqueue): New function.
29775 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29776 RHS has been disqualified and re-queue LHS if necessary. Apart
29777 from that, ignore disqualified RHS.
29779 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29781 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29782 address fetch for z10 or later.
29784 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29786 * config/arc/arc.md (tst_movb): Add guard when splitting.
29788 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29790 * config/arc/arc.c (arc_can_eliminate): Test against
29791 arc_frame_pointer_needed.
29793 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29795 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29796 to prevent store reordering.
29797 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29798 (type): Add block type.
29799 (stack_tie): Define special instruction to be used in
29802 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29804 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29805 constraint. It is not valid for the pattern.
29806 (noncommutative_binary_comparison): Likewise.
29808 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29810 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29813 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29815 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29816 be used by the reg-alloc.
29818 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29820 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29821 reg-alloc when having mul64 or mul32x16 instructions.
29822 (mulsidi3): Likewise.
29823 (umulsidi3): Likewise.
29824 (mulsi32x16): New pattern.
29825 (mulsi64): Likewise.
29826 (mulsidi64): Likewise.
29827 (umulsidi64): Likewise.
29828 (MUL32x16_REG): Define.
29829 (mul64_600): Use MUL32x16_REG.
29830 (mac64_600): Likewise.
29831 (umul64_600): Likewise.
29832 (umac64_600): Likewise.
29834 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29836 * config/arc/arc.md (mulsi3_700): Make it commutative.
29838 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29840 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29842 (*sign_extendsidi2_insn): Likewise for movstosw.
29844 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29846 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29847 the type of the input discriminant value. Convert the
29848 discriminant value of signedness vary.
29850 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29852 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29853 Add to -Wall section.
29855 2017-06-01 Richard Biener <rguenther@suse.de>
29857 PR middle-end/66313
29858 * fold-const.c (fold_plusminus_mult_expr): If the factored
29859 factor may be zero use a wrapping type for the inner operation.
29860 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29861 and handle moved defs.
29862 (process_assignment): Properly guard the unary op case. Return a
29863 tri-state indicating that moving the stmt before the call may allow
29864 to continue. Pass through to_move.
29865 (find_tail_calls): Handle moving unrelated defs before
29868 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29871 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29872 splitter result in the canonical way.
29874 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29876 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29877 also for 32bit target. Update insn attributes.
29878 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29880 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29882 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29883 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29884 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29885 (_mm_maskz_min_ss): New intrinsics.
29887 2017-05-31 Martin Liska <mliska@suse.cz>
29889 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29890 Change comment style to one we normally use.
29891 (vectorizable_reduction): Likewise.
29892 (vectorizable_induction): Likewise.
29893 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29894 (vectorizable_call): Likewise.
29895 (vectorizable_simd_clone_call): Likewise.
29896 (vectorizable_conversion): Likewise.
29897 (vectorizable_assignment): Likewise.
29898 (vectorizable_shift): Likewise.
29899 (vectorizable_operation): Likewise.
29900 (vectorizable_store): Likewise.
29901 (vectorizable_load): Likewise.
29902 * tree-vectorizer.h: Likewise.
29904 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29906 * passes.c (emergency_dump_function): New.
29907 * tree-pass.h (emergency_dump_function): Declare.
29908 * plugin.c (plugins_internal_error_function): Remove.
29909 * plugin.h (plugins_internal_error_function): Remove declaration.
29910 * toplev.c (internal_error_function): New static function. Use it...
29911 (general_init): ...here.
29913 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29915 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29916 (arc_rtx_costs): Add costs for new patterns.
29917 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29918 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29920 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29922 * tree-ssa-strlen.c (get_next_strinfo): New function.
29923 (get_stridx_plus_constant): Use it.
29924 (zero_length_string): Likewise.
29925 (adjust_related_strinfos): Likewise.
29926 (adjust_last_stmt): Likewise.
29928 2017-05-31 Richard Biener <rguenther@suse.de>
29931 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29932 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29934 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29936 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29937 loop_vinfo argument and use of dependence distance vectors.
29938 Check instead whether the two references differ only in their
29939 initial value and assume that they have the same alignment if the
29940 difference is a multiple of the vector alignment.
29941 (vect_analyze_data_refs_alignment): Update call accordingly.
29943 2017-05-31 Martin Liska <mliska@suse.cz>
29946 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29948 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29950 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29951 (create_intersect_range_checks): Move from ...
29952 * tree-data-ref.c (create_intersect_range_checks_index)
29953 (create_intersect_range_checks): ... to here.
29954 (create_runtime_alias_checks): New function factored from ...
29955 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29956 here. Call above function.
29957 * tree-data-ref.h (create_runtime_alias_checks): New function.
29959 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29961 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29962 segment length for dr_b and compute it in wide_int.
29964 2017-05-31 Richard Biener <rguenther@suse.de>
29966 PR tree-optimization/80906
29967 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29968 and pass through iv_map.
29969 (copy_bb_and_scalar_dependences): Adjust.
29970 (translate_pending_phi_nodes): Likewise.
29971 (copy_loop_close_phi_args): Handle code-generating IVs instead
29974 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29976 * diagnostic-color.c (color_dict): Add "type-diff".
29977 (parse_gcc_colors): Update comment.
29978 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29979 -fdiagnostics-show-template-tree and -fno-elide-type.
29980 (GCC_COLORS): Add type-diff to example.
29982 (-fdiagnostics-show-template-tree): New.
29983 (-fno-elide-type): New.
29984 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29985 the pp_format_decoder callback. Call any m_format_postprocessor's
29987 (pretty_printer::pretty_printer): Initialize
29988 m_format_postprocessor.
29989 (pretty_printer::~pretty_printer): Delete any
29990 m_format_postprocessor.
29991 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29992 (class format_postprocessor): New class.
29993 (struct pretty_printer::format_decoder): Document the new parameters.
29994 (struct pretty_printer::m_format_postprocessor): New field.
29995 * tree-diagnostic.c (default_tree_printer): Update for new
29996 bool and const char ** params.
29997 * tree-diagnostic.h (default_tree_printer): Likewise.
29999 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30001 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30002 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30003 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30004 rs6000_gen_cell_microcode code.
30005 (rs6000_final_prescan_insn): Delete.
30006 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30007 "warn-cell-microcode" entries.
30008 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30009 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30010 throughout. Change cc_reg_not_micro_cr0_operand to
30011 cc_reg_not_cr0_operand throughout.
30012 (*extendhi<mode>2_noload): Delete.
30013 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30014 (mwarn-cell-microcode): Delete.
30015 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30016 -mgen-cell-microcode and -mwarn-cell-microcode.
30018 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30021 * config/i386/constraints.md (Yd): New constraint.
30023 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30024 and (?Yd, r) alternatives. Update insn attributes.
30025 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30026 and (?*Yd, r) alternatives. Update insn attributes.
30027 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30029 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30031 * gimplify.c (gimplify_modify_expr): Don't create a
30032 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30035 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30037 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30039 2017-05-30 Richard Biener <rguenther@suse.de>
30041 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30042 and reduc_def fields.
30043 (STMT_VINFO_REDUC_TYPE): New define.
30044 (STMT_VINFO_REDUC_DEF): Likewise.
30045 (vect_force_simple_reduction): Adjust prototype.
30046 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30047 (vect_is_simple_reduction): Remove check_reduction argument.
30048 (vect_force_simple_reduction): Adjust and set
30049 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30050 (vectorizable_reduction): Do not re-do reduction analysis
30051 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30052 * tree-parloops.c (gather_scalar_reductions): Adjust.
30054 2017-05-30 Richard Biener <rguenther@suse.de>
30056 PR middle-end/80901
30057 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30060 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30062 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30063 Introduce unknown_misalignment parameter and remove vf.
30064 (vect_peeling_hash_get_lowest_cost):
30065 Pass unknown_misalignment parameter.
30066 (vect_enhance_data_refs_alignment):
30067 Fix unsupportable data ref treatment.
30069 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30071 * tree-vect-data-refs.c (vect_get_data_access_cost):
30072 Workaround for SLP handling.
30073 (vect_enhance_data_refs_alignment):
30074 Compute costs for doing no peeling at all, compare to the best
30075 peeling costs so far and avoid peeling if cheaper.
30077 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30079 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30080 Return peeling info and set costs to zero for unlimited cost
30082 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30083 with unknown misalignment. Compute and costs for unknown
30084 misalignment, compare them to the costs for known misalignment
30085 and choose the cheapest for peeling.
30087 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30089 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30090 (vect_get_peeling_costs_all_drs): Create function.
30091 (vect_peeling_hash_get_lowest_cost):
30092 Use vect_get_peeling_costs_all_drs.
30093 (vect_peeling_supportable): Create function.
30094 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30096 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30098 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30099 DR_HAS_NEGATIVE_STEP.
30100 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30101 (vect_enhance_data_refs_alignment): Use.
30102 (vect_duplicate_ssa_name_ptr_info): Use.
30103 * tree-vectorizer.h (dr_misalignment): Use.
30104 (known_alignment_for_access_p): Use.
30106 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30109 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30111 (has_section_name): New function.
30113 2017-05-30 Martin Liska <mliska@suse.cz>
30116 * auto-profile.c (get_function_decl_from_block): Fix
30119 2017-05-30 Richard Biener <rguenther@suse.de>
30121 PR middle-end/80876
30122 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30124 2017-05-30 Martin Liska <mliska@suse.cz>
30126 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30127 * dumpfile.h (struct dump_file_info): Remove ctors.
30129 2017-05-30 Martin Liska <mliska@suse.cz>
30131 * predict.def: Fix GNU coding style.
30133 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30135 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30136 Mark 'to' argument with ATTRIBUTE_UNUSED.
30138 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30140 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30141 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30142 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30145 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30147 * doc/install.texi (Options specification): Restore entry of
30148 --enable-sjlj-exceptions.
30150 2017-05-27 Michael Eager <eager@eagercon.com>
30153 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30155 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30157 * config/microblaze/microblaze.h
30158 (FIXED_REGISTERS): Update in macro.
30159 (CALL_USED_REGISTERS): Update in macro.
30161 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30163 * doc/install.texi: Add links to macOS binary distributions.
30165 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30169 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30171 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30173 2017-05-26 Martin Liska <mliska@suse.cz>
30175 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30177 2017-05-26 Martin Liska <mliska@suse.cz>
30179 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30180 always leading ';; '.
30181 (dump_bb_info): Likewise.
30182 (brief_dump_cfg): Likewise.
30183 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30184 * dumpfile.c: Remove usage of TDF_VERBOSE.
30185 * dumpfile.h (enum dump_kind): Likewise.
30186 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30187 * print-tree.c (debug_verbose): Remove.
30188 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30189 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30190 * tree-diagnostic.c (default_tree_printer): Replace
30191 TDF_DIAGNOSTIC with TDF_SLIM.
30193 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30195 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30196 in parameter loop, rather than loop_vinfo.
30197 (create_intersect_range_checks): Ditto.
30198 (vect_create_cond_for_alias_checks): Update call to above functions.
30200 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30202 PR tree-optimization/80815
30203 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30204 for merging runtime alias checks. Handle negative DR_STEPs.
30206 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30208 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30210 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30211 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30212 out code pruning runtime alias checks.
30213 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30214 factored out from above.
30215 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30217 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30219 (prune_runtime_alias_test_list): New decalaration.
30221 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30223 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30224 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30225 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30226 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30227 (operator==, comp_dr_with_seg_len_pair): Ditto.
30228 (vect_prune_runtime_alias_test_list): Ditto.
30230 2017-05-26 Martin Liska <mliska@suse.cz>
30233 * params.def: Bound partial-inlining-entry-probability param.
30235 2017-05-26 Marek Polacek <polacek@redhat.com>
30238 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30241 2017-05-26 Richard Biener <rguenther@suse.de>
30243 PR tree-optimization/80842
30244 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30247 2017-05-26 Richard Biener <rguenther@suse.de>
30249 PR tree-optimization/80844
30250 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30252 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30254 * doc/md.texi (Machine Constraints): Update x86 family
30255 machine constraints section to match 'config/i386/constraints.md'.
30257 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30259 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30261 2017-05-25 Nathan Sidwell <nathan@acm.org>
30263 * doc/invoke.texi (--enable-languages): Update documentation.
30265 2017-05-25 Martin Liska <mliska@suse.cz>
30267 * dumpfile.c: Add TDF_FOLDING.
30268 * dumpfile.h (enum dump_kind): Likewise.
30269 * genmatch.c (dt_simplify::gen_1): Use it.
30271 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30273 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30275 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30277 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30278 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30280 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30282 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30283 * match.pd (X == C): Rewrite it here.
30284 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30285 with_certain_nonzero_bits2): New predicates.
30286 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30288 2017-05-24 Nathan Sidwell <nathan@acm.org>
30290 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30293 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30296 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30298 * config/powerpcspe: New port. Files are copied from the rs6000
30299 port, with "rs6000" in filenames replaced by "powerpcspe".
30301 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30303 PR rtl-optimization/80754
30304 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30306 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30308 * config/sparc/sparc.md (length): Return the correct value for -mflat
30309 sibcalls to match output_sibcall.
30311 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30315 * config/rs6000/rs6000.c (struct machine_function): Add new field
30317 (rs6000_get_separate_components): Init that field, use it.
30318 (rs6000_components_for_bb): Use the field.
30320 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30322 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30324 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30326 PR middle-end/80823
30327 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30329 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30332 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30333 address against address_operand predicate.
30334 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30336 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30338 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30339 parameters passed indirectly.
30341 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30343 * config/i386/i386.md (*movdi_internal): Remove SSE4
30344 alternative 18 (?r, *v). Update insn attributes.
30345 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30346 Update insn attributes.
30347 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30348 Update insn attributes.
30349 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30350 alternative 1 (r, v). Remove isa attribute.
30351 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30352 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30353 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30355 2017-05-23 Tom de Vries <tom@codesourcery.com>
30357 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30360 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30362 * cgraphunit.c (symbol_table::process_new_functions): Update.
30363 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30364 (inline_generate_summary): Rename to ...
30365 (ipa_fn_summary_generate): ... this one.
30366 (inline_read_summary): Rename to ...
30367 (ipa_fn_summary_read): ... this one.
30368 (inline_write_summary): Rename to ...
30369 (ipa_fn_summary_write): ... this one.
30370 (inline_free_summary): Rename to ...
30371 (ipa_free_fn_summary): ... this one.
30372 (pass_data_local_fn_summary, pass_local_fn_summary,
30373 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30374 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30375 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30376 make_pass_ipa_fn_summary): New.
30377 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30378 inline_write_summary, inline_free_summary): Remove.
30379 (ipa_free_fn_summary) : New.
30380 * ipa-inline.c (ipa_inline): Update.
30381 (pass_ipa_inline): Do not generate summaries.
30382 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30384 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30385 and add pass_ipa_fn_summary.
30386 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30388 (make_pass_inline_parameters): Remove.
30390 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30392 * omp-low.c (struct omp_context): Remove "default_kind" member.
30395 * omp-offload.c (execute_oacc_device_lower): Remove the
30396 parallelism dimensions function attributes for unparallelized
30397 OpenACC kernels constructs.
30399 2017-05-23 Martin Liska <mliska@suse.cz>
30401 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30403 (cgraph_edge::make_speculative): Likewise.
30404 (cgraph_edge::resolve_speculation): Likewise.
30405 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30406 (cgraph_node::dump): Likewise.
30407 * cgraph.h: Likewise.
30408 * cgraphunit.c (analyze_functions): Likewise.
30409 (symbol_table::compile): Likewise.
30410 * ipa-cp.c (print_all_lattices): Likewise.
30411 (determine_versionability): Likewise.
30412 (initialize_node_lattices): Likewise.
30413 (ipcp_verify_propagated_values): Likewise.
30414 (estimate_local_effects): Likewise.
30415 (update_profiling_info): Likewise.
30416 (create_specialized_node): Likewise.
30417 (perhaps_add_new_callers): Likewise.
30418 (decide_about_value): Likewise.
30419 (decide_whether_version_node): Likewise.
30420 (identify_dead_nodes): Likewise.
30421 (ipcp_store_bits_results): Likewise.
30422 * ipa-devirt.c (dump_targets): Likewise.
30423 (ipa_devirt): Likewise.
30424 * ipa-icf.c (sem_item::dump): Likewise.
30425 (sem_function::equals): Likewise.
30426 (sem_variable::equals): Likewise.
30427 (sem_item_optimizer::read_section): Likewise.
30428 (sem_item_optimizer::execute): Likewise.
30429 (congruence_class::dump): Likewise.
30430 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30431 (dump_inline_summary): Likewise.
30432 (estimate_node_size_and_time): Likewise.
30433 (inline_analyze_function): Likewise.
30434 * ipa-inline-transform.c (inline_call): Likewise.
30435 * ipa-inline.c (report_inline_failed_reason): Likewise.
30436 (want_early_inline_function_p): Likewise.
30437 (edge_badness): Likewise.
30438 (update_edge_key): Likewise.
30439 (inline_small_functions): Likewise.
30440 * ipa-profile.c (ipa_profile): Likewise.
30441 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30442 (ipa_make_edge_direct_to_target): Likewise.
30443 (remove_described_reference): Likewise.
30444 (ipa_impossible_devirt_target): Likewise.
30445 (propagate_controlled_uses): Likewise.
30446 (ipa_print_node_params): Likewise.
30447 (ipcp_transform_function): Likewise.
30448 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30449 (propagate_pure_const): Likewise.
30450 * ipa-reference.c (generate_summary): Likewise.
30451 (read_write_all_from_decl): Likewise.
30452 (propagate): Likewise.
30453 (ipa_reference_read_optimization_summary): Likewise.
30454 * ipa-utils.c (ipa_merge_profiles): Likewise.
30455 * ipa.c (walk_polymorphic_call_targets): Likewise.
30456 (symbol_table::remove_unreachable_nodes): Likewise.
30457 (ipa_single_use): Likewise.
30458 * passes.c (execute_todo): Likewise.
30459 * predict.c (drop_profile): Likewise.
30460 * symtab.c (symtab_node::get_dump_name): New function.
30461 (symtab_node::dump_name): Likewise.
30462 (symtab_node::dump_asm_name): Likewise.
30463 (symtab_node::dump_references): Likewise.
30464 (symtab_node::dump_referring): Likewise.
30465 (symtab_node::dump_base): Likewise.
30466 (symtab_node::debug_symtab): Likewise.
30467 * tree-sra.c (convert_callers_for_node): Likewise.
30468 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30469 * value-prof.c (init_node_map): Likewise.
30471 2017-05-23 Martin Liska <mliska@suse.cz>
30473 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30474 and symtab_node::debug_symtab to symbol_table::debug.
30475 * cgraphunit.c (analyze_functions): Use the renamed function.
30476 (symbol_table::compile): Likewise.
30477 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30478 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30479 * passes.c (execute_todo): Likewise.
30480 * symtab.c (symbol_table::dump): New function.
30481 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30483 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30485 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30486 that nonconst implies exec.
30488 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30490 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30491 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30492 (inline_edge_summary_vec): Turn into ...
30493 (ipa_call_summaries): ... this one.
30494 (redirect_to_unreachable, edge_set_predicate,
30495 evaluate_properties_for_edge, inline_summary_alloc,
30496 reset_ipa_call_summary, reset_inline_summary,
30497 inline_summary_t::duplicate): Update.
30498 (inline_edge_duplication_hook): Turn to ...
30499 (ipa_call_summary_t::duplicate): ... this one.
30500 (inline_edge_removal_hook): Turn to ...
30501 (ipa_call_summary_t::remove): ... this one.
30502 (dump_inline_edge_summary): Turn to ...
30503 (dump_ipa_call_summary): ... this one.
30504 (estimate_function_body_sizes): Update.
30505 (inline_update_callee_summaries): Update.
30506 (remap_edge_change_prob): Update.
30507 (remap_edge_summaries): Update.
30508 (inline_merge_summary): Update.
30509 (do_estimate_edge_time): Update.
30510 (inline_generate_summary): Update.
30511 (inline_read_section): Update.
30512 (inline_read_summary): Update.
30513 (inline_free_summary): Update.
30514 * ipa-inline.c (can_inline_edge_p): Update.
30515 (compute_inlined_call_time): Update.
30516 (want_inline_small_function_p): Update.
30517 (edge_badness): Update.
30518 (early_inliner): Update.
30519 * ipa-inline.h (inline_edge_summary): Turn to ...
30520 (ipa_call_summary): ... this one.
30521 (ipa_call_summary_t): New class.
30522 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30523 (ipa_call_summaries): New.
30524 (inline_edge_summary): Remove.
30525 (estimate_edge_growth): Update.
30526 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30527 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30528 * ipa-split.c (execute_split_functions): Update.
30529 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30531 2017-05-23 Tom de Vries <tom@codesourcery.com>
30533 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30534 attributes): Document rdrand effective target.
30536 2017-05-23 Tom de Vries <tom@codesourcery.com>
30538 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30539 attributes): Sort alphabetically.
30541 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30543 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30545 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30548 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30549 V2DF/V2DI splat into two separate patterns, one that handles
30550 registers, and the other that only handles memory. Drop support
30551 for splatting from a GPR on ISA 2.07 and then splitting the
30552 splat into direct move and splat.
30553 (vsx_splat_<mode>_reg): Likewise.
30554 (vsx_splat_<mode>_mem): Likewise.
30556 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30558 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30560 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30562 PR middle-end/80809
30563 * omp-low.c (finish_taskreg_remap): New function.
30564 (finish_taskreg_scan): If unit size of ctx->record_type
30565 is non-constant, unshare the size expression and replace
30566 decls in it with possible outer var refs.
30568 PR middle-end/80809
30569 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30570 GOVD_SHARED rather than GOVD_PRIVATE with it.
30571 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30572 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30574 PR middle-end/80853
30575 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30576 as last argument to build_outer_var_ref for pointer bases of array
30577 section reductions.
30579 2017-05-19 Martin Sebor <msebor@redhat.com>
30581 * print-tree.c (print_node): Print DECL_READ_P flag.
30583 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30585 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30586 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30587 * cgraph.c: Likewise.
30588 * cgraphunit.c: Likewise.
30589 * gengtype.c: Likewise.
30590 * ipa-cp.c: Likewise.
30591 * ipa-devirt.c: Likewise.
30592 * ipa-icf.c: Likewise.
30593 * ipa-predicate.c: Likewise.
30594 * ipa-profile.c: Likewise.
30595 * ipa-prop.c: Likewise.
30596 * ipa-split.c: Likewise.
30598 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30599 edge_predicate_pool, dump_inline_hints,
30600 inline_summary::account_size_time, redirect_to_unreachable,
30601 edge_set_predicate, set_hint_predicate,
30602 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30603 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30604 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30605 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30606 ipa_call_summary_t::remove, initialize_growth_caches,
30607 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30608 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30609 mark_modified, unmodified_parm_1, unmodified_parm,
30610 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30611 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30612 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30613 will_be_nonconstant_predicate, record_modified_bb_info,
30614 get_minimal_bb, record_modified, param_change_prob,
30615 phi_result_unknown_predicate, predicate_for_phi_result,
30616 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30617 estimate_function_body_sizes, compute_inline_parameters,
30618 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30619 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30620 inline_update_callee_summaries, remap_edge_change_prob,
30621 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30622 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30623 inline_analyze_function, inline_summary_t::insert,
30624 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30625 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30626 inline_free_summary): Move to ipa-fnsummary.h
30627 (predicate_t): Remove.
30628 * ipa-fnsummary.c: New file.
30629 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30630 (enum inline_hints_vals, inline_hints, agg_position_info,
30631 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30632 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30633 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30634 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30635 inline_read_summary, inline_write_summary, inline_free_summary,
30636 inline_analyze_function, initialize_inline_failed,
30637 inline_merge_summary, inline_update_overall_summary,
30638 compute_inline_parameters): Move to ipa-fnsummary.h
30639 * ipa-fnsummary.h: New file.
30640 * ipa-inline-transform.h: Include ipa-inline.h.
30641 * ipa-inline.c: LIkewise.
30643 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30645 * ipa-inline.c (edge_badness): Use inlined_time instead of
30646 inline_summaries->get.
30648 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30650 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30652 2017-05-22 Nathan Sidwell <nathan@acm.org>
30654 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30655 (fdump-lang): Document 'raw' option.
30656 * dumpfile.h (TDI_tu): Delete.
30657 * dumpfile.c (dump_files): Remove translation-unit.
30658 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30660 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30662 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30663 command option from $(AWK) call.
30664 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30665 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30666 [FORMAT]: Remove handling of variable.
30667 * config/avr/t-multilib: Regenerate.
30669 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30671 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30673 (dump_inline_summary): Do not print self_time.
30674 (estimate_function_body_sizes): Do not set self_time.
30675 (compute_inline_parameters): Likewise.
30676 (inline_read_section, inline_write_summary): Do not stream self_time.
30677 * ipa-inline.h (inline_summary): Drop self_time.
30679 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30681 * ipa-inline-analysis.c (account_size_time): Rename to ...
30682 (inline_summary::account_size_time): ... this one.
30683 (reset_ipa_call_summary): Turn to ...
30684 (ipa_call_summary::reset): ... this one.
30685 (reset_inline_summary): Turn to ...
30686 (inline_summary::reset): ... this one.
30687 (inline_summary_t::remove): Update.
30688 (inline_summary_t::duplicate): Update.
30689 (ipa_call_summary_t::remove): Update.
30690 (dump_inline_summary): Update.
30691 (estimate_function_body_sizes): Update.
30692 (compute_inline_parameters): Update.
30693 (estimate_node_size_and_time): Update.
30694 (inline_merge_summary): Update.
30695 (inline_update_overall_summary): Update.
30696 (inline_read_section): Update.
30697 (inline_write_summary): Update.
30698 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30699 add account_size_time and reset member functions.
30700 (ipa_call_summary): Add reset function.
30701 * ipa-predicate.h (predicate::operator &): Constify.
30703 2017-05-22 Richard Biener <rguenther@suse.de>
30705 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30707 2017-05-19 Jason Merrill <jason@redhat.com>
30709 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30711 2017-05-19 Marek Polacek <polacek@redhat.com>
30714 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30715 TYPE_OVERFLOW_WRAPS checks.
30717 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30719 * tree-core.h (enum omp_clause_default_kind): Add
30720 "OMP_CLAUSE_DEFAULT_PRESENT".
30721 * tree-pretty-print.c (dump_omp_clause): Handle it.
30722 * gimplify.c (enum gimplify_omp_var_data): Add
30723 "GOVD_MAP_FORCE_PRESENT".
30724 (gimplify_adjust_omp_clauses_1): Map it to
30725 "GOMP_MAP_FORCE_PRESENT".
30726 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30728 * gimplify.c (oacc_default_clause): Clarify.
30730 2017-05-19 Nathan Sidwell <nathan@acm.org>
30732 LANG_HOOK_REGISTER_DUMPS
30733 * toplev.c (general_init): Call register dump lang hook.
30734 * doc/invoke.texi: Document -fdump-lang option family.
30735 * dumpfile.c (dump_files): Remove class dump here.
30736 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30737 * dumpfile.h (tree_dump_index): Remove TDI_class.
30738 * langhooks-def.h (lhd_register_dumps): Declare.
30739 (LANG_HOOKS_REGISTER_DUMPS): Define.
30740 (LANG_HOOKS_INITIALIZER): Add it.
30741 * langhooks.c (lhd_register_dumps): Define.
30742 * langhooks.h (struct lang_hooks): Add register_dumps.
30744 2017-05-19 Nathan Sidwell <nathan@acm.org>
30746 * context.h (context::set_passes): New.
30747 * context.c (context::context): Do not create pass manager.
30748 * toplev.c (general_init): Create pass manager here.
30750 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30752 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30753 use this splitter if two add or or instructions would also work for
30754 the constant we want to generate.
30756 2017-05-19 Richard Biener <rguenther@suse.de>
30759 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30760 predicate evaluation.
30762 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30764 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30766 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30769 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30771 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30772 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30773 (inline_edge_summary_vec): Turn into ...
30774 (ipa_call_summaries): ... this one.
30775 (redirect_to_unreachable, edge_set_predicate,
30776 evaluate_properties_for_edge, inline_summary_alloc,
30777 reset_ipa_call_summary, reset_inline_summary,
30778 inline_summary_t::duplicate): Update.
30779 (inline_edge_duplication_hook): Turn to ...
30780 (ipa_call_summary_t::duplicate): ... this one.
30781 (inline_edge_removal_hook): Turn to ...
30782 (ipa_call_summary_t::remove): ... this one.
30783 (dump_inline_edge_summary): Turn to ...
30784 (dump_ipa_call_summary): ... this one.
30785 (estimate_function_body_sizes): Update.
30786 (inline_update_callee_summaries): Update.
30787 (remap_edge_change_prob): Update.
30788 (remap_edge_summaries): Update.
30789 (inline_merge_summary): Update.
30790 (do_estimate_edge_time): Update.
30791 (inline_generate_summary): Update.
30792 (inline_read_section): Update.
30793 (inline_read_summary): Update.
30794 (inline_free_summary): Update.
30795 * ipa-inline.c (can_inline_edge_p): Update.
30796 (compute_inlined_call_time): Update.
30797 (want_inline_small_function_p): Update.
30798 (edge_badness): Update.
30799 (early_inliner): Update.
30800 * ipa-inline.h (inline_edge_summary): Turn to ...
30801 (ipa_call_summary): ... this one.
30802 (ipa_call_summary_t): New class.
30803 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30804 (ipa_call_summaries): New.
30805 (inline_edge_summary): Remove.
30806 (estimate_edge_growth): Update.
30807 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30808 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30809 * ipa-split.c (execute_split_functions): Update.
30810 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30812 2017-05-19 Richard Biener <rguenther@suse.de>
30814 PR middle-end/80764
30815 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30817 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30819 * config/rs6000/rs6000.c (struct machine_function): Add field
30820 fpr_is_wrapped_separately.
30821 (rs6000_get_separate_components): Use 64 components. Handle the
30822 new FPR components.
30823 (rs6000_components_for_bb): Handle the FPR components.
30824 (rs6000_emit_prologue_components): Handle the FPR components.
30825 (rs6000_emit_epilogue_components): Handle the FPR components.
30826 (rs6000_set_handled_components): Handle the FPR components.
30827 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30828 that are already separately shrink-wrapped.
30829 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30830 that are already separately shrink-wrapped.
30832 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30835 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30838 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30839 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30840 cases where the register allocator uses a move and an offsettable
30841 memory operation to/from a FPR register on ISA 2.06/2.07.
30842 (define_peephole2 for Altivec d-form store): Likewise.
30844 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30847 * config/i386/mmx.md (*mov<mode>_internal): Enable
30848 alternatives 11, 12, 13 and 14 also for 32bit targets.
30849 Remove alternatives 15, 16, 17 and 18.
30850 * config/i386/sse.md (vec_concatv2di): Change
30851 alternative (!x, *y) to (x, ?!*Yn).
30853 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30855 * dumpfile.h (enum dump_kind): Remove stray comma.
30857 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30859 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30860 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30861 predicate::num_conditions
30862 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30863 (CHANGED): turn into predicate::changed.
30864 (agg_position_info): Move to ipa-predicate.h
30865 (add_condition, predicate::add_clause, predicate::operator &=,
30866 predicate::or_with, predicate::evaluate, predicate::probability,
30867 dump_condition, dump_clause, predicate::dump,
30868 predicate::remap_after_duplication, predicate::remap_after_inlining,
30869 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30870 (evaluate_conditions_for_known_args): Update.
30871 (set_cond_stmt_execution_predicate): Update.
30872 * ipa-inline.h: Include ipa-predicate.h
30873 (condition, inline_param_summary, conditions, agg_position_info,
30874 predicate): Move to ipa-predicate.h
30875 * ipa-predicate.c: New file.
30876 * ipa-predicate.h: New file.
30878 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30880 * final.c (leaf_function_p): Check we are not in a sequence.
30882 2017-05-18 Martin Liska <mliska@suse.cz>
30884 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30885 * dumpfile.c (dump_register): Use new enum dump_kind.
30886 (get_dump_file_name): Likewise.
30887 (dump_enable_all): Likewise.
30888 (dump_switch_p_1): Likewise.
30889 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30890 * dumpfile.h (enum dump_kind): New enum type.
30891 (struct dump_file_info): Create constructor and
30892 format fields and comments.
30893 * passes.c (pass_manager::register_one_dump_file):
30895 * statistics.c (statistics_early_init): Likewise.
30896 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30897 TDF_TREE with TDF_SLIM.
30898 (gather_memory_references_ref): Likewise.
30900 2017-05-18 Martin Liska <mliska@suse.cz>
30902 * vec.h (struct vnull): Use it.
30904 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30906 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30907 (true_predicate, false_predicate, true_predicate_p,
30908 false_predicate_p): Remove.
30909 (single_cond_predicate, not_inlined_predicate): Turn to member function
30911 (add_condition): Update.
30912 (add_clause): Turn to...
30913 (predicate::add_clause): ... this one; update; allow passing NULL
30915 (and_predicates): Turn to ...
30916 (predicate::operator &=): ... this one.
30917 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30918 (or_predicates): Turn to ...
30919 (predicate::or_with): ... this one.
30920 (evaluate_predicate): Turn to ...
30921 (predicate::evaluate): ... this one.
30922 (predicate_probability): Turn to ...
30923 (predicate::probability): ... this one.
30924 (dump_condition): Update.
30925 (dump_predicate): Turn to ...
30926 (predicate::dump): ... this one.
30927 (account_size_time): Update.
30928 (edge_set_predicate): Update.
30929 (set_hint_predicate): UPdate.
30930 (evaluate_conditions_for_known_args): Update.
30931 (evaluate_properties_for_edge): Update.
30932 (remap_predicate_after_duplication): Turn to...
30933 (predicate::remap_after_duplication): ... this one.
30934 (remap_hint_predicate_after_duplication): Update.
30935 (inline_summary_t::duplicate): UPdate.
30936 (dump_inline_edge_summary): Update.
30937 (dump_inline_summary): Update.
30938 (set_cond_stmt_execution_predicate): Update.
30939 (set_switch_stmt_execution_predicate): Update.
30940 (compute_bb_predicates): Update.
30941 (will_be_nonconstant_expr_predicate): Update.
30942 (will_be_nonconstant_predicate): Update.
30943 (phi_result_unknown_predicate): Update.
30944 (predicate_for_phi_result): Update.
30945 (array_index_predicate): Update.
30946 (estimate_function_body_sizes): Update.
30947 (estimate_node_size_and_time): Update.
30948 (estimate_ipcp_clone_size_and_time): Update.
30949 (remap_predicate): Rename to ...
30950 (predicate::remap_after_inlining): ... this one.
30951 (remap_hint_predicate): Update.
30952 (inline_merge_summary): Update.
30953 (inline_update_overall_summary): Update.
30954 (estimate_size_after_inlining): Update.
30955 (read_predicate): Rename to ...
30956 (predicate::stream_in): ... this one.
30957 (read_inline_edge_summary): Update.
30958 (write_predicate): Rename to ...
30959 (predicate::stream_out): ... this one.
30960 (write_inline_edge_summary): Update.
30961 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30962 (clause_t): Turn to uint32_t
30963 (predicate): Turn to class; implement constructor and operators
30965 (size_time_entry): Update.
30966 (inline_summary): Update.
30967 (inline_edge_summary): Update.
30969 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30971 * fold-const.c (fold_binary_loc): Move transformation...
30972 * match.pd (C - X CMP X): ... here.
30974 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30976 * config/sparc/sparc.c (sparc_option_override): Set function
30977 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30978 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30980 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30982 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30984 2017-05-18 Marek Polacek <polacek@redhat.com>
30987 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30988 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30989 gimple_assign_load_p.
30991 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30993 PR middle-end/80692
30994 * real.c (do_compare): Give decimal_do_compare preference over
30995 comparing just the signs.
30997 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30999 * doc/md.texi (Canonicalization of Instructions): Describe the
31000 canonical form of instructions that inherently set a condition
31003 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31005 PR middle-end/80775
31006 * tree-cfg.c: Move deletion of unreachable case statements to after
31007 the merging of consecutive case labels.
31009 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31011 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31012 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31013 restoring of callee-saved registers.
31015 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31017 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31018 * config/visium/visium.c (single_set_and_flags): Likewise.
31019 * config/visium/visium.md (Substitutions): Likewise.
31021 2017-05-17 Martin Liska <mliska@suse.cz>
31023 * cfg.c: Introduce dump_flags_t type and
31024 use it instead of int type.
31026 * cfghooks.c: Likewise.
31027 * cfghooks.h (struct cfg_hooks): Likewise.
31028 * cfgrtl.c: Likewise.
31029 * cfgrtl.h: Likewise.
31030 * cgraph.c (cgraph_node::get_body): Likewise.
31031 * coretypes.h: Likewise.
31032 * domwalk.c: Likewise.
31033 * domwalk.h: Likewise.
31034 * dumpfile.c (struct dump_option_value_info): Likewise.
31035 (dump_enable_all): Likewise.
31036 (dump_switch_p_1): Likewise.
31037 (opt_info_switch_p): Likewise.
31038 * dumpfile.h (enum tree_dump_index): Likewise.
31039 (struct dump_file_info): Likewise.
31040 * genemit.c: Likewise.
31041 * generic-match-head.c: Likewise.
31042 * gengtype.c (open_base_files): Likewise.
31043 * gimple-pretty-print.c: Likewise.
31044 * gimple-pretty-print.h: Likewise.
31045 * graph.c (print_graph_cfg): Likewise.
31046 * graphite-scop-detection.c (dot_all_sese): Likewise.
31047 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31048 * loop-unroll.c (report_unroll): Likewise.
31049 * passes.c (pass_manager::register_one_dump_file): Likewise.
31050 * print-tree.c: Likewise.
31051 * statistics.c: Likewise.
31052 * tree-cfg.c: Likewise.
31053 * tree-cfg.h: Likewise.
31054 * tree-dfa.c: Likewise.
31055 * tree-dfa.h: Likewise.
31056 * tree-dump.c (dump_function): Likewise.
31057 * tree-dump.h (struct dump_info): Likewise.
31058 * tree-pretty-print.c: Likewise.
31059 * tree-pretty-print.h: Likewise.
31060 * tree-ssa-live.c: Likewise.
31061 * tree-ssa-live.h: Likewise.
31062 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31063 * tree-vect-loop.c: Likewise.
31064 * tree-vect-slp.c: Likewise.
31066 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31067 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31069 PR tree-optimization/80457
31070 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31071 of all arguments to a statement as scalar_to_vec operations.
31072 (vectorizable_call): Adjust call to vect_model_simple_cost for
31074 (vectorizable_conversion): Likewise.
31075 (vectorizable_assignment): Likewise.
31076 (vectorizable_shift): Likewise.
31077 (vectorizable_operation): Likewise.
31078 (vectorizable_comparison): Likewise.
31079 (vect_is_simple_cond): Record the def types for operands.
31080 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31081 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31082 for statement argument count.
31084 2017-05-16 Carl Love <cel@us.ibm.com>
31086 * config/rs6000/rs6000-c: Add support for built-in functions
31087 vector unsigned long long vec_bperm (vector unsigned long long,
31088 vector unsigned char)
31089 vector signed long long vec_mule (vector signed int,
31091 vector unsigned long long vec_mule (vector unsigned int,
31092 vector unsigned int)
31093 vector signed long long vec_mulo (vector signed int,
31095 vector unsigned long long vec_mulo (vector unsigned int,
31096 vector unsigned int)
31097 vector signed char vec_sldw (vector signed char,
31098 vector signed char,
31100 vector unsigned char vec_sldw (vector unsigned char,
31101 vector unsigned char,
31103 vector signed short vec_sldw (vector signed short,
31104 vector signed short,
31106 vector unsigned short vec_sldw (vector unsigned short,
31107 vector unsigned short,
31109 vector signed int vec_sldw (vector signed int,
31112 vector unsigned int vec_sldw (vector unsigned int,
31113 vector unsigned int,
31115 vector signed long long vec_sldw (vector signed long long,
31116 vector signed long long,
31118 vector unsigned long long vec_sldw (vector unsigned long long,
31119 vector unsigned long long,
31121 * config/rs6000/rs6000-c: Add support for built-in functions
31122 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31123 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31124 * doc/extend.texi: Update the built-in documentation for the
31125 new built-in functions.
31127 2017-05-16 Marek Polacek <polacek@redhat.com>
31131 * tree.c (save_expr): Don't fold the expression.
31133 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31135 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31136 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31137 and (?*y,m). Update insn attributes.
31139 2017-05-16 Martin Liska <mliska@suse.cz>
31141 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31142 flags argument of print_gimple_stmt, print_gimple_expr,
31143 print_generic_stmt and print_generic_expr.
31144 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31145 * coretypes.h: Likewise.
31146 * except.c (dump_eh_tree): Likewise.
31147 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31148 * gimple-pretty-print.h: Likewise.
31149 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31150 (backprop::push_to_worklist): Likewise.
31151 (backprop::pop_from_worklist): Likewise.
31152 (backprop::process_use): Likewise.
31153 (backprop::intersect_uses): Likewise.
31154 (note_replacement): Likewise.
31155 * gimple-ssa-store-merging.c
31156 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31157 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31158 (pass_store_merging::execute): Likewise.
31159 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31160 (ssa_base_cand_dump_callback): Likewise.
31161 (dump_incr_vec): Likewise.
31162 (replace_refs): Likewise.
31163 (replace_mult_candidate): Likewise.
31164 (create_add_on_incoming_edge): Likewise.
31165 (create_phi_basis): Likewise.
31166 (insert_initializers): Likewise.
31167 (all_phi_incrs_profitable): Likewise.
31168 (introduce_cast_before_cand): Likewise.
31169 (replace_one_candidate): Likewise.
31170 * gimplify.c (gimplify_expr): Likewise.
31171 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31172 (set_rename): Likewise.
31173 (rename_uses): Likewise.
31174 (copy_loop_phi_nodes): Likewise.
31175 (add_close_phis_to_merge_points): Likewise.
31176 (copy_loop_close_phi_args): Likewise.
31177 (copy_cond_phi_args): Likewise.
31178 (graphite_copy_stmts_from_block): Likewise.
31179 (translate_pending_phi_nodes): Likewise.
31180 * graphite-poly.c (print_pdr): Likewise.
31181 (dump_gbb_cases): Likewise.
31182 (dump_gbb_conditions): Likewise.
31183 (print_scop_params): Likewise.
31184 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31185 (build_cross_bb_scalars_use): Likewise.
31186 (gather_bbs::before_dom_children): Likewise.
31187 * hsa-dump.c (dump_hsa_immed): Likewise.
31188 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31189 (get_replacement_map): Likewise.
31190 * ipa-inline-analysis.c (dump_condition): Likewise.
31191 (estimate_function_body_sizes): Likewise.
31192 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31193 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31194 * ipa-prop.c (ipa_dump_param): Likewise.
31195 (ipa_print_node_jump_functions_for_edge): Likewise.
31196 (ipa_modify_call_arguments): Likewise.
31197 (ipa_modify_expr): Likewise.
31198 (ipa_dump_param_adjustments): Likewise.
31199 (ipa_dump_agg_replacement_values): Likewise.
31200 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31201 * ipa-pure-const.c (check_stmt): Likewise.
31202 (pass_nothrow::execute): Likewise.
31203 * ipa-split.c (execute_split_functions): Likewise.
31204 * omp-offload.c (dump_oacc_loop_part): Likewise.
31205 (dump_oacc_loop): Likewise.
31206 * trans-mem.c (tm_log_emit): Likewise.
31207 (tm_memopt_accumulate_memops): Likewise.
31208 (dump_tm_memopt_set): Likewise.
31209 (dump_tm_memopt_transform): Likewise.
31210 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31211 (print_loop): Likewise.
31212 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31213 (chkp_gather_checks_info): Likewise.
31214 (chkp_get_check_result): Likewise.
31215 (chkp_remove_check_if_pass): Likewise.
31216 (chkp_use_outer_bounds_if_possible): Likewise.
31217 (chkp_reduce_bounds_lifetime): Likewise.
31218 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31219 (chkp_mark_completed_bounds): Likewise.
31220 (chkp_register_incomplete_bounds): Likewise.
31221 (chkp_mark_invalid_bounds): Likewise.
31222 (chkp_maybe_copy_and_register_bounds): Likewise.
31223 (chkp_build_returned_bound): Likewise.
31224 (chkp_get_bound_for_parm): Likewise.
31225 (chkp_build_bndldx): Likewise.
31226 (chkp_get_bounds_by_definition): Likewise.
31227 (chkp_generate_extern_var_bounds): Likewise.
31228 (chkp_get_bounds_for_decl_addr): Likewise.
31229 * tree-chrec.c (chrec_apply): Likewise.
31230 * tree-data-ref.c (dump_data_reference): Likewise.
31231 (dump_subscript): Likewise.
31232 (dump_data_dependence_relation): Likewise.
31233 (analyze_overlapping_iterations): Likewise.
31234 * tree-inline.c (expand_call_inline): Likewise.
31235 (tree_function_versioning): Likewise.
31236 * tree-into-ssa.c (dump_defs_stack): Likewise.
31237 (dump_currdefs): Likewise.
31238 (dump_names_replaced_by): Likewise.
31239 (dump_update_ssa): Likewise.
31240 (update_ssa): Likewise.
31241 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31242 * tree-parloops.c (build_new_reduction): Likewise.
31243 (try_create_reduction_list): Likewise.
31244 (ref_conflicts_with_region): Likewise.
31245 (oacc_entry_exit_ok_1): Likewise.
31246 (oacc_entry_exit_single_gang): Likewise.
31247 * tree-pretty-print.h: Likewise.
31248 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31249 (get_scalar_evolution): Likewise.
31250 (add_to_evolution): Likewise.
31251 (get_loop_exit_condition): Likewise.
31252 (analyze_evolution_in_loop): Likewise.
31253 (analyze_initial_condition): Likewise.
31254 (analyze_scalar_evolution): Likewise.
31255 (instantiate_scev): Likewise.
31256 (number_of_latch_executions): Likewise.
31257 (gather_chrec_stats): Likewise.
31258 (final_value_replacement_loop): Likewise.
31259 (scev_const_prop): Likewise.
31260 * tree-sra.c (dump_access): Likewise.
31261 (disqualify_candidate): Likewise.
31262 (create_access): Likewise.
31263 (reject): Likewise.
31264 (maybe_add_sra_candidate): Likewise.
31265 (create_access_replacement): Likewise.
31266 (analyze_access_subtree): Likewise.
31267 (analyze_all_variable_accesses): Likewise.
31268 (sra_modify_assign): Likewise.
31269 (initialize_constant_pool_replacements): Likewise.
31270 (find_param_candidates): Likewise.
31271 (decide_one_param_reduction): Likewise.
31272 (replace_removed_params_ssa_names): Likewise.
31273 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31274 * tree-ssa-copy.c (dump_copy_of): Likewise.
31275 (copy_prop_visit_cond_stmt): Likewise.
31276 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31277 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31278 (record_equivalences_from_stmt): Likewise.
31279 * tree-ssa-dse.c (compute_trims): Likewise.
31280 (delete_dead_call): Likewise.
31281 (delete_dead_assignment): Likewise.
31282 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31283 (forward_propagate_into_cond): Likewise.
31284 (pass_forwprop::execute): Likewise.
31285 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31286 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31288 (move_computations_worker): Likewise.
31289 (execute_sm): Likewise.
31290 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31291 (remove_exits_and_undefined_stmts): Likewise.
31292 (remove_redundant_iv_tests): Likewise.
31293 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31294 (adjust_iv_update_pos): Likewise.
31295 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31296 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31297 (value_replacement): Likewise.
31298 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31299 * tree-ssa-pre.c (print_pre_expr): Likewise.
31300 (get_representative_for): Likewise.
31301 (create_expression_by_pieces): Likewise.
31302 (insert_into_preds_of_block): Likewise.
31303 (eliminate_insert): Likewise.
31304 (eliminate_dom_walker::before_dom_children): Likewise.
31305 (eliminate): Likewise.
31306 (remove_dead_inserted_code): Likewise.
31307 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31308 * tree-ssa-reassoc.c (get_rank): Likewise.
31309 (eliminate_duplicate_pair): Likewise.
31310 (eliminate_plus_minus_pair): Likewise.
31311 (eliminate_not_pairs): Likewise.
31312 (undistribute_ops_list): Likewise.
31313 (eliminate_redundant_comparison): Likewise.
31314 (update_range_test): Likewise.
31315 (optimize_range_tests_var_bound): Likewise.
31316 (optimize_vec_cond_expr): Likewise.
31317 (rewrite_expr_tree): Likewise.
31318 (rewrite_expr_tree_parallel): Likewise.
31319 (linearize_expr): Likewise.
31320 (break_up_subtract): Likewise.
31321 (linearize_expr_tree): Likewise.
31322 (attempt_builtin_powi): Likewise.
31323 (attempt_builtin_copysign): Likewise.
31324 (transform_stmt_to_copy): Likewise.
31325 (transform_stmt_to_multiply): Likewise.
31326 (dump_ops_vector): Likewise.
31327 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31328 (print_scc): Likewise.
31329 (set_ssa_val_to): Likewise.
31330 (visit_reference_op_store): Likewise.
31331 (visit_use): Likewise.
31332 (sccvn_dom_walker::before_dom_children): Likewise.
31333 (run_scc_vn): Likewise.
31334 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31336 (expr_hash_elt::print): Likewise.
31337 (const_and_copies::pop_to_marker): Likewise.
31338 (const_and_copies::record_const_or_copy_raw): Likewise.
31339 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31340 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31341 (dump_predicates): Likewise.
31342 (find_uninit_use): Likewise.
31343 (warn_uninitialized_phi): Likewise.
31344 (pass_late_warn_uninitialized::execute): Likewise.
31345 * tree-ssa.c (verify_vssa): Likewise.
31346 (verify_ssa): Likewise.
31347 (maybe_optimize_var): Likewise.
31348 * tree-vrp.c (dump_value_range): Likewise.
31349 (dump_all_value_ranges): Likewise.
31350 (dump_asserts_for): Likewise.
31351 (register_edge_assert_for_2): Likewise.
31352 (vrp_visit_cond_stmt): Likewise.
31353 (vrp_visit_switch_stmt): Likewise.
31354 (vrp_visit_stmt): Likewise.
31355 (vrp_visit_phi_node): Likewise.
31356 (simplify_cond_using_ranges_1): Likewise.
31357 (fold_predicate_in): Likewise.
31358 (evrp_dom_walker::before_dom_children): Likewise.
31359 (evrp_dom_walker::push_value_range): Likewise.
31360 (evrp_dom_walker::pop_value_range): Likewise.
31361 (execute_early_vrp): Likewise.
31363 2017-05-16 Richard Biener <rguenther@suse.de>
31365 * dwarf2out.c (loc_list_from_tree_1): Do not create
31366 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31368 2017-05-16 Richard Biener <rguenther@suse.de>
31370 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31372 (note_variable_value_in_expr): If we resolved the decl ref
31373 do not push to the stack.
31375 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31377 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31378 operations in fast-math mode.
31379 (vaddq_f16): Likewise.
31380 (vmul_f16): Likewise.
31381 (vmulq_f16): Likewise.
31382 (vsub_f16): Likewise.
31383 (vsubq_f16): Likewise.
31384 * config/arm/neon.md (add<mode>3): New.
31386 (fma:<VH:mode>3): New. Also remove outdated comment.
31389 2017-05-16 Martin Liska <mliska@suse.cz>
31393 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31394 (odr_types_equivalent_p): Likewise.
31396 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31398 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31400 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31403 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31404 non-interunit SSE move alternatives with '?'.
31405 (zero-extendsidi peephole2): New peephole to skip intermediate
31406 general register in SSE zero-extend sequence.
31408 2017-05-15 Jeff Law <law@redhat.com>
31410 * reorg.c (relax_delay_slots): Create a new variable to hold
31411 the temporary target rather than clobbering TARGET_LABEL.
31413 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31414 missing argument to extract_bit_field call.
31415 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31417 2017-05-15 Martin Liska <mliska@suse.cz>
31420 * gcc.c (process_command): Do not allow empty argument of -o option.
31422 2017-05-15 Renlin Li <renlin.li@arm.com>
31424 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31425 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31426 * config/aarch64/constraints.md (Usf): Add long call check.
31427 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31428 (call_value): Likewise.
31429 (sibcall): Likewise.
31430 (sibcall_value): Likewise.
31432 (call_value_insn): New.
31433 (sibcall_insn): Update rtx pattern.
31434 (sibcall_value_insn): Likewise.
31435 (call_internal): Remove.
31436 (call_value_internal): Likewise.
31437 (sibcall_internal): Likewise.
31438 (sibcall_value_internal): Likewise.
31439 (call_reg): Likewise.
31440 (call_symbol): Likewise.
31441 (call_value_reg): Likewise.
31442 (call_value_symbol): Likewise.
31444 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31447 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31449 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31451 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31452 compatible with CCGOCmode and with CCZmode.
31454 2017-05-14 Martin Sebor <msebor@redhat.com>
31456 PR middle-end/77671
31457 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31458 (gimple_fold_builtin_snprintf): Same.
31459 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31460 (gimple_fold_builtin_snprintf): Same.
31461 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31462 of character types.
31463 (is_call_safe): New function.
31464 (try_substitute_return_value): Call it.
31465 (try_simplify_call): New function.
31466 (pass_sprintf_length::handle_gimple_call): Call it.
31468 2017-05-14 Martin Sebor <msebor@redhat.com>
31470 PR middle-end/80669
31471 * builtins.c (expand_builtin_stpncpy): Simplify.
31473 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31475 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31476 * config/i386/i386.h
31477 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31478 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31479 (struct machine_function): Add new members call_ms2sysv,
31480 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31481 (struct machine_frame_state): New fields sp_realigned and
31482 sp_realigned_offset.
31483 * config/i386/i386.c
31484 (enum xlogue_stub): New enum.
31485 (enum xlogue_stub_sets): New enum.
31486 (class xlogue_layout): New class.
31487 (struct ix86_frame): New fields stack_realign_allocate_offset,
31488 stack_realign_offset and outlined_save_offset. Modify comments to
31489 detail stack layout when using out-of-line stubs.
31490 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31491 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31492 -mcall-ms2sysv-xlogues.
31493 (stub_managed_regs): New static variable.
31494 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31495 registers managed by out-of-line stub.
31496 (disable_call_ms2sysv_xlogues): New function.
31497 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31498 m->call_ms2sysv when appropriate and compute frame layout for
31500 (sp_valid_at, fp_valid_at): New inline functions.
31501 (choose_basereg): New function.
31502 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31504 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31505 Use align parameter of choose_baseaddr to generated aligned SSE movs
31507 (pro_epilogue_adjust_stack): Modify to track
31508 machine_frame_state::sp_realigned.
31509 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31510 (ix86_nsaved_sseregs): Likewise.
31511 (ix86_emit_save_regs): Likewise.
31512 (ix86_emit_save_regs_using_mov): Likewise.
31513 (ix86_emit_save_sse_regs_using_mov): Likewise.
31514 (get_scratch_register_on_entry): Likewise.
31515 (gen_frame_set): New function.
31516 (gen_frame_load): Likewise.
31517 (gen_frame_store): Likewise.
31518 (emit_outlined_ms2sysv_save): Likewise.
31519 (emit_outlined_ms2sysv_restore): Likewise.
31520 (ix86_expand_prologue): Modify stack re-alignment code and call
31521 emit_outlined_ms2sysv_save when appropriate.
31522 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31523 parameter rtx_insn *insn, which allows the function to be used to only
31524 generate the notes.
31525 (ix86_expand_epilogue): Modify validity checks of frame and stack
31526 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31527 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31528 * config/i386/predicates.md
31529 (save_multiple): New predicate.
31530 (restore_multiple): Likewise.
31531 * config/i386/sse.md
31532 (save_multiple<mode>): New pattern.
31533 (save_multiple_realign<mode>): Likewise.
31534 (restore_multiple<mode>): Likewise.
31535 (restore_multiple_and_return<mode>): Likewise.
31536 (restore_multiple_leave_return<mode>): Likewise.
31537 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31539 2017-05-14 Julia Koval <julia.koval@intel.com>
31541 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31542 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31543 (__builtin_ia32_xsetbv): New builtins.
31544 * config/i386/i386.c (ix86_expand_special_args_builtin):
31546 (ix86_expand_builtin): Special expand for new intrinsics.
31547 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31548 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31549 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31551 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31553 * cfganal.c (inverted_post_order_compute): Change argument type
31555 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31556 * df-core.c (rest_of_handle_df_initialize): Adjust.
31557 (rest_of_handle_df_finish): Likewise.
31558 (df_analyze_1): Likewise.
31559 (df_analyze): Likewise.
31560 (loop_inverted_post_order_compute): Change argument to be a vec *.
31561 (df_analyze_loop): Adjust.
31562 (df_get_n_blocks): Likewise.
31563 (df_get_postorder): Likewise.
31564 * df.h (struct df_d): Change field to be a vec.
31565 * lcm.c (compute_laterin): Adjust.
31566 (compute_available): Likewise.
31567 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31568 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31569 * tree-ssa-pre.c (compute_antic): Likewise.
31571 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31573 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31574 (depth_first_search::depth_first_search): Change structure init
31575 function to this constructor.
31576 (depth_first_search::add_bb): Rename function to this member.
31577 (depth_first_search::execute): Likewise.
31578 (flow_dfs_compute_reverse_finish): Adjust.
31580 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31582 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31583 (longest_simple_path): Likewise.
31584 * shrink-wrap.c (spread_components): Likewise.
31585 (disqualify_problematic_components): Likewise.
31586 (emit_common_heads_for_components): Likewise.
31587 (emit_common_tails_for_components): Likewise.
31588 (insert_prologue_epilogue_for_components): Likewise.
31590 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31592 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31595 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31597 * df-core.c (df_set_blocks): Start using auto_bitmap.
31598 (df_compact_blocks): Likewise.
31599 * df-problems.c (df_rd_confluence_n): Likewise.
31600 * df-scan.c (df_insn_rescan_all): Likewise.
31601 (df_process_deferred_rescans): Likewise.
31602 (df_update_entry_block_defs): Likewise.
31603 (df_update_exit_block_uses): Likewise.
31604 (df_entry_block_bitmap_verify): Likewise.
31605 (df_exit_block_bitmap_verify): Likewise.
31606 (df_scan_verify): Likewise.
31607 * lra-constraints.c (lra_constraints): Likewise.
31608 (undo_optional_reloads): Likewise.
31609 (lra_undo_inheritance): Likewise.
31610 * lra-remat.c (calculate_gen_cands): Likewise.
31611 (do_remat): Likewise.
31612 * lra-spills.c (assign_spill_hard_regs): Likewise.
31613 (spill_pseudos): Likewise.
31614 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31615 (bitmap_set_subtract_values): Likewise.
31617 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31619 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31620 management with auto_bitmap.
31621 (fix_inter_tick): Likewise.
31622 (fix_recovery_deps): Likewise.
31623 * ira.c (add_store_equivs): Likewise.
31624 (find_moveable_pseudos): Likewise.
31625 (split_live_ranges_for_shrink_wrap): Likewise.
31626 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31627 (rtx_reuse_manager::seen_def_p): Likewise.
31628 (rtx_reuse_manager::set_seen_def): Likewise.
31629 * print-rtl.h (class rtx_reuse_manager): Likewise.
31631 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31633 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31635 (migrate_btr_def): Likewise.
31636 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31637 * df-core.c (loop_post_order_compute): Likewise.
31638 (loop_inverted_post_order_compute): Likewise.
31639 * hsa-common.h: Likewise.
31640 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31641 * init-regs.c (initialize_uninitialized_regs): Likewise.
31642 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31643 (inline_small_functions): Likewise.
31644 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31645 * ira.c (combine_and_move_insns): Likewise.
31646 (build_insn_chain): Likewise.
31647 * loop-invariant.c (find_invariants): Likewise.
31648 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31649 * predict.c (tree_predict_by_opcode): Likewise.
31650 (predict_paths_leading_to): Likewise.
31651 (predict_paths_leading_to_edge): Likewise.
31652 (estimate_loops_at_level): Likewise.
31653 (estimate_loops): Likewise.
31654 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31655 (spread_components): Likewise.
31656 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31657 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31658 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31659 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31660 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31661 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31662 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31663 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31664 (mark_threaded_blocks): Likewise.
31665 (thread_through_all_blocks): Likewise.
31666 * tree-ssa.c (verify_ssa): Likewise.
31667 (execute_update_addresses_taken): Likewise.
31668 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31670 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31672 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31674 (post_order_compute): Likewise.
31675 (inverted_post_order_compute): Likewise.
31676 (pre_and_rev_post_order_compute_fn): Likewise.
31678 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31680 * genrecog.c (int_set::int_set): Explicitly construct our
31681 auto_vec base class.
31682 * vec.h (auto_vec::auto_vec): New constructor.
31684 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31686 * bitmap.h (class auto_bitmap): New constructor taking
31687 bitmap_obstack * argument.
31689 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31691 * bitmap.h (class auto_bitmap): Change type of m_bits to
31692 bitmap_head, and adjust ctor / dtor and member operators.
31694 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31696 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31697 when returned register mode doesn't match original mode.
31699 2017-05-12 Jeff Law <law@redhat.com>
31700 Jakub Jelinek <jakub@redhat.com>
31702 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31703 we look for cc setter after the compare-elim changes.
31704 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31705 within the vector to match what compare-elim now expects.
31706 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31707 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31709 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31710 after the compare-elim changes.
31711 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31712 the vector to match what compare-elim now expects.
31713 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31714 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31715 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31716 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31717 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31719 * config/visium/visium.c (single_set_and_flags): Fix where
31720 we look for cc setter after the compare-elim changes.
31721 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31722 with the vector to match what compare-elim now expects.
31723 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31724 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31725 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31726 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31727 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31728 (neg<mode>2_insn_set_overflow): Likewise.
31730 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31732 PR middle-end/79794
31733 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31734 maybe_expand_insn call, set ops[0].target. If still set after call,
31735 set alt_rtl. Add extra arg to recursive calls.
31736 (extract_bit_field): Add alt_rtl argument. Pass to
31738 * expmed.h (extract_bit_field): Fix prototype.
31739 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31740 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31741 to extract_bit_field_calls.
31742 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31743 Pass alt_rtl to extract_bit_field calls.
31744 * calls.c (store_unaligned_arguments_into_psuedos)
31745 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31746 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31748 * optabs.h (struct expand_operand): Add target bitfield.
31750 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31752 * compare-elim.c (try_eliminate_compare): Canonicalize
31753 operation with embedded compare to
31754 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31755 (set (reg) (operation)].
31757 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31759 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31762 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31763 cost of adding a carry flag for ADC instruction.
31764 [case MINUS]: Ignore the cost of subtracting a carry flag
31765 for SBB instruction.
31767 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31769 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31771 * config/rs6000/bmiintrin.h: New file.
31772 * config/rs6000/bmi2intrin.h: New file.
31773 * config/rs6000/x86intrin.h: New file.
31775 2017-05-12 Jeff Law <law@redhat.com>
31777 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31780 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31782 PR middle-end/80707
31783 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31785 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31787 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31788 early expansion of vector divide builtins.
31789 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31790 builtins identified as having unsigned arguments.
31792 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31794 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31795 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31796 expansion of vector logical operations (and, andc, or, xor,
31799 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31801 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31802 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31804 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31806 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31807 early GIMPLE expansion of vector multiplies.
31809 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31811 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31812 TARGET_HAVE_MOVT conditional.
31813 (movt splitter): Likewise.
31815 2017-05-12 Richard Biener <rguenther@suse.de>
31817 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31818 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31819 Fold all stmts not inplace.
31821 2017-05-12 Richard Biener <rguenther@suse.de>
31823 PR tree-optimization/80713
31824 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31825 inserted_exprs bit for not removed stmts.
31827 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31829 PR middle-end/69921
31830 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31831 parallelized" attribute for parallelized OpenACC kernels.
31832 * omp-offload.c (execute_oacc_device_lower): Use it.
31834 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31835 Set "oacc kernels" attribute.
31836 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31837 parameter. Adjust all users.
31838 (oacc_fn_attrib_kernels_p): Remove function.
31839 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31840 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31841 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31842 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31843 assert "oacc kernels" attribute is set.
31845 2017-05-11 Carl Love <cel@us.ibm.com>
31847 * config/rs6000/rs6000-c: Add support for built-in functions
31848 vector unsigned char vec_popcnt (vector signed char)
31849 vector unsigned char vec_popcnt (vector unsigned char)
31850 vector unsigned short vec_popcnt (vector signed short)
31851 vector unsigned short vec_popcnt (vector unsigned short)
31852 vector unsigned int vec_popcnt (vector signed int)
31853 vector unsigned int vec_popcnt (vector unsigned int)
31854 vector unsigned long long vec_popcnt (vector signed long long)
31855 vector unsigned long long vec_popcnt (vector unsigned long long)
31856 vector signed long long vec_slo (vector signed long long,
31857 vector signed char)
31858 vector signed long long vec_slo (vector signed long long,
31859 vector unsigned char)
31860 vector unsigned long long vec_slo (vector unsigned long long,
31861 vector signed char)
31862 vector unsigned long long vec_slo (vector unsigned long long,
31863 vector unsigned char)
31864 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31865 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31866 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31867 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31868 * doc/extend.texi: Update the built-in documentation file for the
31869 new built-in functions.
31871 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31873 * attribs.h (sorted_attr_string): Move machine independent
31874 functions for target clone support from the i386 port to common
31875 code. Rename ix86_function_versions to common_function_versions.
31876 Rename make_name to make_unique_name.
31877 (common_function_versions): Likewise.
31878 (make_unique_name): Likewise.
31879 (make_dispatcher_decl): Likewise.
31880 (is_function_default_version): Likewise.
31881 * attribs.c (attr_strcmp): Likewise.
31882 (sorted_attr_string): Likewise.
31883 (common_function_versions): Likewise.
31884 (make_unique_name): Likewise.
31885 (make_dispatcher_decl): Likewise.
31886 (is_function_default_version): Likewise.
31887 * config/i386/i386.c (attr_strcmp): Likewise.
31888 (sorted_attr_string): Likewise.
31889 (ix86_function_versions): Likewise.
31890 (make_name): Likewise.
31891 (make_dispatcher_decl): Likewise.
31892 (is_function_default_version): Likewise.
31893 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31895 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31898 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31899 Account for direct move costs for vec_construct of integer
31902 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31905 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31906 (UNSPEC_STX_ATOMIC): Ditto.
31907 (loaddi_via_sse): New insn.
31908 (storedi_via_sse): Ditto.
31909 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31910 Update corresponding peephole2 patterns.
31911 (atomic_storedi_fpu): Ditto.
31913 2017-05-11 Julia Koval <julia.koval@intel.com>
31915 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31916 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31918 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31919 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31920 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31922 2017-05-11 Nathan Sidwell <nathan@acm.org>
31924 * graphite-poly.c: Include dumpfile.h.
31926 * dumpfle.h (dump_function): Declare here ...
31927 * tree-dump.h (dump_function): ... not here.
31928 * dumpfile.c: #include tree-cfg.h.
31929 (dump_function): Move here from ...
31930 * tree-dump.c (dump_function): ... here.
31931 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31932 * graphite-poly.c: Don't include tree-dump.h.
31933 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31934 * print-tree.c: Likewise.
31935 * stor-layout.c: Likewise.
31936 * tree-nested.c: Likewise.
31938 * dumpfile.c (dump_start): Use TDF_FLAGS.
31939 (dump_enable_all): Fix TDF_KIND check thinko.
31941 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31943 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31944 array entries to represent two legal parameterizations of the
31945 overloaded __builtin_cmpb function, as represented by the
31946 P6_OV_BUILTIN_CMPB constant.
31947 (altivec_resolve_overloaded_builtin): Add special case handling
31948 for the __builtin_cmpb function, as represented by the
31949 P6_OV_BUILTIN_CMPB constant.
31950 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31951 (BU_P6_64BIT_2): New macro.
31952 (BU_P6_OVERLOAD_2): New macro
31953 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31954 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31955 (CMPB): Add overload support to represent both 32-bit and 64-bit
31956 compare-bytes function.
31957 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31958 support for TARGET_CMPB.
31959 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31960 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31961 documentation of the __builtin_cmpb overloaded built-in function.
31963 2017-05-11 Richard Biener <rguenther@suse.de>
31965 PR tree-optimization/80705
31966 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31967 bases are not vectorizable.
31969 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31971 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31972 when counting register pressure.
31974 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31976 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31977 (iv_ca_more_deps): Renamed to ...
31978 (iv_ca_compare_deps): ... this.
31979 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31981 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31983 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31985 (determine_group_iv_costs): ... here.
31986 (find_inv_vars_cb): Record inv var if it's not recorded before.
31988 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31990 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31991 (get_shiftadd_cost): Ditto.
31993 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31995 * tree-ssa-address.c: Include header file.
31996 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31998 (add_to_parts): Refactor.
31999 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32000 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32003 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32005 PR tree-optimization/53090
32006 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32008 (extract_cond_operands): Detect condition with IV on both sides
32009 and return COMP_IV_EXPR_2.
32010 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32011 (rewrite_use_compare): Simplify by removing call to function
32012 extract_cond_operands.
32014 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32016 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32017 (extract_cond_operands): Detect condition comparing against non-
32018 invariant bound and return appropriate enum value.
32019 (find_interesting_uses_cond): Update use of extract_cond_operands.
32020 Handle its return value accordingly.
32021 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32023 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32025 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32026 nonlinear iv_use computation in loop invariant sensitive way.
32028 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32030 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32031 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32033 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32035 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32036 (dump_cand): Support iv_cand.inv_exprs.
32037 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32039 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32042 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32044 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32046 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32047 as local function. Include necessary header files.
32048 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32050 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32052 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32054 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32056 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32057 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32058 RSHIFT_EXPR and BIT_NOT_EXPR.
32060 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32062 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32063 (adjust_setup_cost): New parameter supporting round up adjustment.
32064 (struct address_cost_data): Delete.
32065 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32066 (split_address_cost, ptr_difference_cost): Delete.
32067 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32068 (struct ainc_cost_data): New struct.
32069 (get_address_cost_ainc): New function.
32070 (get_address_cost, get_computation_cost): Reimplement.
32071 (determine_group_iv_cost_address): Record inv_expr for all uses of
32073 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32074 (iv_ca_has_deps): Reimplemented to ...
32075 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32077 (iv_ca_extend): Call iv_ca_more_deps.
32079 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32081 * tree-ssa-address.c (struct mem_address): Move to header file.
32082 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32083 * tree-ssa-address.h (struct mem_address): Move from C file.
32084 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32086 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32088 * tree-affine.h (aff_combination_type): New interface.
32089 (aff_combination_zero_p): Remove static.
32090 (aff_combination_const_p): New interface.
32091 (aff_combination_singleton_var_p): New interfaces.
32093 2017-05-11 Richard Biener <rguenther@suse.de>
32095 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32096 Skip unreachable blocks and destinations.
32097 (eliminate): Move stmt removal and fixup ...
32098 (fini_eliminate): ... here. Skip inserted exprs.
32099 (pass_pre::execute): Move fini_pre after fini_eliminate.
32100 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32101 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32102 PRE to get rid of dead code that has invalid SSA form and
32103 split critical edges again.
32105 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32107 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32109 2017-05-11 Richard Biener <rguenther@suse.de>
32111 * passes.c (execute_function_todo): Verify loops if they are
32112 said to be up-to-date.
32113 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32114 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32116 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32119 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32120 handle calling assemble_external ourself.
32123 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32124 modes with zero size. Enhance comment.
32126 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32128 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32129 built-ins for vec_xl and vec_xst with short and char pointer
32132 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32134 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32135 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32136 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32137 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32138 (_mm_maskz_min_round_ss): New intrinsics.
32139 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32140 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32141 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32142 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32143 (__builtin_ia32_minss_mask_round): New builtins.
32144 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32145 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32146 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32148 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32149 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32151 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32154 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32156 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32157 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32158 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32159 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32160 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32161 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32162 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32163 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32164 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32165 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32166 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32167 (__builtin_ia32_mulss_mask_round): New builtins.
32168 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32169 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32170 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32172 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32173 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32175 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32178 2017-05-10 Julia Koval <julia.koval@intel.com>
32180 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32181 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32182 (_mm256_setr_m128i): New intrinsics.
32184 2017-05-10 Julia Koval <julia.koval@intel.com>
32186 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32187 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32188 (_mm_maskz_rcp14_ss): New intrinsics.
32189 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32190 (__builtin_ia32_rcp14ss_mask): New builtins.
32191 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32193 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32195 PR tree-optimization/51513
32196 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32197 (assert_unreachable_fallthru_edge_p): Use it.
32198 (group_case_labels_stmt): Likewise.
32199 * tree-cfg.h: Prototype it.
32200 * stmt.c: Include cfghooks.h and tree-cfg.h.
32201 (emit_case_dispatch_table) <gap_label>: New local variable.
32202 Use it to fill dispatch table gaps.
32203 Test for default_label before updating probabilities.
32204 (expand_case) <default_label>: Remove unneeded initialization.
32205 Test for unreachable default case statement and remove its edge.
32206 Set default_label accordingly.
32207 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32209 2017-05-10 Carl Love <cel@us.ibm.com>
32211 * config/rs6000/rs6000-c: Add support for built-in functions
32212 vector signed char vec_neg (vector signed char)
32213 vector signed short int vec_neg (vector short int)
32214 vector signed int vec_neg (vector signed int)
32215 vector signed long long vec_neg (vector signed long long)
32216 vector float vec_neg (vector float)
32217 vector double vec_neg (vector double)
32218 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32220 * config/rs6000/altivec.h: Add define for vec_neg
32221 * doc/extend.texi: Update the built-in documentation for the
32222 new built-in functions.
32224 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32226 PR tree-optimization/77644
32227 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32229 2017-05-10 Nathan Sidwell <nathan@acm.org>
32231 * dumpfile.h (TDI_lang_all): New.
32232 (TDF_KIND): New. Renumber others
32233 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32235 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32237 (get_dump_file_name): Adjust suffix generation.
32238 (dump_enable_all): Use TDF_KIND.
32239 * doc/invoke.texi (-fdump-lang-all): Document.
32241 * dumpfile.h: Tabify.
32243 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32246 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32247 Move member access before delete.
32249 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32251 * tree-inline.c (expand_call_inline): Split block at stmt
32254 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32257 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32258 are now unused after splitting mov{sf,sd}_hardfloat.
32259 (f32_lr2): Likewise.
32260 (f32_lm): Likewise.
32261 (f32_lm2): Likewise.
32262 (f32_li): Likewise.
32263 (f32_li2): Likewise.
32264 (f32_lv): Likewise.
32265 (f32_sr): Likewise.
32266 (f32_sr2): Likewise.
32267 (f32_sm): Likewise.
32268 (f32_sm2): Likewise.
32269 (f32_si): Likewise.
32270 (f32_si2): Likewise.
32271 (f32_sv): Likewise.
32272 (f32_dm): Likewise.
32273 (f32_vsx): Likewise.
32274 (f32_av): Likewise.
32275 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32276 For movsf, order stores so the VSX stores occur before the GPR
32277 store which encourages the register allocator to use a traditional
32278 FPR instead of a GPR. For movsd, order the stores so that the GPR
32279 store comes before the VSX stores to allow the power6 to work.
32280 This is due to the power6 not having a 32-bit integer store
32281 instruction from a FPR.
32282 (movsf_hardfloat): Likewise.
32283 (movsd_hardfloat): Likewise.
32285 2017-05-09 Martin Sebor <msebor@redhat.com>
32287 PR translation/80280
32288 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32291 PR translation/80280
32292 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32293 data member added in r247778.
32294 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32296 2017-05-09 Nathan Sidwell <nathan@acm.org>
32298 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32300 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32303 2017-05-09 Marek Polacek <polacek@redhat.com>
32305 * doc/invoke.texi: Fix typo.
32307 2017-05-09 Richard Biener <rguenther@suse.de>
32309 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32310 (vrp_val_is_min): Likewise.
32311 (set_value_range_to_value): Likewise.
32312 (set_value_range_to_nonnegative): Likewise.
32313 (gimple_assign_nonzero_p): Likewise.
32314 (gimple_stmt_nonzero_p): Likewise.
32315 (vrp_int_const_binop): Likewise. Remove unreachable case.
32316 (adjust_range_with_scev): Adjust comments.
32317 (compare_range_with_value): Likewise.
32318 (extract_range_from_phi_node): Likewise.
32319 (test_for_singularity): Likewise.
32321 2017-05-09 Richard Biener <rguenther@suse.de>
32323 * tree-vrp.c (get_single_symbol): Add assert that we don't
32324 get overflowed constants as invariant part.
32325 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32326 checks. Use wi::cmp instead of recursing for integer constants.
32327 (compare_values): Just ignore whether we assumed undefined
32328 overflow instead of failing the compare.
32329 (extract_range_for_var_from_comparison_expr): Add comment before the
32330 TREE_NO_WARNING sets.
32331 (test_for_singularity): Likewise.
32332 (extract_range_from_comparison): Do not disable optimization
32333 when we assumed undefined overflow.
32334 (extract_range_basic): Remove init of unused var.
32336 2017-05-09 Richard Biener <rguenther@suse.de>
32338 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32339 (extract_range_from_multiplicative_op_1): Adjust.
32340 (extract_range_from_binary_expr_1): Use int_const_binop.
32342 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32345 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32346 rs6000_store_data_bypass_p in seven define_bypass directives and
32347 in several comments.
32348 * config/rs6000/rs6000-protos.h: Add prototype for
32349 rs6000_store_data_bypass_p function.
32350 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32351 function implements slightly different (rs6000-specific) semantics
32352 than store_data_bypass_p, returning false rather than aborting
32353 with assertion error when arguments do not satisfy the
32354 requirements of store data bypass.
32355 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32356 rs6000_store_data_bypass_p.
32358 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32360 * config/xtensa/xtensa-protos.h
32361 (xtensa_initial_elimination_offset): New declaration.
32362 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32363 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32364 macro definition, add case for FRAME_POINTER_REGNUM when
32365 FRAME_GROWS_DOWNWARD.
32366 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32367 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32368 xtensa_initial_elimination_offset.
32370 2017-05-08 Nathan Sidwell <nathan@acm.org>
32372 * doc/invoke.texi: Alphabetize -fdump options.
32374 2017-05-08 Martin Sebor <msebor@redhat.com>
32376 PR translation/80280
32377 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32379 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32381 * target.def (compute_frame_layout): New optional target hook.
32382 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32383 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32384 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32386 * reload1.c (verify_initial_elim_offsets): Likewise.
32387 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32388 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32389 (arm_get_frame_offsets): Split up into this ...
32390 (arm_compute_frame_layout): ... and this function.
32392 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32394 * config/aarch64/constraints.md (Usa): New constraint.
32395 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32397 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32399 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32400 with_multilib_list after it has been checked.
32402 2017-05-08 Richard Biener <rguenther@suse.de>
32404 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32405 (bitmap_set_subtract_values): Likewise.
32407 2017-05-08 Richard Biener <rguenther@suse.de>
32409 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32410 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32412 (gimple_stmt_nonzero_warnv_p): Rename to ...
32413 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32415 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32416 (extract_range_basic): Adjust, do not disable propagation on
32417 strict overflow sensitive simplification.
32418 (vrp_visit_cond_stmt): Likewise.
32420 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32422 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32423 body size unconditionally.
32425 2017-05-07 Jeff Law <law@redhat.com>
32428 2017-05-06 Jeff Law <law@redhat.com>
32429 PR tree-optimization/78496
32430 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32433 PR tree-optimization/78496
32434 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32435 (simplify_stmt_using_ranges): Call it.
32436 (vrp_dom_walker::before_dom_children): Extract equivalences
32437 from an ASSERT_EXPR with an equality comparison against a
32440 2017-05-06 Jeff Law <law@redhat.com>
32442 PR tree-optimization/78496
32443 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32446 PR tree-optimization/78496
32447 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32448 (simplify_stmt_using_ranges): Call it.
32449 (vrp_dom_walker::before_dom_children): Extract equivalences
32450 from an ASSERT_EXPR with an equality comparison against a
32453 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32455 * lra-constraints.c (lra_copy_reg_equiv): New function.
32456 (split_reg): Use it to copy equivalence information from the
32457 original register to the spill register.
32459 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32461 PR rtl-optimization/75964
32462 * simplify-rtx.c (simplify_const_relational_operation): Remove
32463 invalid handling of comparisons of integer ABS.
32465 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32467 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32468 initialize to zero.
32469 (init_regs): Remove declaration.
32470 (function_arg_advance_32): Initialize error_p as boolean variable.
32472 2017-05-05 Nathan Sidwell <nathan@acm.org>
32474 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32475 lines. Use for (;;).
32477 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32479 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32480 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32481 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32482 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32483 VF=2 that require versioning.
32485 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32487 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32490 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32492 * diagnostic.h (diagnostic_override_option_index): Convert from
32493 macro to inline function.
32495 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32497 * diagnostic.c (last_module_changed_p): New function.
32498 (set_last_module): New function.
32499 (diagnostic_report_current_module): Convert macro usage to
32500 the above functions.
32501 * diagnostic.h (diagnostic_context::last_module): Strengthen
32502 from const line_map * to const line_map_ordinary *.
32503 (diagnostic_last_module_changed): Delete macro.
32504 (diagnostic_set_last_module): Delete macro.
32506 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32508 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32509 with diagnostic_report_diagnostic.
32510 (diagnostic_n_impl_richloc): Likewise.
32511 * diagnostic.h (report_diagnostic): Delete macro.
32512 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32513 with diagnostic_report_diagnostic.
32514 * substring-locations.c (format_warning_va): Likewise.
32516 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32518 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32519 save/restor of format_spec. Move option-printing code to...
32520 (print_option_information): ...this new function, and
32521 reimplement by simply printing to the pretty_printer,
32522 rather than appending to the format string.
32524 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32526 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32527 handling logic into...
32528 (update_effective_level_from_pragmas): ...this new function.
32530 2017-05-04 Andrew Waterman <andrew@sifive.com>
32532 * config/riscv/riscv.opt (mstrict-align): New option.
32533 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32534 (SLOW_UNALIGNED_ACCESS): Define.
32535 (riscv_slow_unaligned_access): Declare.
32536 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32538 (riscv_slow_unaligned_access): New variable.
32539 (rocket_tune_info): Set slow_unaligned_access to true.
32540 (optimize_size_tune_info): Set slow_unaligned_access to false.
32541 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32542 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32543 (riscv_option_override): Set riscv_slow_unaligned_access.
32544 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32546 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32548 * config/riscv/riscv.md: Unify indentation.
32550 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32555 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32557 (extendsi<mode>2): Add support for doing sign extension via
32558 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32559 don't have ISA 3.0 instructions.
32560 (extendsi<mode>2 splitter): Likewise.
32561 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32562 generate the normal insns since SImode can now go in vector
32563 registers. Disallow the special UNSPECs needed for previous
32564 machines to hide SImode being used. Add new insns
32565 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32566 (fix_trunc<mode>si2_stfiwx): Likewise.
32567 (fix_trunc<mode>si2_internal): Likewise.
32568 (fixuns_trunc<mode>si2): Likewise.
32569 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32570 (fctiw<u>z_<mode>_smallint): Likewise.
32571 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32572 of floating point to 32-bit integer from doing a direct move to
32573 the GPR registers to do a store.
32574 (fctiwz_<mode>): Break long line.
32576 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32578 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32579 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32580 (addr_list, addr_offset_valid_p): New.
32581 (split_address_groups): Check offset validity with above function.
32582 (gt-tree-ssa-loop-ivopts.h): Include header file.
32584 2017-05-05 Nathan Sidwell <nathan@acm.org>
32586 * config.gcc (arm*-*-*): Add missing 'fi'.
32588 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32590 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32591 included in -fopt-info does not matter.
32592 * doc/optinfo.texi (-fopt-info): Fix description of default
32593 behavour. Explicitly say order of options included in -fopt-info
32596 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32598 * config.gcc: Allow combinations of aprofile and rmprofile values for
32599 --with-multilib-list.
32600 * config/arm/t-multilib: New file.
32601 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32602 variables. Remove setting of ISA and floating-point ABI in
32603 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32604 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32605 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32606 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32608 * config/arm/t-rmprofile: Likewise except for the matches changes.
32609 * doc/install.texi (--with-multilib-list): Document the combination of
32610 aprofile and rmprofile values and warn about pitfalls in doing that.
32612 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32614 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32615 (movdi_aarch64): Likewise.
32617 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32619 PR tree-optimization/80632
32620 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32622 (build_arrays): Initialize it for virtual phis.
32623 (fix_phi_nodes): Use it for virtual phis.
32625 PR tree-optimization/80558
32626 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32627 [x, y] op z into [x op, y op z] for op & or | if conditions
32630 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32631 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32634 * config/arm/arm.md (use_literal_pool): Remove.
32635 (64-bit immediate split): No longer takes cost into consideration
32636 if arm_disable_literal_pool is enabled.
32637 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32638 used when arm_disable_literal_pool is enabled.
32639 (arm_max_const_double_inline_cost): Remove use of
32640 arm_disable_literal_pool.
32641 (push_minipool_fix): Add assert.
32642 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32643 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32644 (no_literal_pool_sf_immediate): New.
32646 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32648 PR tree-optimization/80613
32649 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32650 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32652 2017-05-05 Richard Biener <rguenther@suse.de>
32654 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32656 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32658 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32659 of this flag from insn conditions due to removal from r247495.
32661 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32663 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32665 (arm_early_store_addr_dep_ptr): Likewise.
32666 * config/arm/aarch-common-protos.h
32667 (arm_early_load_addr_dep_ptr): Add prototype.
32668 (arm_early_store_addr_dep_ptr): Likewise.
32669 * config/arm/cortex-a53.md: Add new bypasses.
32671 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32673 * tree.c (next_type_uid): Change type to unsigned.
32674 (type_hash_canon): Decrement back next_type_uid if
32675 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32676 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32679 2017-05-04 Martin Sebor <msebor@redhat.com>
32681 * builtins.c: Fix a trivial typo in a comment.
32683 PR middle-end/79234
32684 * builtins.c (check_sizes): Adjust to handle reading past the end.
32685 Avoid printing excessive upper bound of ranges. Use %E to print
32686 tree nodes instead of converting them to %wu.
32687 (expand_builtin_memchr): New function.
32688 (compute_dest_size): Rename...
32689 (compute_objsize): ...to this.
32690 (expand_builtin_memcpy): Adjust.
32691 (expand_builtin_mempcpy): Adjust.
32692 (expand_builtin_strcat): Adjust.
32693 (expand_builtin_strcpy): Adjust.
32694 (check_strncat_sizes): Adjust.
32695 (expand_builtin_strncat): Adjust.
32696 (expand_builtin_strncpy): Adjust and simplify.
32697 (expand_builtin_memset): Adjust.
32698 (expand_builtin_bzero): Adjust.
32699 (expand_builtin_memcmp): Adjust.
32700 (expand_builtin): Handle memcmp.
32701 (maybe_emit_chk_warning): Check strncat just once.
32703 2017-05-04 Martin Sebor <msebor@redhat.com>
32705 PR preprocessor/79214
32706 PR middle-end/79222
32707 PR middle-end/79223
32708 * builtins.c (check_sizes): Add inlining context and issue
32709 warnings even when -Wno-system-headers is set.
32710 (check_strncat_sizes): Same.
32711 (expand_builtin_strncat): Same.
32712 (expand_builtin_memmove): New function.
32713 (expand_builtin_stpncpy): Same.
32714 (expand_builtin): Handle memmove and stpncpy.
32716 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32718 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32719 which is not used any more.
32721 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32723 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32725 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32727 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32728 (cortexa53_tunings): Likewise.
32729 (cortexa57_tunings): Likewise.
32730 (cortexa72_tunings): Likewise.
32731 (cortexa73_tunings): Likewise.
32733 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32735 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32736 Set loop alignment to 8.
32738 2017-05-04 Martin Sebor <msebor@redhat.com>
32740 PR translation/80280
32741 * builtins.c (expand_builtin_object_size): Add missing quoting to
32742 %D and like directives.
32743 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32744 (hsa_type_for_tree_type): Same.
32745 (verify_function_arguments): Same.
32746 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32747 * varasm.c (get_section): Same.
32750 2017-05-04 Martin Sebor <msebor@redhat.com>
32752 PR translation/80280
32753 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32755 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32757 * config/aarch64/aarch64.c (generic_addrcost_table):
32758 Change HI/TI mode setting.
32760 2017-05-04 Martin Jambor <mjambor@suse.cz>
32762 PR tree-optimization/80622
32763 * tree-sra.c (comes_initialized_p): New function.
32764 (build_accesses_from_assign): Only set write lazily when
32765 comes_initialized_p is false.
32766 (analyze_access_subtree): Use comes_initialized_p.
32767 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32768 instead of testing for PARM_DECL.
32770 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32772 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32773 constraint on operand 0 to allow more general addressing modes.
32774 Adjust output template.
32775 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32777 * config/aarch64/aarch64-protos.h
32778 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32779 * config/aarch64/constraints.md (Dp): New address constraint.
32780 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32783 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32785 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32786 update use of estimate_ipcp_clone_size_and_time.
32787 (estimate_local_effects): Update use of
32788 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32789 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32790 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32791 Return nonspecialized time.
32793 2017-05-04 Richard Biener <rguenther@suse.de>
32795 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32796 for the last VUSE which def dominates the PHI. Directly call
32798 (get_continuation_for_phi_1): Remove.
32800 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32802 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32803 to explain the use of truncating division. Cap the number of
32804 iterations to the maximum given by nb_iterations_upper_bound,
32807 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32809 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32810 * configure: Regenerate.
32811 * config.in: Regenerate.
32812 * config/i386/driver-mingw32.c: new file.
32813 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32814 * config.host: Link driver-mingw32.o on MinGW host.
32815 * doc/install.texi: Document new --enable-mingw-wildcard configure
32818 2017-05-04 Marek Polacek <polacek@redhat.com>
32820 PR tree-optimization/80612
32821 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32823 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32824 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32826 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32827 (movt splitter): Likewise.
32828 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32829 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32830 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32831 block for Thumb-1 with MOVT.
32832 (thumb2_legitimate_address_p): Move code block ...
32833 (can_avoid_literal_pool_for_label_p): ... into this new function.
32834 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32836 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32837 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32838 "M-profile targets with the MOVT instruction".
32840 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32842 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32843 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32844 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32846 2017-05-04 Martin Liska <mliska@suse.cz>
32848 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32849 variable cond_code.
32851 2017-05-04 Richard Biener <rguenther@suse.de>
32853 * tree.c (array_at_struct_end_p): Handle arrays at struct
32854 end with flexarrays more conservatively. Refactor and treat
32855 arrays of arrays or aggregates more strict. Fix
32856 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32857 * tree.c (array_at_struct_end_p): Adjust prototype.
32858 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32859 * gimple-fold.c (get_range_strlen): Likewise.
32860 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32862 2017-05-04 Richard Biener <rguenther@suse.de>
32864 PR tree-optimization/31130
32865 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32867 (supports_overflow_infinity): Likewise.
32868 (is_negative_overflow_infinity): Likewise.
32869 (is_positive_overflow_infinity): Likewise.
32870 (is_overflow_infinity): Likewise.
32871 (stmt_overflow_infinity): Likewise.
32872 (overflow_infinity_range_p): Likewise.
32873 (usable_range_p): Remove as always returning true.
32874 (make_overflow_infinity): Remove.
32875 (negative_overflow_infinity): Likewise.
32876 (positive_overflow_infinity): Likewise.
32877 (avoid_overflow_infinity): Likewise.
32878 (set_value_range): Adjust accordingly.
32879 (set_value_range_to_nonnegative): Likewise, remove now unused
32880 overflow_infinity arg.
32881 (vrp_operand_equal_p): Adjust.
32882 (update_value_range): Likewise.
32883 (range_int_cst_singleton_p): Likewise.
32884 (operand_less_p): Likewise.
32885 (compare_values_warnv): Likewise.
32886 (extract_range_for_var_from_comparison_expr): Likewise.
32887 (vrp_int_const_binop): Likewise.
32888 (zero_nonzero_bits_from_vr): Likewise.
32889 (extract_range_from_multiplicative_op_1): Likewise.
32890 (extract_range_from_binary_expr_1): Likewise.
32891 (extract_range_from_unary_expr): Likewise.
32892 (extract_range_from_comparison): Likewise.
32893 (extract_range_basic): Likewise.
32894 (adjust_range_with_scev): Likewise.
32895 (compare_ranges): Likewise.
32896 (compare_range_with_value): Likewise.
32897 (dump_value_range): Likewise.
32898 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32900 (simplify_cond_using_ranges): Adjust.
32902 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32904 * brig-builtins.def: Added a builtin for class_f64.
32905 * builtin-types.def: Added a builtin type needed by class_f64.
32907 2017-05-03 Jason Merrill <jason@redhat.com>
32909 * timevar.def: Add TV_CONSTEXPR.
32911 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32913 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32915 2017-05-03 Martin Jambor <mjambor@suse.cz>
32917 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32918 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32919 * ipa-cp.c (ipcp_propagate_stage): Do not call
32920 ipa_update_after_lto_read.
32921 * ipa-inline.c (ipa_inline): Likewise.
32923 2017-05-03 Martin Jambor <mjambor@suse.cz>
32925 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32926 tag. Added a default constructor and a destructor.
32927 (ipa_edge_args_sum_t): New class;
32928 (ipa_edge_args_sum): Declare.
32929 (ipa_edge_args_vector): Remove declaration.
32930 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32931 (ipa_free_edge_args_substructures): Remove declaration.
32932 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32933 (ipa_edge_args_info_available_for_edge_p): Likewise.
32934 * ipa-prop.c (ipa_edge_args_vector): Removed.
32935 (edge_removal_hook_holder): Likewise.
32936 (edge_duplication_hook_holder): Likewise.
32937 (ipa_edge_args_sum): New variable.
32938 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32939 ipa_edge_args_vector.
32940 (ipa_free_edge_args_substructures): Likewise.
32941 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32942 ipa_edge_args_vector.
32943 (ipa_edge_removal_hook): Turned into method
32944 ipa_edge_args_sum_t::remove.
32945 (ipa_edge_duplication_hook): Turned into method
32946 ipa_edge_args_sum_t::duplicate.
32947 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32948 registering edge hooks.
32949 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32950 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32951 ipa_edge_args_sum instead of ipa_edge_args_vector.
32952 * ipa-profile.c (ipa_profile): Likewise.
32954 2017-05-03 Martin Jambor <mjambor@suse.cz>
32956 * symbol-summary.h (function_summary): New method exists.
32957 (function_summary::symtab_removal): Deallocate through release.
32958 (call_summary): New class.
32959 (gt_ggc_mx): New overload.
32960 (gt_pch_nx): Likewise.
32961 (gt_pch_nx): Likewise.
32963 2017-05-03 Jeff Law <law@redhat.com>
32965 PR tree-optimization/78496
32966 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32967 from simplify_cond_using_ranges. Split off code to walk
32968 backwards through casts into ...
32969 (simplify_cond_using_ranges_2): New function.
32970 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32971 (execute_vrp): After identifying jump threads, call
32972 simplify_cond_using_ranges_2.
32974 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32977 * ipa-inline.h (inline_summary): Add ctor.
32978 (create_ggc): Do not use ggc_cleared_alloc.
32980 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32981 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32983 * gcc.c (handle_braces): Support escaping in switch matching
32985 * doc/invoke.texi (Spec Files): Document it.
32986 Remove superfluous @code markup in items.
32988 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32990 * diagnostic-show-locus.c (struct column_range): New struct.
32991 (get_affected_columns): New function.
32992 (get_printed_columns): New function.
32993 (struct correction): New struct.
32994 (correction::ensure_capacity): New function.
32995 (correction::ensure_terminated): New function.
32996 (struct line_corrections): New struct.
32997 (line_corrections::~line_corrections): New dtor.
32998 (line_corrections::add_hint): New function.
32999 (layout::print_trailing_fixits): Reimplement in terms of the new
33001 (selftest::test_overlapped_fixit_printing): New function.
33002 (selftest::diagnostic_show_locus_c_tests): Call it.
33004 2017-05-03 Nathan Sidwell <nathan@acm.org>
33006 Canonicalize canonical type hashing
33007 * tree.h (type_hash_canon_hash): Declare.
33008 * tree.c (type_hash_list, attribute_hash_list): Move into
33009 type_hash_canon_hash.
33010 (build_type_attribute_qual_variant): Break out hash code calc into
33011 type_hash_canon_hash.
33012 (type_hash_canon_hash): New. Generic type hash computation.
33013 (build_range_type_1, build_array_type_1, build_function_type,
33014 build_method_type_directly, build_offset_type, build_complex_type,
33015 make_vector_type): Call it.
33017 2017-05-03 Richard Biener <rguenther@suse.de>
33019 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33020 When all DRs have unknown misaligned do not always peel
33021 when there is a store but apply the same costing model as if
33022 there were only loads.
33024 2017-05-03 Richard Biener <rguenther@suse.de>
33027 PR tree-optimization/80492
33028 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33029 compare_base_decls returning dont-know properly.
33031 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33033 * config/arm/iterators.md (CCSI): New mode iterator.
33034 (arch): New mode attribute.
33035 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33036 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33037 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33038 code iterator for success result mode.
33039 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33040 the corresponding new insn generators.
33042 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33045 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33046 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33048 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33050 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33051 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33052 (DDR_A): Wrap DDR argument in brackets.
33053 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33054 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33055 (DDR_REVERSED_P): Likewise.
33057 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33059 PR tree-optimization/79472
33060 * tree-switch-conversion.c (struct switch_conv_info): Add
33061 contiguous_range and default_case_nonstandard fields.
33062 (collect_switch_conv_info): Compute contiguous_range and
33063 default_case_nonstandard fields, don't clear final_bb if
33064 contiguous_range and only the default case doesn't have the required
33066 (check_all_empty_except_final): Set default_case_nonstandard instead
33067 of failing if contiguous_range and the default case doesn't have empty
33069 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33070 and only the default case doesn't have the required constants. Skip
33072 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33073 if default_case_nonstandard.
33074 (build_constructors): Build constant 1 just once. Assert that default
33075 values aren't inserted in between cases if contiguous_range. Skip
33077 (build_arrays): Skip virtual phis.
33078 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33079 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33080 Handle virtual phis.
33081 (gen_inbound_check): Handle default_case_nonstandard case.
33082 (process_switch): Adjust check_final_bb caller. Call
33083 gather_default_values with the first non-default case instead of
33084 default case if default_case_nonstandard.
33086 2017-05-02 Nathan Sidwell <nathan@acm.org>
33088 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33089 check. Fix formatting.
33091 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33093 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33094 errors when comparing specialized and unspecialized times.
33096 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33098 * diagnostic-show-locus.c
33099 (layout::should_print_annotation_line_p): Make private.
33100 (layout::print_annotation_line): Make private.
33101 (layout::annotation_line_showed_range_p): Make private.
33102 (layout::show_ruler): Make private.
33103 (layout::print_source_line): Make private. Pass in line and
33104 line_width, rather than calling location_get_source_line. Drop
33106 (layout::print_leading_fixits): New method.
33107 (layout::print_any_fixits): Rename to...
33108 (layout::print_trailing_fixits): ...this, and make private.
33109 Don't print newline fixits.
33110 (diagnostic_show_locus): Move logic for printing one row into...
33111 (layout::print_line): ...this new function. Move the
33112 location_get_source_line call and error-handling from
33113 print_source_line to here. Call print_leading_fixits, and rename
33114 print_any_fixits to print_trailing_fixits.
33115 (selftest::test_fixit_insert_containing_newline): Update now that
33116 newlines are partially supported.
33117 (selftest::test_fixit_insert_containing_newline_2): New test.
33118 (selftest::test_fixit_replace_containing_newline): Update comments.
33119 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33120 * edit-context.c (class added_line): New class.
33121 (class edited_line): Describe newline handling in comment.
33122 (edited_line::actually_edited_p): New method.
33123 (edited_line::print_content): Delete redundant decl.
33124 (edited_line::m_predecessors): New field.
33125 (edited_file::print_content): Call edited_line::print_content.
33126 (edited_file::print_diff): Update to support newlines.
33127 (edited_file::print_diff_hunk): Likewise.
33128 (edited_file::print_run_of_changed_lines): New function.
33129 (edited_file::print_diff_line): Convert to...
33130 (print_diff_line): ...this.
33131 (edited_file::get_effective_line_count): New function.
33132 (edited_line::edited_line): Initialize new field m_predecessors.
33133 (edited_line::~edited_line): Clean up m_predecessors.
33134 (edited_line::apply_fixit): Handle newlines.
33135 (edited_line::get_effective_line_count): New function.
33136 (edited_line::print_content): New function.
33137 (edited_line::print_diff_lines): New function.
33138 (selftest::test_applying_fixits_insert_containing_newline): New
33140 (selftest::test_applying_fixits_replace_containing_newline): New
33142 (selftest::insert_line): New function.
33143 (selftest::test_applying_fixits_multiple_lines): Add example of
33145 (selftest::edit_context_c_tests): Call the new tests.
33147 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33149 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33150 parameter cand. Update dump information.
33151 (get_computation_cost): Update uses.
33153 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33155 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33156 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33157 (get_computation_at, rewrite_use_address): Update use of
33158 get_computation_aff.
33160 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33162 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33163 (get_computation): Delete.
33164 (get_computation_cost): Implement like get_computation_cost_at.
33165 Use get_computation_at.
33166 (get_computation_cost_at): Delete.
33167 (rewrite_use_nonlinear_expr): Use get_computation_at.
33168 (rewrite_use_compare, remove_unused_ivs): Ditto.
33170 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33172 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33174 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33176 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33177 (ivopts_global_cost_for_size): Rename parameter and update uses.
33178 (iv_ca_recount_cost): Update uses.
33179 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33180 candidates seperately in n_invs and n_cands.
33181 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33183 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33185 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33186 (find_inv_vars_cb): New.
33187 (find_depends): Renamed to ...
33188 (find_inv_vars): ... this.
33189 (add_candidate_1, force_var_cost): Call find_inv_vars.
33190 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33192 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33194 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33195 inv_vars. Add inv_exprs.
33196 (struct iv_cand): Rename depends_on to inv_vars.
33197 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33198 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33199 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33200 (dump_cand): Dump inv_vars.
33201 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33202 (record_invariant, find_depends, add_candidate_1): Ditto.
33203 (set_group_iv_cost, force_var_cost): Ditto.
33204 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33205 (get_computation_cost_at, get_computation_cost): Ditto.
33206 (determine_group_iv_cost_generic): Ditto.
33207 (determine_group_iv_cost_address): Ditto.
33208 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33209 (determine_group_iv_costs): Ditto.
33210 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33211 (iv_ca_set_remove_invariants): Renamed to ...
33212 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33213 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33214 (iv_ca_set_add_invariants): Renamed to ...
33215 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33216 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33217 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33218 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33219 (create_new_ivs): Remove useless dump.
33221 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33223 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33225 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33226 (iv_ca_set_no_cp, create_new_iv): Ditto.
33228 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33230 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33232 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33234 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33235 function tree_check2.
33237 2017-05-02 Martin Liska <mliska@suse.cz>
33239 * doc/gcov.texi: Add missing preposition.
33240 * gcov.c (function_info::function_info): Properly fill up
33241 all member variables.
33243 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33245 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33247 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33249 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33251 2017-05-02 Martin Liska <mliska@suse.cz>
33254 * lto-streamer-in.c (lto_read_tree_1): Remove
33255 LTO_STREAMER_DEBUG.
33256 * lto-streamer.c (struct tree_hash_entry): Likewise.
33257 (struct tree_entry_hasher): Likewise.
33258 (tree_entry_hasher::hash): Likewise.
33259 (tree_entry_hasher::equal): Likewise.
33260 (lto_streamer_init): Likewise.
33261 (lto_orig_address_map): Likewise.
33262 (lto_orig_address_get): Likewise.
33263 (lto_orig_address_remove): Likewise.
33264 * lto-streamer.h: Likewise.
33265 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33266 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33268 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33270 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33271 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33272 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33273 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33274 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33275 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33276 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33277 (mm_maskz_sub_ss): New intrinsics.
33278 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33279 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33280 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33281 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33282 (__builtin_ia32_subss_mask_round): New builtins.
33283 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33284 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33285 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33287 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33288 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33290 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33293 2017-05-02 Martin Jambor <mjambor@suse.cz>
33295 PR tree-optimization/78687
33296 * tree-sra.c (access): New field parent.
33297 (process_subtree_disqualification): New function.
33298 (disqualify_candidate): Call it.
33299 (build_accesses_from_assign): Reset write flag if creating an
33301 (build_access_subtree): Fill in parent field and also prpagate
33302 down grp_write flag.
33303 (create_artificial_child_access): New parameter set_grp_write, set
33304 grp_write to its value.
33305 (propagate_subaccesses_across_link): Also propagate grp_write flag
33307 (propagate_all_subaccesses): Push the closest parent back to work
33308 queue if add_access_to_work_queue returned true.
33310 2017-05-02 Richard Biener <rguenther@suse.de>
33312 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33313 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33314 -fstrict-overflow documentation.
33315 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33316 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33317 flag_strict_overflow.
33318 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33319 * lto-opts.c (lto_write_options): Do not stream it.
33320 * lto-wrapper.c (merge_and_complain): Do not handle it.
33321 * opts.c (default_options_table): Do not set -fstrict-overflow.
33322 (finish_options): Likewise do not clear it when sanitizing.
33323 * simplify-rtx.c (simplify_const_relational_operation): Do not
33324 test flag_strict_overflow.
33326 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33328 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33329 using enabled attribute.
33330 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33331 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33332 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33333 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33334 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33335 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33336 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33337 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33338 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33339 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33341 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33343 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33345 2017-05-02 Richard Biener <rguenther@suse.de>
33347 PR tree-optimization/80591
33349 2017-04-10 Richard Biener <rguenther@suse.de>
33351 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33354 2017-05-02 Richard Biener <rguenther@suse.de>
33356 PR tree-optimization/80549
33357 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33358 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33359 headers if they do not have a preheader.
33361 2017-05-02 Martin Liska <mliska@suse.cz>
33364 * common.opt: Fix typo.
33365 * doc/invoke.texi: Likewise.
33367 2017-05-01 Jan Beulich <jbeulich@suse.com>
33369 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33370 swapping, add (x,x,m,x,n) alternative.
33372 2017-05-01 Nathan Sidwell <nathan@acm.org>
33374 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33375 unnecessary unadjusted_alignment check.
33377 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33380 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33382 * gimplify.c (gimplify_cilk_detach): New function.
33383 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33384 * tree-core.h: Document EXPR_CILK_SPAWN.
33385 * tree.h (EXPR_CILK_SPAWN): Define.
33387 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33389 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33390 to use new fixit_hint representation, using the "replace" logic.
33391 (get_line_span_for_fixit_hint): Likewise.
33392 (layout::print_any_fixits): Likewise.
33393 (selftest::test_one_liner_many_fixits): Rename to...
33394 (selftest::test_one_liner_many_fixits_1): ...this, and update
33395 comment and expected output to reflect that the multiple fix-it
33396 hints are now consolidated into one insertion.
33397 (selftest::test_one_liner_many_fixits_2): New test.
33398 (selftest::test_diagnostic_show_locus_one_liner): Update for
33400 (selftest::test_fixit_consolidation): Update for fix-it API
33402 * diagnostic.c (print_parseable_fixits): Likewise.
33403 * edit-context.c (edited_line::m_line_events): Convert from
33404 auto_vec <line_event *> to auto_vec <line_event>.
33405 (class line_event): Convert from abstract base class to a concrete
33406 class, taking over the role of replace_event.
33407 (class insert_event): Delete.
33408 (class replace_event): Rename to class line_event. Convert to
33410 (edit_context::add_fixits): Reimplement.
33411 (edit_context::apply_insert): Delete.
33412 (edit_context::apply_replace): Rename to...
33413 (edit_context::apply_fixit): ...this. Convert to half-open range.
33414 (edited_file::apply_insert): Delete.
33415 (edited_file::apply_replace): Rename to...
33416 (edited_file::apply_fixit): ...this.
33417 (edited_line::~edited_line): Drop deletion of events.
33418 (edited_line::apply_insert): Delete.
33419 (edited_line::apply_replace): Rename to...
33420 (edited_line::apply_fixit): ...this. Convert to half-open range.
33421 Update for change to type of m_line_events.
33422 * edit-context.h (edit_context::apply_insert): Delete.
33423 (edit_context::apply_replace): Rename to...
33424 (edit_context::apply_fixit): ...this.
33426 2017-05-01 Martin Sebor <msebor@redhat.com>
33428 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33431 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33434 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33435 __get_cpuid_max returns 0.
33436 (__get_cpuid_count): Ditto.
33438 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33440 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33441 replacement expression is another instance of one of its arguments.
33443 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33446 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33447 check for stack push/pop autoinc.
33448 * config/i386/i386.c (ix86_agi_dependent): Return false
33449 if the only reason why modified_in_p returned true is that
33450 addr is SP based and set_insn is a push or pop.
33452 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33454 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33457 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33460 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33461 (account_size_time): Use two predicates - exec_pred and
33463 (evaluate_conditions_for_known_args): Compute both clause and
33465 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33466 (inline_summary_t::duplicate): Update.
33467 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33469 (compute_inline_parameters): Likewise.
33470 (estimate_edge_size_and_time): Update caluclation of time.
33471 (estimate_node_size_and_time): Compute both time and nonspecialized
33473 (estimate_ipcp_clone_size_and_time): Update.
33474 (inline_merge_summary): Update.
33475 (do_estimate_edge_time): Update.
33476 (do_estimate_edge_size): Update.
33477 (do_estimate_edge_hints): Update.
33478 (inline_read_section, inline_write_summary): Stream both new predicates.
33479 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33481 (compute_inlined_call_time): Cleanup.
33482 (big_speedup_p): Update.
33483 (edge_badness): Update.
33484 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33485 (size_time_entry): Replace predicate by exec_predicate and
33486 nonconst_predicate.
33487 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33488 (estimate_edge_time): Return also nonspec_time.
33489 (reset_edge_growth_cache): Update.
33491 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33493 PR rtl-optimization/80491
33494 * ifcvt.c (noce_process_if_block): When looking for x setter
33495 with missing else_bb, don't check only the insn right before
33496 cond_earliest, but look for the last insn that x is modified in
33497 within the same bb.
33499 PR rtl-optimization/80491
33500 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33502 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33504 PR tree-optimization/80487
33505 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33507 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33509 PR tree-optimization/79697
33510 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33511 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33512 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33514 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33515 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33517 2017-04-28 Martin Sebor <msebor@redhat.com>
33519 PR tree-optimization/80523
33520 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33521 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33523 (maybe_warn, format_directive, parse_directive): Use new functions.
33524 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33526 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33528 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33530 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33532 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33533 target_header_dir): Set correctly.
33534 * configure: Regenerated.
33535 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33536 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33537 instead of SYSTEM_HEADER_DIR.
33539 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33541 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33542 (estimate_local_effects): Likewise.
33543 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33544 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33545 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33546 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33547 do_estimate_edge_time, estimate_edge_time): Likewise.
33548 * ipa-inline-analysis.c (estimate_node_size_and_time,
33549 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33550 (estimate_time_after_inlining): Remove.
33552 2017-04-28 Martin Liska <mliska@suse.cz>
33554 * doc/gcov.texi: Enhance documentation of gcov.
33556 2017-04-28 Martin Liska <mliska@suse.cz>
33558 * doc/gcov.texi: Sort options in alphabetic order.
33559 * doc/gcov-dump.texi: Likewise.
33560 * doc/gcov-tool.texi: Likewise.
33561 * gcov.c (print_usage): Likewise.
33562 * gcov-dump.c (print_usage): Likewise.
33563 * gcov-tool.c (print_merge_usage_message): Likewise.
33564 (print_rewrite_usage_message): Likewise.
33565 (print_overlap_usage_message): Likewise.
33567 2017-04-28 Martin Liska <mliska@suse.cz>
33569 PR gcov-profile/53915
33570 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33572 2017-04-28 Martin Liska <mliska@suse.cz>
33574 PR gcov-profile/79891
33575 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33576 is marked by compiler as living on a line.
33577 (get_cycles_count): Remove usage of the union.
33578 (output_intermediate_file): Likewise.
33579 (find_source): Fix GNU coding style.
33580 (accumulate_line_counts): Remove old non-all block mode.
33581 (output_lines): Remove usage of the union.
33582 * profile.c (output_location): Include all BBs, even if
33583 belonging to a same line (and file) as a previous BB.
33585 2017-04-28 Martin Liska <mliska@suse.cz>
33587 * gcov.c (process_args): Handle new argument 'w'.
33588 (read_graph_file): Assign ID to BBs.
33589 (output_branch_count): Display BB # if verbose flag is set.
33590 (output_lines): Likewise for arcs.
33591 (print_usage): Add '--verbose' option help.
33592 * doc/gcov.texi: Document --verbose (-w) option.
33594 2017-04-28 Martin Liska <mliska@suse.cz>
33596 * gcov.c (struct block_location_info): New struct.
33597 (process_file): Fill up the new structure.
33598 (read_graph_file): Replace usage of encoding by the newly added
33600 (add_line_counts): Likewise.
33601 (accumulate_line_counts): Remove usage of the union.
33602 (function_info::function_info): New function.
33603 (function_info::~function_info): Likewise.
33604 (process_file): Call delete instead of release_function.
33605 (release_function): Release the function.
33606 (release_structures): Call delete instead of release_function.
33607 (solve_flow_graph): Replace usage of num_blocks.
33608 (find_exception_blocks): Likewise.
33609 (output_lines): Fix GNU coding style.
33611 2017-04-28 Martin Liska <mliska@suse.cz>
33614 * coverage.c (coverage_remove_note_file): New function.
33615 * coverage.h: Declare the function.
33616 * toplev.c (finalize): Clean if an error has been seen.
33618 2017-04-28 Martin Liska <mliska@suse.cz>
33620 PR gcov-profile/80031
33621 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33622 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33623 * gcov.c (read_graph_file): Read just number of blocks.
33624 * profile.c (branch_prob): Do not stream 0 flags per a basic
33627 2017-04-28 Martin Liska <mliska@suse.cz>
33629 * gcov-dump.c (tag_*): Add new argument to declarations.
33630 (dump_gcov_file): Likewise.
33631 (tag_blocks): Add and use new argument depth.
33632 (tag_arcs): Likewise.
33633 (tag_lines): Likewise.
33634 (tag_counters): Likewise.
33635 (tag_summary): Likewise.
33636 (dump_working_sets): Use depth to do a proper indentation.
33638 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33641 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33642 * symtab.c (symtab_node::debug_symtab): Move definition here.
33644 2017-04-28 Richard Biener <rguenther@suse.de>
33646 * lto-streamer.h (LTO_major_version): Bump to 7.
33648 2017-04-28 Richard Biener <rguenther@suse.de>
33650 * tree-vrp.c (assert_info): New struct.
33651 (add_assert_info): New helper.
33652 (register_edge_assert_for_2): Refactor to add asserts to a vector
33654 (register_edge_assert_for_1): Likewise.
33655 (register_edge_assert_for): Likewise.
33656 (finish_register_edge_assert_for): New helper actually registering
33657 asserts where live on edge.
33658 (find_conditional_asserts): Adjust.
33659 (find_switch_asserts): Likewise.
33660 (evrp_dom_walker::try_find_new_range): Generalize.
33661 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33663 2017-04-27 Marek Polacek <polacek@redhat.com>
33666 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33667 arg10 and arg11 to itype.
33669 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33671 * doc/extend.texi (Object Size Checking): Improve grammar.
33673 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33676 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33677 that the logic for permitting reciprocal estimates matches that
33680 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33683 * tree.c (type_cache_hasher::equal): Only compare
33684 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33685 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33686 non-aggregate element types.
33687 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33688 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33690 2017-04-27 Richard Biener <rguenther@suse.de>
33692 PR middle-end/80533
33693 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33694 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33695 keeping a reference to a trailing array.
33697 2017-04-27 Richard Biener <rguenther@suse.de>
33699 PR middle-end/80539
33700 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33701 being in loop-closed SSA form conservatively.
33702 (chrec_fold_multiply_poly_poly): Likewise.
33704 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33706 PR middle-end/79665
33707 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33708 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33710 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33713 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33714 (aarch64_function_arg_alignment): Return unsigned int again, but still
33715 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33716 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33717 Don't emit -Wpsabi note.
33718 (aarch64_function_arg_boundary): Likewise.
33719 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33722 2017-04-26 Nathan Sidwell <nathan@acm.org>
33724 * tree.h (crc32_unsigned_n): Declare.
33725 (crc32_unsigned, crc32_unsigned): Make inline.
33726 * tree.c (crc32_unsigned_bits): Replace with ...
33727 (crc32_unsigned_n): ... this.
33728 (crc32_unsigned, crc32_byte): Remove.
33729 (crc32_string): Remove unnecessary braces.
33731 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33733 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33734 * ipa-inline-analysis.c (MAX_TIME): Remove.
33735 (account_size_time): Use sreal for time.
33736 (dump_inline_summary): Update.
33737 (estimate_function_body_sizes): Update.
33738 (estimate_edge_size_and_time): Update.
33739 (estimate_calls_size_and_time): Update.
33740 (estimate_node_size_and_time): Update.
33741 (inline_merge_summary): Update.
33742 (inline_update_overall_summary): Update.
33743 (estimate_time_after_inlining): Update.
33744 (inline_read_section): Update.
33745 (inline_write_summary): Update.
33746 * ipa-inline.c (compute_uninlined_call_time): Update.
33747 (compute_inlined_call_time): Update.
33748 (recursive_inlining): Update.
33749 (inline_small_functions): Update.
33750 (dump_overall_stats): Update.
33751 * ipa-inline.h: Include sreal.h.
33752 (size_time_entry): Turn time to sreal.
33753 (inline_summary): Turn self_time nad time to sreal.
33755 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33757 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33759 (sreal::stream_out, sreal::stream_in): New.
33760 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33762 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33764 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33767 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33770 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33772 (dimode_scalar_chain::compute_convert_gain): Ditto.
33773 (dimode_scalar_chain::make_vector_copies): Ditto.
33774 (dimode_scalar_chain::convert_reg): Ditto.
33775 (dimode_scalar_chain::convert_insn): Ditto.
33776 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33777 (VI248_AVX512BW_1): New mode iterator.
33778 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33779 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33782 2017-04-25 Martin Sebor <msebor@redhat.com>
33784 PR tree-optimization/80497
33785 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33786 constants are representable in HOST_WIDE_INT.
33787 (parse_directive): Ditto.
33789 2017-04-25 Martin Sebor <msebor@redhat.com>
33792 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33793 (new_zero_array): Adjust signature.
33794 (dom_info::dom_init): Used unsigned rather that size_t.
33795 (dom_info::dom_info): Same.
33797 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33798 Jakub Jelinek <jakub@redhat.com>
33801 * config/arm/arm.c: Include gimple.h.
33802 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33803 returns negative, increment ncrn only if it returned positive.
33804 (arm_needs_doubleword_align): Return int instead of bool,
33805 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33806 members, but if there is any such non-FIELD_DECL
33807 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33808 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33809 returns negative, increment nregs only if it returned positive.
33810 (arm_setup_incoming_varargs): Likewise.
33811 (arm_function_arg_boundary): Emit -Wpsabi note if
33812 arm_needs_doubleword_align returns negative, return
33813 DOUBLEWORD_ALIGNMENT only if it returned positive.
33815 2017-04-25 Marek Polacek <polacek@redhat.com>
33818 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33819 first argument to type.
33821 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33824 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33825 type checks to test for compatibility instead of equality.
33827 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33828 Jakub Jelinek <jakub@redhat.com>
33831 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33833 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33834 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33835 the alignment computation, but return their maximum in warn_alignment.
33836 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33837 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33839 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33840 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33843 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33845 * config/arc/simdext.md (dmpyh): Fix typo.
33847 2017-04-25 Richard Biener <rguenther@suse.de>
33849 PR tree-optimization/80492
33850 * alias.c (compare_base_decls): Handle registers with asm
33851 specification conservatively.
33852 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33853 compare_base_decls returning dont-know properly.
33855 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33857 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33858 (legitimate_offset_address_p): New function.
33859 (arc_legitimate_address_p): Use above function.
33861 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33863 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33865 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33867 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33868 ACCH registers whenever they are available.
33870 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33872 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33873 double regs fix when not used.
33875 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33877 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33879 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33880 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33882 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33884 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33885 when emitting short ADD instructions.
33887 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33889 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33891 (cmpsi_cc_c_insn): Likewise.
33892 (cbranchsi4_scratch): Compute proper instruction length using
33893 compact_hreg_operand.
33894 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33896 2017-04-25 Richard Biener <rguenther@suse.de>
33898 PR middle-end/80509
33899 * passes.c (pass_manager::pass_manager): Initialize
33900 m_name_to_pass_map.
33902 2017-04-25 Richard Biener <rguenther@suse.de>
33904 PR tree-optimization/79201
33905 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33907 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33910 * config/s390/vector.md: Split MEM->GPR vector moves for
33911 non-s_operand addresses.
33913 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33916 * config/s390/predicates.md (reload_const_wide_int_operand): New
33918 * config/s390/s390.md ("movti"): Remove d/P alternative.
33919 ("movti_bigconst"): New pattern definition.
33921 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33924 * s390-protos.h (s390_expand_cs_hqi): Removed.
33925 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33926 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33927 modes as well as CCZ1mode and CCZmode.
33928 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33929 signature of s390_emit_compare_and_swap.
33930 (s390_expand_cs_hqi): Likewise, make static.
33931 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33932 compare-and-swap, in some cases.
33933 (s390_expand_cs): Wrapper function.
33934 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33936 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33937 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33938 patterns for small and large integers. Forbid symref memory operands.
33939 Move expander to s390.c. Require cc register.
33940 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33941 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33942 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33943 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33944 symref memory operands. Remove CC mode and call s390_match_ccmode
33946 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33948 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33950 * config/s390/s390.md (define_peephole2): New peephole to help
33951 combining the load-and-test pattern with volatile memory.
33953 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33955 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33956 with CCZmode for TARGET_Z196.
33958 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33960 PR rtl-optimization/80501
33961 * combine.c (make_compound_operation_int): Set subreg_code to SET
33962 even for AND with mask of the sign bit of mode.
33964 PR rtl-optimization/80500
33965 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33966 sum's initial value.
33968 2017-04-25 Julian Brown <julian@codesourcery.com>
33969 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33971 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33973 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33975 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33977 2017-04-25 Julian Brown <julian@codesourcery.com>
33978 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33980 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33981 (thunderx2t99_sha): New Reservation.
33983 2017-04-25 Julian Brown <julian@codesourcery.com>
33984 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33986 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33987 type for 1-element load.
33989 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33991 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33993 2017-04-24 Martin Jambor <mjambor@suse.cz>
33995 PR tree-optimization/80293
33996 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33997 char arrays not totally scalarizable if it is false.
33998 (analyze_all_variable_accesses): Pass correct value in the new
33999 parameter. Add a statistics counter.
34001 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34003 PR middle-end/79931
34004 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34006 2017-04-24 Richard Biener <rguenther@suse.de>
34008 PR tree-optimization/80494
34009 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34010 out for complex types.
34012 2017-04-24 Richard Biener <rguenther@suse.de>
34014 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34015 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34016 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34017 (DFS): Adjust and never fail.
34018 (sccvn_dom_walker::fail): Remove.
34019 (sccvn_dom_walker::before_dom_children): Adjust.
34020 (run_scc_vn): Likewise and never fail.
34021 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34022 (pass_fre::execute): Likewise.
34024 2017-04-24 Richard Biener <rguenther@suse.de>
34026 PR tree-optimization/79725
34027 * tree-ssa-sink.c (statement_sink_location): Return whether
34028 failure reason was zero uses. Move that check later.
34029 (sink_code_in_bb): Deal with zero uses by removing the stmt
34032 2017-04-24 Richard Biener <rguenther@suse.de>
34035 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34036 pointer-based references.
34038 2017-04-24 Richard Biener <rguenther@suse.de>
34041 * pass_manager.h (pass_manager::operator new): Remove.
34042 (pass_manager::operator delete): Likewise.
34043 * passes.c (pass_manager::operator new): Remove.
34044 (pass_manager::operator delete): Likewise.
34045 (pass_manager::pass_manager): Zero individual pass members.
34047 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34050 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34051 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34052 Check "XEXP (src, 1)" operand here.
34053 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34054 Check "XEXP (src, 1)" operand here.
34055 (dimode_scalar_chain::make_vector_copies): Detect count register
34056 of a shift instruction. Zero extend count register from QImode
34057 to DImode to satisfy vector shift pattern count operand predicate.
34058 Substitute vector shift count operand with a DImode copy.
34059 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34062 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34064 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34065 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34066 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34067 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34068 (UNSPEC_NOREX_MEM): Remove definition.
34070 2017-04-21 Richard Biener <rguenther@suse.de>
34072 PR tree-optimization/79547
34073 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34074 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34075 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34076 without any constraints.
34078 2017-04-21 Richard Biener <rguenther@suse.de>
34080 PR tree-optimization/78847
34081 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34083 2017-04-21 Richard Biener <rguenther@suse.de>
34085 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34086 (build_distinct_type_copy): Likewise.
34087 (build_variant_type_copy): Likewise.
34088 * tree.c (build_qualified_type): Pass down mem-stat info.
34089 (build_distinct_type_copy): Likewise.
34090 (build_variant_type_copy): Likewise.
34092 2017-04-21 Richard Biener <rguenther@suse.de>
34094 PR tree-optimization/80237
34095 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34097 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34098 for a simplified result.
34100 2016-04-21 Richard Biener <rguenther@suse.de>
34102 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34103 sth as strict as a simple_iv but a chrec without symbols and an
34104 operand defined in the loop we are peeling (and not some subloop).
34105 (propagate_constants_for_unrolling): Propagate all constants.
34107 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34110 * config/i386/i386.c (print_reg): Remove assert for disalowed
34111 regno values, call output_operand_lossage instead.
34113 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34116 * config/i386/constraints.md (Yc): New register constraint.
34117 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34118 Use Yc constraint for alternative 2 of operand 0. Remove
34119 preferred_for_speed attribute.
34121 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34123 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34124 lastprivate clauses in SIMT case.
34126 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34128 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34130 2017-04-20 Richard Biener <rguenther@suse.de>
34132 PR tree-optimization/57796
34133 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34134 as N scalar stores.
34135 (vect_model_load_cost): Cost gathers as N scalar loads.
34137 2017-04-20 Richard Biener <rguenther@suse.de>
34139 * ggc-page.c (ggc_allocated_p): Rename to ...
34140 (safe_lookup_page_table_entry): ... this and return the lookup
34142 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34144 2017-04-20 Richard Biener <rguenther@suse.de>
34146 PR tree-optimization/80453
34147 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34148 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34149 from the conditions.
34150 (vn_phi_eq): Pass them down.
34151 (vn_phi_lookup): Record them.
34152 (vn_phi_insert): Likewise.
34154 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34156 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34157 uninitialized variable warning to avoid buffer overrun.
34159 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34162 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34163 is suppressed for '{ 0 }' in C.
34165 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34167 * BASE-VER: Set to 8.0.0.
34169 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34171 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34172 priority .init_array and .fini_array section with SECTION_NOTYPE
34175 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34177 PR middle-end/80423
34178 * tree.h (build_array_type): Add typeless_storage default argument.
34179 * tree.c (type_cache_hasher::equal): Also compare
34180 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34181 (build_array_type): Add typeless_storage argument, set
34182 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34184 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34185 (build_array_type): Likewise. Add typeless_storage argument.
34187 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34188 Jakub Jelinek <jakub@redhat.com>
34190 PR tree-optimization/80426
34191 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34192 operation on symbolic operands, also compute the overflow for the
34193 invariant part when the operation degenerates into a negation.
34195 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34198 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34199 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34202 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34204 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34207 * config/avr/avr.c (tree.h): Include it.
34208 (cgraph.h): Include it.
34209 (avr_encode_section_info): Don't warn for uninitialized progmem
34210 variable if it's just an alias.
34212 2017-04-19 Richard Biener <rguenther@suse.de>
34215 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34216 when needed by AutoPGO.
34218 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34221 * doc/lto.texi: Remove an extra 'that'.
34223 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34225 PR rtl-optimization/80429
34226 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34227 are only used in debug insns.
34229 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34230 Vladimir Makarov <vmakarov@redhat.com>
34232 * config/sparc/predicates.md (input_operand): Add comment. Return
34233 true for any memory operand when LRA is in progress.
34234 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34236 2017-04-18 Jeff Law <law@redhat.com>
34239 * mips.md ({return,simple_return}_internal): Do not overwrite
34242 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34244 PR tree-optimization/80443
34245 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34246 instead of adding 1, subtract -1 and similarly instead of subtracting
34249 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34251 PR rtl-optimization/80357
34252 * haifa-sched.c (tmp_bitmap): New variable.
34253 (model_recompute): Handle duplicate use records.
34254 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34255 (free_global_sched_pressure_data): Free it.
34257 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34260 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34261 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34262 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34263 instead of SYSTEM_HEADER_DIR.
34265 2017-04-18 Jeff Law <law@redhat.com>
34267 PR middle-end/80422
34268 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34269 predecessors after walking up the insn chain.
34271 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34274 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34275 sizetype type into debug info.
34277 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34280 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34281 unneeded test for TARGET_UPPER_REGS_SF.
34282 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34284 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34287 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34288 instead of gsi_after_labels.
34290 2017-04-18 Jeff Law <law@redhat.com>
34292 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34296 2017-04-13 Jeff Law <law@redhat.com>
34297 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34298 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34300 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34303 * config/avr/avr.c (intl.h): Include it.
34304 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34306 2017-04-18 Martin Liska <mliska@suse.cz>
34308 PR gcov-profile/78783
34309 * gcov-tool.c (gcov_output_files): Validate that destination
34310 file is either removed by the tool or by a user.
34312 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34313 Guy Benyei <guybe@mellanox.com>
34315 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34316 block, and do not negate it, the stored id is already negative.
34318 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34320 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34322 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34325 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34326 masks of options that should be turned off if the VSX vector
34327 options are turned off.
34328 (OTHER_P8_VECTOR_MASKS): Likewise.
34329 (OTHER_VSX_VECTOR_MASKS): Likewise.
34330 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34331 rs6000_disable_incompatible_switches to validate no type switches
34333 (rs6000_incompatible_switch): New function to disallow turning on
34334 other vector options if -mno-vsx, -mno-power8-vector, or
34335 -mno-power9-vector are specified.
34337 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34339 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34341 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34343 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34344 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34345 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34346 (ARG_POINTER_CFA_OFFSET): Likewise.
34348 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34350 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34351 conditions to take advantage of various optimizations.
34353 2017-04-13 Jeff Law <law@redhat.com>
34355 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34356 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34357 (zero_extendsidi2_dext): Likewise.
34359 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34362 * fold-const.c (fold_ternary_loc): Revert
34363 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34366 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34368 PR rtl-optimization/80343
34369 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34370 new scratch pseudo.
34372 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34375 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34376 to ubsan_encode_value.
34378 2017-04-13 Jeff Law <law@redhat.com>
34380 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34381 appearing in DEBUG_INSNs.
34383 2017-04-13 Martin Liska <mliska@suse.cz>
34385 PR gcov-profile/80413
34386 * gcov-io.c (gcov_write_string): Copy to buffer just when
34387 allocated size is greater than zero.
34389 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34392 * dwarf2out.c (decls_for_scope): Ignore declarations of
34393 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34395 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34398 * ipa-visibility.c (non_local_p): Fix typos.
34399 (localize_node): When localizing symbol in same comdat group,
34400 dissolve the group only when we know external symbols are going
34402 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34404 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34406 PR tree-optimization/79390
34407 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34408 order does not result in usable sequence, retry with reversed operand
34414 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34415 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34416 op0 instead of fold_convert_loc (loc, type, arg0).
34418 2017-04-12 Jeff Law <law@redhat.com>
34420 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34421 has a delay slot in the generated code.
34423 * config/cris/cris.md (cris_preferred_reload_class): Return
34424 GENNONACR_REGS rather than GENERAL_REGS.
34426 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34429 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34430 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34431 signedness of the result type.
34433 2017-04-12 Richard Biener <rguenther@suse.de>
34434 Jeff Law <law@redhat.com>
34436 PR tree-optimization/80359
34437 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34438 trim stores to TARGET_MEM_REFs.
34440 2017-04-12 Richard Biener <rguenther@suse.de>
34442 PR tree-optimization/79390
34443 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34444 threading case even more.
34446 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34449 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34450 for quad_address_p for TImode, instead of just not indexed_address.
34452 2017-04-12 Richard Biener <rguenther@suse.de>
34453 Bernd Edlinger <bernd.edlinger@hotmail.de>
34455 PR middle-end/79671
34456 * alias.c (component_uses_parent_alias_set_from): Handle
34457 TYPE_TYPELESS_STORAGE.
34458 (get_alias_set): Likewise.
34459 * tree-core.h (tree_type_common): Add typeless_storage flag.
34460 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34461 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34462 for types containing members with TYPE_TYPELESS_STORAGE.
34463 (place_field): Likewise.
34464 (layout_type): Likewise for ARRAY_TYPE.
34465 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34466 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34467 TYPE_TYPELESS_STORAGE.
34468 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34470 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34473 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34474 first argument to type.
34476 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34480 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34481 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34482 (rs6000_expand_binop_builtin): Likewise.
34483 (rs6000_expand_ternop_builtin): Likewise; also add missing
34484 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34485 vshasigma built-ins.
34486 * doc/extend.texi: Document that vec_xxpermdi's third argument
34487 must be a constant.
34489 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34491 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34492 Use shift_const cost parameter when calculating gain of STV shifts.
34494 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34496 PR rtl-optimization/70478
34497 * lra-constraints.c (process_alt_operands): Check memory for
34498 disfavoring memory insn operand.
34500 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34502 PR middle-end/80100
34503 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34504 left shift in unsigned HOST_WIDE_INT type.
34506 PR rtl-optimization/80385
34507 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34508 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34511 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34512 if they have any depend clauses.
34514 2017-04-11 Martin Liska <mliska@suse.cz>
34517 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34518 * ipa-split.c (split_function): Create a local comdat symbol
34519 if caller is in a comdat group.
34521 2017-04-11 Martin Liska <mliska@suse.cz>
34524 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34527 2017-04-11 Martin Sebor <msebor@redhat.com>
34529 PR middle-end/80364
34530 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34531 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34533 (directive::set_width, directive::set_precision, format_character):
34535 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34538 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34541 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34542 conflict, set target->arch_name instead of target->cpu_name.
34544 2017-04-11 Richard Biener <rguenther@suse.de>
34546 PR tree-optimization/80374
34547 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34548 build_zero_cst, remove fold_convertible_p check again.
34550 2017-04-11 Martin Liska <mliska@suse.cz>
34553 * ubsan.c (instrument_object_size): Do not instrument register
34556 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34559 * config/i386/i386-builtin-types.def
34560 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34561 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34562 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34563 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34564 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34565 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34566 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34567 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34568 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34569 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34570 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34571 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34572 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34573 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34574 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34575 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34576 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34577 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34578 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34579 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34580 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34581 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34582 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34583 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34584 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34585 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34586 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34587 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34588 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34589 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34590 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34591 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34592 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34593 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34594 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34595 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34596 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34597 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34598 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34599 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34600 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34601 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34602 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34603 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34604 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34605 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34607 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34608 flag to second_arg_count, handle 4 argument function type _COUNT
34609 aliases, handle second_arg_count on second argument rather than last.
34611 2017-04-10 Jeff Law <law@redhat.com>
34613 PR tree-optimization/80374
34614 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34615 record anything if we can not convert integer_zero_node to the
34618 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34621 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34622 Enhance special handling given to the TARGET_P9_MINMAX option in
34623 relation to certain other options.
34625 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34627 PR tree-optimization/80153
34628 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34629 remove POINTER_PLUS_EXPR's base part directly, rather than through
34632 2017-04-10 Richard Biener <rguenther@suse.de>
34633 Bin Cheng <bin.cheng@arm.com>
34635 PR tree-optimization/80153
34636 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34637 the first element of pointer type aff_tree. Build result expr in
34639 (add_elt_to_tree): Convert to type unconditionally. Remove other
34640 fold_convert calls.
34641 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34642 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34644 2017-04-10 Richard Biener <rguenther@suse.de>
34646 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34649 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34651 PR rtl-optimization/70478
34652 * lra-constraints.c (curr_small_class_check): New.
34653 (update_and_check_small_class_inputs): New.
34654 (process_alt_operands): Update curr_small_class_check. Disfavor
34655 alternative insn memory operands. Check available regs for small
34658 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34661 * config/mips/mips.opt (-mvirt): Update description.
34662 * doc/invoke.texi (-mvirt): Likewise.
34664 2017-04-10 Richard Biener <rguenther@suse.de>
34666 PR middle-end/80362
34667 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34668 looking for NEGATE_EXPR in -A / -B to A / B folding.
34670 2017-04-10 Martin Liska <mliska@suse.cz>
34672 PR gcov-profile/80224
34673 * gcov.c (print_usage): Fix usage string.
34674 (get_gcov_intermediate_filename): Remove.
34675 (output_gcov_file): Use both for normal and intermediate format.
34676 (generate_results): Do not initialize special file for
34677 intermediate format.
34679 2017-04-10 Richard Biener <rguenther@suse.de>
34681 PR tree-optimization/80304
34682 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34685 2017-04-10 Nathan Sidwell <nathan@acm.org>
34688 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34689 (rs6000_init_builtins): Use it.
34691 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34693 * config/arm/arm.md (<mrc>): Add mode to SET source.
34694 (<mrrc>): Likewise.
34696 2017-04-10 Richard Biener <rguenther@suse.de>
34698 PR middle-end/80344
34699 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34701 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34704 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34705 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34706 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34707 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34708 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34709 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34710 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34711 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34712 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34713 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34714 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34715 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34716 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34717 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34718 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34719 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34720 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34721 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34722 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34723 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34724 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34725 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34726 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34728 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34730 PR rtl-optimization/70478
34731 * lra-constraints.c: Reverse the last patch.
34733 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34735 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34736 Add comment for WCHAR_T.
34738 2017-04-08 Martin Liska <mliska@suse.cz>
34741 2017-04-07 Martin Liska <mliska@suse.cz>
34744 * ipa-split.c (split_function): Add function part to a same comdat
34747 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34750 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34752 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34754 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34755 (vec_pairing): ...to this.
34756 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34757 (rs6000_sched_init): Adjust for name change.
34758 (struct rs6000_sched_context): Likewise.
34759 (rs6000_init_sched_context): Likewise.
34760 (rs6000_set_sched_context): Likewise.
34762 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34768 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34770 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34771 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34772 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34774 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34776 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34778 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34780 PR rtl-optimization/70703
34781 * ira-color.c (update_conflict_hard_regno_costs): Use
34782 int64_t instead of HOST_WIDE_INT.
34784 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34786 PR rtl-optimization/70478
34787 * lra-constraints.c (process_alt_operands): Disfavor alternative
34788 insn memory operands.
34790 2017-04-07 Jeff Law <law@redhat.com>
34792 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34793 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34795 2017-04-07 Martin Liska <mliska@suse.cz>
34798 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34799 Show error message instead of an ICE.
34801 2017-04-07 Martin Liska <mliska@suse.cz>
34804 * ipa-split.c (split_function): Add function part to a same comdat
34807 2017-04-07 Richard Biener <rguenther@suse.de>
34809 PR middle-end/80341
34810 * tree.c (get_unwidened): Also handle ! for_type case for
34812 * convert.c (do_narrow): Split out from ...
34813 (convert_to_integer_1): ... here. Do not pass final truncation
34814 type to get_unwidened for TRUNC_DIV_EXPR.
34816 2017-04-07 Richard Biener <rguenther@suse.de>
34818 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34820 (aff_combination_const): Adjust.
34821 (aff_combination_scale): Likewise.
34822 (aff_combination_add_elt): Likewise.
34823 (aff_combination_add_cst): Likewise.
34824 (aff_combination_convert): Likewise.
34825 (add_elt_to_tree): Likewise. Remove unused argument.
34826 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34828 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34830 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34832 * config/arm/arm.c (arm_default_short_enums): Use
34833 ARM_DEFAULT_SHORT_ENUMS.
34834 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34836 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34839 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34840 members with redundant out-of-class redeclaration.
34842 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34845 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34846 * config/i386/i386.md (*zero_extendsidi2):
34847 Add (?*x,*x) and (?*v,*v) alternatives.
34849 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34852 * config/i386/i386.c (ix86_expand_builtin)
34853 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34854 mode from insn data. Convert operands to insn operand mode.
34855 Copy operands that don't satisfy insn predicate to a register.
34857 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34859 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34862 2017-04-06 Richard Biener <rguenther@suse.de>
34864 PR tree-optimization/80334
34865 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34866 preserve alignment of accesses.
34868 2017-04-06 Richard Biener <rguenther@suse.de>
34870 PR tree-optimization/80262
34871 * tree-sra.c (build_ref_for_offset): Preserve address-space
34873 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34874 Drop useless address-space information on MEM_REF offsets.
34876 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34878 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34880 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34882 PR rtl-optimization/70703
34883 * ira-color.c (update_conflict_hard_regno_costs): Use
34884 HOST_WIDE_INT instead of long.
34886 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34889 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34890 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34892 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34893 for x86_64 target. Handle -m3dnowa option.
34895 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34897 PR rtl-optimization/70703
34898 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34899 (update_conflict_hard_regno_costs): Use long instead of unsigned
34900 arithmetic for cost calculation.
34902 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34903 Bernd Edlinger <bernd.edlinger@hotmail.de>
34906 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34909 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34912 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34913 ptr_mode with Pmode throughout.
34914 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34915 into probe_stack_range and use DImode.
34917 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34920 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34921 call_eh_return is true.
34923 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34925 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34926 Initialize last_match_fntype_index.
34928 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34931 * tree-nvr.c: Include internal-fn.h.
34932 (pass_return_slot::execute): Ignore internal calls without
34935 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34936 Richard Biener <rguenther@suse.de>
34939 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34940 captures used multiple times, except for the last use.
34941 * generic-match-head.c: Include gimplify.h.
34943 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34945 PR tree-optimization/79390
34946 * target.h (struct noce_if_info): Declare.
34947 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34948 * target.def (noce_conversion_profitable_p): New target hook.
34949 * ifcvt.h (struct noce_if_info): New type, moved from ...
34950 * ifcvt.c (struct noce_if_info): ... here.
34951 (noce_conversion_profitable_p): Renamed to ...
34952 (default_noce_conversion_profitable_p): ... this. No longer
34954 (noce_try_store_flag_constants, noce_try_addcc,
34955 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34956 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34957 instead of noce_conversion_profitable_p.
34958 * config/i386/i386.c: Include ifcvt.h.
34959 (ix86_option_override_internal): Don't override
34960 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34961 (ix86_noce_conversion_profitable_p): New function.
34962 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34963 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34964 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34965 * doc/tm.texi: Regenerated.
34967 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34969 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34972 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34975 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34976 instructions for small multiply cores.
34978 2017-04-04 Jeff Law <law@redhat.com>
34980 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34982 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34983 that are not set by the loop over the elements.
34985 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34988 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34989 int mode, convert_modes it to mode as unsigned, otherwise use
34990 lowpart_subreg to mode rather than SImode.
34991 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34992 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34993 Use DImode instead of SImode for the shift count operand.
34994 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34997 2017-04-04 Richard Biener <rguenther@suse.de>
34999 PR middle-end/80281
35000 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35001 arithmetic done for the negate or the plus. Simplify.
35002 (A - (-B) -> A + B): Likewise.
35003 * fold-const.c (split_tree): Make sure to not negate pointers.
35005 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35007 PR rtl-optimization/60818
35008 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35009 a compare of comparisons with the thing compared if this results
35010 in a different machine mode.
35012 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35014 * alias.c (base_alias_check): Fix typo in comment.
35015 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35016 * cgraphunit.c (symbol_table::compile): Likewise.
35017 * collect2.c (maybe_run_lto_and_relink): Likewise.
35018 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35019 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35020 * config/avr/avr.c (avr_map_op_t): Likewise.
35021 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35022 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35023 * config/epiphany/epiphany.md (movcc): Likewise.
35024 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35025 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35027 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35028 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35029 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35030 * config/sh/sh.c (sh_rtx_costs): Likewise.
35031 * fold-const.c (fold_truth_andor): Likewise.
35032 * genautomata.c (collapse_flag): Likewise.
35033 * gengtype.h (struct type::u::s): Likewise.
35034 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35035 * input.c (FORMAT_AMOUNT): Likewise.
35036 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35037 (known_aggs_to_agg_replacement_list): Likewise.
35038 * ipa-inline-analysis.c: Likewise.
35039 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35040 * ipa-polymorphic-call.c
35041 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35042 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35043 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35045 * modulo-sched.c (apply_reg_moves): Likewise.
35046 * omp-expand.c (build_omp_regions_1): Likewise.
35047 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35048 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35049 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35050 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35051 * value-prof.c: Likewise.
35052 * var-tracking.c (val_reset): Likewise.
35054 2017-04-03 Richard Biener <rguenther@suse.de>
35056 PR tree-optimization/80275
35057 * fold-const.c (split_address_to_core_and_offset): Handle
35060 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35062 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35063 descriptors is at least equal to that of functions.
35065 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35067 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35069 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35072 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35073 (mov<IMOD4:mode>): New expander.
35074 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35076 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35078 PR rtl-optimization/79405
35079 * fwprop.c (propagations_left): New variable.
35080 (forward_propagate_into): Decrement it.
35081 (fwprop_init): Initialize it.
35082 (fw_prop): If the variable has reached zero, stop propagating.
35083 (fwprop_addr): Ditto.
35085 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35088 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35089 a FUNCTION_DECL, pass it as decl instead of origin to
35092 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35094 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35097 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35100 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35101 TARGET_VSX_SMALL_INTEGER.
35103 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35105 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35106 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35108 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35110 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35111 extraction from odd-numbered MSA register.
35113 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35115 PR middle-end/80173
35116 * expmed.c (store_bit_field_1): Don't attempt to create
35117 a word subreg out of hard registers wider than word if they
35118 have HARD_REGNO_NREGS of 1 for their mode.
35120 PR middle-end/80163
35121 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35122 conversions to integer types wider than word and pointer.
35125 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35126 (rtx_equal_for_cselib_p): Pass 0 to it.
35127 * cselib.c (cselib_hasher::equal): Likewise.
35128 (rtx_equal_for_cselib_1): Add depth argument. If depth
35129 is 128, don't look up VALUE locs and punt. Increment
35130 depth in recursive calls when walking VALUE locs.
35132 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35134 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35135 (make_gcov_file_name): Use the canonical path name for generating
35137 (read_line): Fix handling of files with ascii null bytes.
35139 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35141 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35142 to initialise a vector register instead
35143 of using a const_int.
35145 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35147 PR translation/80189
35148 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35149 diagnostic messages.
35151 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35154 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35155 (dfp_diex_<mode>): Update mode of operand 1.
35156 * doc/extend.texi (dxex, dxexq): Document change to return type.
35157 (diex, diexq): Document change to argument type.
35159 2017-03-30 Martin Jambor <mjambor@suse.cz>
35162 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35163 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35164 it reflects the signature changes performed at the callee side.
35165 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35166 to cgraph_build_function_type_skip_args.
35167 (build_function_decl_skip_args): Adjust call to the above function.
35169 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35172 * config/i386/sse.md
35173 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35174 register as dest whenever it is a MEM not rtx_equal_p to the
35175 corresponding dup operand, and when forcing into reg move the
35176 reg into the memory afterwards.
35177 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35178 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35179 for the force_reg mode.
35180 (avx512vl_vextractf128<mode>): Use register as dest either
35181 always when a MEM, or when it is a MEM not rtx_equal_p to the
35182 corresponding dup operand, or even not when it is a CONST_VECTOR
35183 depending on the mode and lo vs. hi.
35184 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35186 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35187 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35188 Likewise. Require that operands[2] is even.
35189 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35190 Remove extraneous parens. Require that operands[2] is a multiple
35192 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35193 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35194 disallow memory then.
35196 2017-03-30 Richard Biener <rguenther@suse.de>
35198 PR tree-optimization/77498
35199 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35200 to non-constants over backedges.
35202 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35204 PR rtl-optimization/80233
35205 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35206 as last_combined_insn. Do not test for BARRIER_P separately.
35208 2017-03-29 Andreas Schwab <schwab@suse.de>
35211 * calls.c (prepare_call_address): Convert funexp to Pmode before
35212 copying to temp reg.
35214 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35216 PR tree-optimization/80158
35217 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35218 Handle possible future case of more than one alternate
35220 (replace_rhs_if_not_dup): Likewise.
35221 (replace_one_candidate): Likewise.
35223 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35225 PR rtl-optimization/80193
35226 * ira.c (ira): Do not check allocation for LRA.
35228 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35230 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35231 (nvptx_output_simt_exit): Declare.
35232 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35233 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35234 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35235 (nvptx_declare_function_name): ...here. Emit declaration of local
35236 memory space buffer for omp_simt_enter insn.
35237 (nvptx_output_unisimt_switch): New.
35238 (nvptx_output_softstack_switch): New.
35239 (nvptx_output_simt_enter): New.
35240 (nvptx_output_simt_exit): New.
35241 * config/nvptx/nvptx.h (struct machine_function): New fields
35242 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35243 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35244 (UNSPECV_SIMT_EXIT): Ditto.
35245 (omp_simt_enter_insn): New insn.
35246 (omp_simt_enter): New expansion.
35247 (omp_simt_exit): New insn.
35248 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35250 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35251 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35252 (expand_GOMP_SIMT_EXIT): New.
35253 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35254 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35255 (GOMP_SIMT_EXIT): Ditto.
35256 * target-insns.def (omp_simt_enter): New insn.
35257 (omp_simt_exit): Ditto.
35258 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35260 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35261 (lower_rec_input_clauses): Likewise.
35262 (lower_lastprivate_clauses): Handle SIMT privatization.
35264 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35265 (ompdevlow_adjust_simt_enter): New.
35266 (find_simtpriv_var_op): New.
35267 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35268 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35270 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35271 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35272 (copy_decl_for_dup_finish): Ditto.
35274 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35276 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35279 * config/i386/i386.c (ix86_option_override_internal): Always
35280 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35282 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35284 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35286 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35288 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35289 mark new edge's irreducible flag accordign to it.
35290 (vect_do_peeling): Check loop preheader edge's irreducible flag
35291 and pass it to function slpeel_add_loop_guard.
35293 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35295 PR tree-optimization/80218
35296 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35297 Update block frequencies and counts.
35299 2017-03-28 Richard Biener <rguenther@suse.de>
35301 PR tree-optimization/78644
35302 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35303 of a simplification result we may not use it at all.
35305 2017-03-28 Richard Biener <rguenther@suse.de>
35308 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35309 without arguments, generate default definition of a SSA name.
35311 2017-03-28 Richard Biener <rguenther@suse.de>
35313 PR middle-end/80222
35314 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35315 TYPE_REF_CAN_ALIAS_ALL references.
35316 * fold-const.c (fold_indirect_ref_1): Likewise.
35318 2017-03-28 Martin Liska <mliska@suse.cz>
35321 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35322 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35324 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35325 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35327 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35328 (EXTRA_SPECS): Define.
35329 (SUBTARGET_EXTRA_SPECS): Likewise.
35330 (SUBTARGET_CPP_SPEC): Likewise.
35331 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35332 SUBTARGET_EXTRA_SPECS.
35333 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35335 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35337 * config/arc/simdext.md (vst64_insn): Update pattern.
35338 (vld32wh_insn): Likewise.
35339 (vld32wl_insn): Likewise.
35340 (vld64_insn): Likewise.
35341 (vld32_insn): Likewise.
35343 2017-03-28 Marek Polacek <polacek@redhat.com>
35346 * fold-const.c (fold_comparison): Use protected_set_expr_location
35347 instead of SET_EXPR_LOCATION.
35349 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35351 * tree.c (add_expr): Avoid name lookup warning.
35353 2017-03-27 Jeff Law <law@redhat.com>
35355 PR tree-optimization/80216
35356 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35357 function name. Limit recursion depth.
35358 (record_temporary_equivalences): Corresponding changes.
35360 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35362 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35365 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35368 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35370 * cfgcleanup.c (reg_note_cfa_p): New array.
35371 (insns_have_identical_cfa_notes): New function.
35372 (old_insns_match_p): Don't cross-jump in between /f
35373 and non-/f instructions. If both i1 and i2 are frame related,
35374 verify all CFA notes, their order and content.
35376 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35379 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35380 HImode and SImode with zero extend to DImode to one insn.
35381 (bswap<mode>2_extenddi): Likewise.
35382 (bswapsi2_extenddi): Likewise.
35383 (bswaphi2_extendsi): Likewise.
35384 (bswaphi2): Combine bswap HImode and SImode into one insn.
35385 Separate memory insns from swapping register.
35386 (bswapsi2): Likewise.
35387 (bswap<mode>2): Likewise.
35388 (bswaphi2_internal): Delete, no longer used.
35389 (bswapsi2_internal): Likewise.
35390 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35391 store, and gpr<-gpr swap insns.
35392 (bswap<mode>2_store): Likewise.
35393 (bswaphi2_reg): Register only splitter, combine with the splitter.
35394 (bswaphi2 splitter): Likewise.
35395 (bswapsi2_reg): Likewise.
35396 (bswapsi2 splitter): Likewise.
35397 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35398 the insns into load, store, and register/register insns.
35399 (bswapdi2_ldbrx): Likewise.
35400 (bswapdi2_load): Likewise.
35401 (bswapdi2_store): Likewise.
35402 (bswapdi2_reg): Likewise.
35404 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35406 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35407 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35409 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35412 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35414 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35415 special handling for target option conflicts between dform
35416 options (-mpower9-dform, -mpower9-dform-vector,
35417 -mpower9-dform-scalar) and -mno-direct-move.
35419 2017-03-27 Richard Biener <rguenther@suse.de>
35421 PR tree-optimization/80181
35422 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35424 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35426 * config/arc/predicates.md (move_double_src_operand): Replace the
35427 call to move_double_src_operand with a call to address_operand.
35429 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35431 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35432 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35433 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35435 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35437 * config/arc/predicates.md (long_immediate_loadstore_operand):
35438 Consider scaled addresses cases.
35440 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35442 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35443 restored when in interrupt.
35444 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35445 doesn't have delay slot.
35447 2017-03-27 Richard Biener <rguenther@suse.de>
35450 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35451 inlined thunk clones.
35453 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35456 * asan.c (instrument_derefs): Copy over last operand from
35457 original COMPONENT_REF to the new COMPONENT_REF with
35458 DECL_BIT_FIELD_REPRESENTATIVE.
35459 * ubsan.c (instrument_object_size): Likewise.
35461 2017-03-27 Richard Biener <rguenther@suse.de>
35463 PR tree-optimization/80170
35464 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35465 sure DR/SCEV didnt fold in constants we do not see when looking
35466 at the reference base alignment.
35468 2017-03-27 Richard Biener <rguenther@suse.de>
35470 PR middle-end/80171
35471 * gimple-fold.c (fold_ctor_reference): Properly guard against
35472 NULL return value from canonicalize_constructor_val.
35474 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35477 * config/i386/i386.c (ix86_expand_builtin)
35478 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35479 flags reg setting and flags reg using instructions.
35480 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35481 clobbering instructions to zero extend op2.
35483 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35485 * doc/install.texi (Configuration) <--with-aix-soname>:
35486 Update link to AIX ld.
35488 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35490 PR rtl-optimization/80160
35491 PR rtl-optimization/80159
35492 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35493 reg_alternate_class into account.
35495 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35498 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35499 to consider in curr_insn_transform.
35501 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35503 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35504 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35505 and emit_mode_inner.
35507 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35509 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35510 argument to the overloaded builtin variants. Use the new flag to
35511 deprecate certain builtin variants.
35512 * config/s390/s390-builtin-types.def: Add new builtin types.
35513 * config/s390/s390-builtins.h: Support new flags field for
35514 overloaded builtins.
35515 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35516 (s390_macro_to_expand): Enable vector float data type.
35517 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35518 builtins by incrementing the __VEC__ version number.
35519 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35521 (s390_resolve_overloaded_builtin): Emit error messages depending
35522 on the builtin flags.
35523 * config/s390/s390.c (s390_expand_builtin): Support additional
35524 flags argument. Change error message to match the messages
35525 emitted in s390-c.c.
35526 * config/s390/s390.md: New UNSPEC_* constants.
35527 (op_type): Add new instruction types.
35528 * config/s390/vecintrin.h: Add new builtins and test data class
35530 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35531 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35532 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35533 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35534 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35535 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35537 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35538 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35539 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35540 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35542 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35543 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35544 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35545 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35547 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35548 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35549 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35550 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35551 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35552 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35553 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35555 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35556 ("vec_scatter_element<V_HW_4:mode>_DI")
35557 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35558 ("vec_fpint<mode>", "vflls")
35559 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35560 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35561 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35562 ("*vec_cmphe<mode>_cc"): ... these.
35564 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35565 mode constant instead of magic value.
35567 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35569 * config/s390/s390.c (s390_expand_vec_compare): Support other
35570 vector floating point modes than just V2DF.
35571 (s390_expand_vcond): Likewise.
35572 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35573 (s390_cannot_change_mode_class): Prevent mode changes between TF
35574 and V1TF in vector registers.
35575 * config/s390/s390.md (DF, SF): New mode attributes.
35576 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35577 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35578 SFmode support for VRs.
35579 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35581 (VFT, VF_HW): New mode iterators.
35582 (vw, sdx): New mode attributes.
35583 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35584 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35585 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35586 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35587 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35588 also the new vector floating point modes. Renaming to ...
35590 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35591 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35592 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35593 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35594 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35595 ("vec_unordered<mode>"): ... these.
35597 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35598 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35599 ("*vec_extendv2df"): New insn definitions.
35601 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35603 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35604 ("mulditi3_2", "*muldi3_sign"): New patterns.
35605 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35606 rename the pattern definition.
35608 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35610 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35612 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35614 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35616 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35617 instruction if possible.
35618 * config/s390/vector.md (vec_halfnumelts): New mode
35620 ("*vec_vllezlf<mode>"): New pattern.
35622 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35624 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35625 ("popcountv4si2", "popcountv2di2"): Rename to ...
35626 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35627 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35629 ("popcount<mode>2_vxe"): New pattern.
35631 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35633 * common/config/s390/s390-common.c (processor_flags_table): Add
35635 * config.gcc: Add arch12.
35636 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35637 Default to arch12 for unknown CPU model numbers.
35638 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35639 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35640 PROCESSOR_max sanity check.
35641 * config/s390/s390-opts.h (enum processor_type): Add
35643 * config/s390/s390.c (processor_table): Add arch12.
35644 (s390_expand_builtin): Add check for B_VXE flag.
35645 (s390_issue_rate): Add PROCESSOR_ARCH12.
35646 (s390_get_sched_attrmask): Likewise.
35647 (s390_get_unit_mask): Likewise.
35648 (s390_sched_score): Enable z13 scheduling for arch12.
35649 (s390_sched_reorder): Likewise.
35650 (s390_sched_variable_issue): Likewise.
35651 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35653 (s390_tune_attr): Use z13 scheduling also for arch12.
35654 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35655 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35656 (TARGET_VXE_P): New macros.
35657 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35658 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35659 * config/s390/s390.opt: Add arch12 as processor_type.
35661 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35663 * config/s390/s390.md
35664 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35665 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35666 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35668 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35669 Rename expanders to ...
35671 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35672 ("fixuns_truncdddi2_emu"): ... these.
35674 ("fixuns_trunc<mode>si2_emu"): New expander.
35676 ("*fixuns_truncdfdi2_z13"): Rename to ...
35677 ("*fixuns_truncdfdi2_vx"): ... this.
35679 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35681 * config/s390/2964.md: Remove the single element vector compare
35682 instructions which are no longer used.
35683 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35685 (s390_canonicalize_comparison): Remove handling of DFmode
35687 (s390_expand_vec_compare_scalar): Remove function.
35688 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35689 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35691 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35693 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35695 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35697 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35698 will anyway by matched by mov<mode>_64dfp.
35700 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35702 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35703 vlef/vstef. Add missing operand to vleif.
35705 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35707 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35708 pair for all vector types with 64 bit elements.
35709 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35710 * config/s390/vector.md (V_HW_64): ... here.
35711 (V_128_NOSINGLE): New mode iterator.
35712 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35713 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35714 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35715 ("*vec_load_pairv2di"): Change to ...
35716 ("*vec_load_pair<mode>"): ... this one.
35718 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35720 * config/s390/constraints.md: Add comments.
35721 (jKK): Reject element sizes > 8 bytes.
35722 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35724 * config/s390/s390.md: Add the s_operand checks formerly in
35725 s390_split_ok_p to various splitters where they are still
35727 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35728 for 128 bit vectors. Plus two splitters.
35730 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35732 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35735 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35738 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35739 error if the boundary argument is not constant.
35741 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35743 PR rtl-optimization/80112
35744 * loop-doloop.c (doloop_condition_get): Don't check condition
35745 if cmp isn't SET with IF_THEN_ELSE src.
35747 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35749 PR tree-optimization/80158
35750 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35751 replacing a candidate statement, also replace it for the
35752 candidate's alternate interpretation.
35753 (replace_rhs_if_not_dup): Likewise.
35754 (replace_one_candidate): Likewise.
35756 2017-03-24 Richard Biener <rguenther@suse.de>
35758 PR tree-optimization/80167
35759 * graphite-isl-ast-to-gimple.c
35760 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35762 (translate_isl_ast_to_gimple::get_rename): Likewise.
35764 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35766 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35767 handling of certain combinations of target options, including the
35768 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35769 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35771 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35774 * config/arm/arm.md (*load_multiple): Add reload_completed to
35775 matching condition.
35777 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35778 Richard Biener <rguenth@suse.de>
35780 PR tree-optimization/79908
35781 PR tree-optimization/80136
35782 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35783 been cast away, gimplify_and_add suffices.
35785 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35787 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35789 2017-03-23 Richard Biener <rguenther@suse.de>
35791 PR tree-optimization/80032
35792 * gimplify.c (gimple_push_cleanup): Forced unconditional
35793 cleanups still have to go to the conditional_cleanups
35796 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35798 PR tree-optimization/80072
35799 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35801 (next_operand_entry_id): Change type to unsigned int.
35802 (sort_by_operand_rank): Make sure to return the right return value
35803 even if unsigned fields are bigger than INT_MAX.
35804 (struct oecount): Change cnt and id type to unsigned int.
35805 (oecount_hasher::equal): Formatting fix.
35806 (oecount_cmp): Make sure to return the right return value
35807 even if unsigned fields are bigger than INT_MAX.
35808 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35811 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35812 TREE_READONLY on result if writing it more than once.
35815 * doc/invoke.texi (-fsanitize=thread): Document that with
35816 -fnon-call-exceptions atomics are not able to throw
35820 * tsan.c: Include tree-eh.h.
35821 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35822 maybe_clean_or_replace_eh_stmt where needed.
35823 (instrument_memory_accesses): Add cfg_changed argument.
35824 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35825 if it returned true.
35826 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35828 PR rtl-optimization/63191
35829 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35830 wrapper function, moved the whole old content into ...
35831 (ix86_delegitimize_address_1): ... this. New inline function.
35832 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35833 true as last argument instead of ix86_delegitimize_address.
35835 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35837 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35838 cortexa57_branch_cost.
35840 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35842 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35844 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35847 * doc/md.texi (Constraints): Document wA constraint.
35848 * config/rs6000/constraints.md (wA): New.
35849 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35850 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35851 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35852 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35854 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35857 * gimplify.c (is_oacc_declared): New function.
35858 (oacc_default_clause): Use it to set default flags for acc declared
35859 variables inside parallel regions.
35860 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35861 declared variables.
35862 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35863 declare attribute to any decl as necessary.
35865 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35868 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35869 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35870 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35871 (arm_arch_lpae): This.
35872 * config/arm/arm.c (arm_arch7ve): Rename into ...
35873 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35874 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35877 2017-03-22 Martin Liska <mliska@suse.cz>
35880 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35881 error message instead of an ICE.
35883 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35885 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35887 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35889 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35891 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35894 2017-03-21 Martin Sebor <msebor@redhat.com>
35896 * doc/extend.texi: Use "cannot" instead of "can't."
35897 * doc/hostconfig.texi: Same.
35898 * doc/install.texi: Same.
35899 * doc/invoke.texi: Same.
35900 * doc/loop.texi: Same.
35901 * doc/md.texi: Same.
35902 * doc/objc.texi: Same.
35903 * doc/rtl.texi: Same.
35904 * doc/tm.texi: Same.
35905 * doc/tm.texi.in: Same.
35906 * doc/trouble.texi: Same.
35908 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35911 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35912 (collect_checksum_attributes): Set it.
35913 (die_checksum_ordered): Use it.
35915 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35917 PR tree-optimization/79908
35918 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35919 change: For a VA_ARG whose LHS has been cast away, use
35920 force_gimple_operand to construct the side effects.
35922 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35924 PR translation/80001
35925 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35926 more amenable to translation.
35927 (oacc_loop_auto_partitions): Likewise.
35929 2017-03-21 Marek Polacek <polacek@redhat.com>
35930 Martin Sebor <msebor@redhat.com>
35932 PR tree-optimization/80109
35933 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35934 on INTEGRAL_TYPE_P.
35936 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35937 Segher Boessenkool <segher@kernel.crashing.org>
35940 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35941 check reg_used_between_p between insn and one of succ or succ2
35942 depending on if succ is artificial insn not inserted into insn
35945 2017-03-21 Martin Liska <mliska@suse.cz>
35947 PR gcov-profile/80081
35948 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35949 * doc/gcc.texi: Include gcov-dump stuff.
35950 * doc/gcov-dump.texi: New file.
35952 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35954 PR rtl-optimization/79150
35955 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35956 conditional jump, if the jump is the last insn of the loop.
35958 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35959 Richard Biener <rguenth@suse.de>
35961 PR tree-optimization/79908
35962 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35963 been cast away, use force_gimple_operand to construct the side
35966 2017-03-21 Martin Liska <mliska@suse.cz>
35968 PR libfortran/79956
35969 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35972 2017-03-21 Brad Spengler <spender@grsecurity.net>
35975 * plugin.c (htab_hash_plugin): New function.
35976 (add_new_plugin): Use it and adjust.
35977 (parse_plugin_arg_opt): Adjust.
35978 (init_one_plugin): Likewise.
35980 2017-03-21 Richard Biener <rguenther@suse.de>
35982 PR tree-optimization/80032
35983 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35984 if set force the cleanup to happen unconditionally.
35985 (gimplify_target_expr): Push inserted clobbers with force_uncond
35986 to avoid them being removed by control-dependent DCE.
35988 2017-03-21 Richard Biener <rguenther@suse.de>
35990 PR tree-optimization/80122
35991 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35992 va_arg_pack_len when the inlined call stmt requires pack
35994 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35996 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35999 * tsan.c (instrument_builtin_call): If the memory model argument
36000 is not a constant, assume it is valid.
36003 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36006 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36008 PR rtl-optimization/79910
36009 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36010 if its dest is used by an insn before I2 (other than the combined
36011 insns themselves, which are properly handled already).
36013 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36016 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36018 * combine.c (record_used_regs): New static function.
36019 (try_combine): Handle situations where there is an additional
36020 instruction between I2 and I3 which needs to have a LOG_LINK
36024 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36026 * combine.c (try_combine): Delete redundant i1 test. Call
36027 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36029 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36032 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36033 alternatives 13/14.
36035 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36037 PR tree-optimization/80054
36038 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36039 the optimization if a PHI or any of its arguments is not dominated
36040 by the candidate's basis. Use gphi* rather than gimple* as
36042 (replace_profitable_candidates): Clean up a gimple* variable that
36043 should be a gphi* variable.
36045 2017-03-20 Martin Sebor <msebor@redhat.com>
36048 * doc/extend.texi (attribute constructor): Document present limitation.
36050 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36053 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36054 __POWER9_VECTOR__ #ifdef control, change template definition to
36055 use Power9-specific built-in function.
36056 (vec_any_eq): Likewise.
36057 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36058 to control outcomes from this test.
36059 (vector_ae_<mode>p): For VEC_F modes, likewise.
36061 2017-03-20 Ian Lance Taylor <iant@google.com>
36063 * config/i386/i386.c (ix86_function_regparm): Save an extra
36064 register for -fsplit-stack with DECL_STATIC_CHAIN.
36066 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36069 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36070 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36072 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36074 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36076 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36079 2017-03-20 Marek Polacek <polacek@redhat.com>
36082 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36084 2017-03-20 Richard Biener <rguenther@suse.de>
36086 PR tree-optimization/80113
36087 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36088 allocate extra SSA name for PHI def.
36089 (add_close_phis_to_outer_loops): Likewise.
36090 (add_close_phis_to_merge_points): Likewise.
36091 (copy_loop_close_phi_args): Likewise.
36092 (copy_cond_phi_nodes): Likewise.
36094 2017-03-20 Martin Liska <mliska@suse.cz>
36096 PR middle-end/79753
36097 * tree-chkp.c (chkp_build_returned_bound): Do not build
36098 returned bounds for a LHS that's not a BOUNDED_P type.
36100 2017-03-20 Martin Liska <mliska@suse.cz>
36104 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36105 COMPLEX_CST and VECTOR_CST.
36107 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36110 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36111 target operand. A new splitter adds the clobber statement in case
36112 the target operand is dead anyway.
36114 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36116 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36117 to age-old versions of binutils and glibc.
36119 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36121 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36123 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36125 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36127 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36129 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36130 requirement for binutils 2.13.
36132 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36134 * combine.c (try_combine): Delete redundant i1 test. Call
36135 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36137 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36139 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36140 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36142 <riscv64-*-elf>: Re-arrange section
36143 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36144 <riscv32-*-linux>: Likewise.
36145 <riscv64-*-elf>: Likewise
36146 <riscv64-*-linux>: Likewise.
36148 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36151 * aarch64.opt(verbose-cost-dump): Fix typo.
36153 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36156 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36157 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36159 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36161 * reload.c (find_reloads): When reloading a nonoffsettable address,
36162 use RELOAD_OTHER for it and its address reloads.
36164 PR rtl-optimization/79910
36165 * combine.c (record_used_regs): New static function.
36166 (try_combine): Handle situations where there is an additional
36167 instruction between I2 and I3 which needs to have a LOG_LINK
36170 2017-03-17 Jeff Law <law@redhat.com>
36172 PR tree-optimization/71437
36173 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36174 conditional in the hash table first.
36175 (vrp_dom_walker::before_dom_children): Extract condition from
36176 ASSERT_EXPR. Record condition, its inverion and any implied
36177 conditions as well.
36179 2017-03-17 Marek Polacek <polacek@redhat.com>
36180 Markus Trippelsdorf <markus@trippelsdorf.de>
36182 PR tree-optimization/80079
36183 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36186 2017-03-17 Richard Biener <rguenther@suse.de>
36188 PR middle-end/80075
36189 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36190 Properly verify the LHS before the RHS possibly claims to be
36192 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36195 2017-03-17 Martin Jambor <mjambor@suse.cz>
36197 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36198 (List of -O2 options): Likewise.
36199 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36202 2017-03-17 Tom de Vries <tom@codesourcery.com>
36204 * gcov-dump.c (print_usage): Print bug_report_url.
36206 2017-03-17 Richard Biener <rguenther@suse.de>
36208 PR middle-end/80050
36209 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36210 (parser::peek): Likewise.
36212 2017-03-17 Richard Biener <rguenther@suse.de>
36214 PR tree-optimization/80048
36215 * sese.c (free_sese_info): Properly release rename_map and
36216 copied_bb_map elements.
36218 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36220 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36221 Add linked-list forward and backlinks. Insert on
36222 construction, remove on destruction.
36223 (class pass_store_merging): Add m_stores_head field.
36224 (pass_store_merging::terminate_and_process_all_chains):
36225 Iterate over m_stores_head list.
36226 (pass_store_merging::terminate_all_aliasing_chains):
36228 (pass_store_merging::execute): Check for debug stmts first.
36229 Push new chains onto the m_stores_head stack.
36231 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36234 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36235 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36236 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36238 2017-03-16 Jeff Law <law@redhat.com>
36240 PR tree-optimization/71437
36241 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36242 member function. Implementation moved into after_dom_children
36243 member function and into the threader's thread_outgoing_edges
36245 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36246 some code into new thread_outgoing_edges.
36247 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36248 definition. Simplify marker handling (do it here). Assume we always
36249 have the available expression and the const/copies tables.
36250 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36252 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36253 * tree-vrp.c (equiv_stack): No longer file scoped.
36254 (vrp_dom_walker): New class.
36255 (vrp_dom_walker::before_dom_children): New member function.
36256 (vrp_dom_walker::after_dom_children): Likewise.
36257 (identify_jump_threads): Setup domwalker. Use it rather than
36258 walking edges in a random order by hand. Simplify setup/finalization.
36259 (finalize_jump_threads): Remove.
36260 (vrp_finalize): Do not call identify_jump_threads here.
36261 (execute_vrp): Do it here instead and call thread_through_all_blocks
36264 PR tree-optimization/71437
36265 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36267 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36269 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36270 (dom_opt_dom_walker::thread_across_edge): Remove
36271 handle_dominating_asserts argument. All callers changed.
36272 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36273 changes. Remove calls to lhs_of_dominating_assert. Other
36274 uses of handle_dominating_asserts turn into unconditional code
36275 (simplify_control_stmt_condition_1): Likewise.
36276 (simplify_control_stmt_condition): Likewise.
36277 (thread_through_normal_block, thread_across_edge): Likewise.
36278 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36279 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36280 object if it is not an SSA_NAME.
36281 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36282 before calling into the VRP specific simplifiers.
36283 (identify_jump_threads): Remove handle_dominating_asserts
36286 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36289 * tree-diagnostic.c (default_tree_printer): No longer static.
36290 * tree-diagnostic.h (default_tree_printer): New prototype.
36292 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36294 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36295 Change ins into fmov.
36297 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36299 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36300 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36301 Use h_con constraint for operand 1.
36302 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36303 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36305 2017-03-15 Jeff Law <law@redhat.com>
36307 PR tree-optimization/71437
36308 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36309 (record_temporary_equivalences): Use it.
36311 PR tree-optimization/71437
36312 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36313 tree-ssa-scopedtables.
36314 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36315 (record_conditions, record_cond, vuse_eq): Likewise.
36316 (record_edge_info): Adjust to API tweak of record_conditions.
36317 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36318 (record_temporary_equivalences, optimize_stmt): Likewise.
36319 (eliminate_redundant_computations): Likewise.
36320 (record_equivalences_from_stmt): Likewise.
36321 * tree-ssa-scopedtables.c: Include options.h and params.h.
36322 (vuse_eq): New function, moved from tree-ssa-dom.c
36323 (build_and_record_new_cond): Likewise.
36324 (record_conditions): Likewise. Accept vector of conditions rather
36325 than edge_equivalence structure for first argument.
36326 for the first argument.
36327 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36328 from tree-ssa-dom.c.
36329 (avail_exprs_stack::record_cond): Likewise.
36330 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36331 from tree-ssa-dom.c.
36332 (avail_exprs_stack): Add new member functions lookup_avail_expr
36334 (record_conditions): Declare.
36336 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36339 * lra-constraints.c (process_alt_operands): Increase reject for
36340 reloading an input/output operand.
36342 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36345 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36346 insns to convert from signed/unsigned char/short to IEEE 128-bit
36348 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36350 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36353 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36354 subreg of inner mode for values already in registers.
36356 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36358 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36359 iteration reg is used after the loop.
36361 2017-03-14 Martin Sebor <msebor@redhat.com>
36363 PR tree-optimization/79800
36364 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36365 precision in negative-positive range.
36366 (format_floating): Call non-const overload with adjusted precision.
36368 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36371 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36374 2017-03-14 Martin Sebor <msebor@redhat.com>
36376 PR middle-end/80020
36377 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36378 * builtins.def (aligned_alloc): Use it.
36381 * Makefile.in (GTFILES): Add calls.c.
36382 * calls.c: Include "gt-calls.h".
36384 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36386 PR rtl-optimization/79728
36387 * regs.h (struct target_regs): New field
36388 x_contains_allocatable_regs_of_mode.
36389 (contains_allocatable_regs_of_mode): New macro.
36390 * reginfo.c (init_reg_sets_1): Initialize it, and change
36391 contains_reg_of_mode so it includes global regs as well.
36392 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36393 rather than contains_regs_of_mode.
36395 2017-03-14 Martin Liska <mliska@suse.cz>
36397 * doc/invoke.texi: Document options that can't be combined with
36398 -fcheck-pointer-bounds.
36400 2017-03-14 Martin Liska <mliska@suse.cz>
36402 PR middle-end/79831
36403 * doc/invoke.texi (-Wchkp): Document the option.
36405 2017-03-14 Martin Liska <mliska@suse.cz>
36407 * Makefile.in: Install gcov-dump.
36409 2017-03-14 Martin Liska <mliska@suse.cz>
36411 * multiple_target.c (expand_target_clones): Bail out for
36412 an invalid attribute.
36414 2017-03-14 Richard Biener <rguenther@suse.de>
36416 * alias.c (struct alias_set_entry): Pack properly.
36417 * cfgloop.h (struct loop): Likewise.
36418 * cse.c (struct set): Likewise.
36419 * ipa-utils.c (struct searchc_env): Likewise.
36420 * loop-invariant.c (struct invariant): Likewise.
36421 * lra-remat.c (struct cand): Likewise.
36422 * recog.c (struct change_t): Likewise.
36423 * rtl.h (struct address_info): Likewise.
36424 * symbol-summary.h (function_summary): Likewise.
36425 * tree-loop-distribution.c (struct partition): Likewise.
36426 * tree-object-size.c (struct object_size_info): Likewise.
36427 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36428 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36429 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36430 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36431 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36432 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36433 (struct _stmt_vec_info): Likewise.
36435 2017-03-14 Martin Liska <mliska@suse.cz>
36438 * multiple_target.c (create_dispatcher_calls): Check that
36439 a target can create a function dispatcher.
36441 2017-03-14 Martin Liska <mliska@suse.cz>
36444 * multiple_target.c (expand_target_clones): Drop local.local
36445 flag for default implementation.
36447 2017-03-14 Richard Biener <rguenther@suse.de>
36449 PR tree-optimization/80030
36450 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36452 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36454 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36455 gcc_fallthrough() instead of __attribute__((fallthrough));
36457 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36459 * doc/gcc.texi: Remove "up" link to (DIR).
36460 * doc/gccint.texi: Ditto.
36462 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36464 * doc/install.texi (Specific) <avr>: Remove reference to
36467 2017-03-13 Jeff Law <law@redhat.com>
36469 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36470 attribute rather than comments.
36472 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36473 match_scratch operand is highest.
36475 2017-03-13 Martin Liska <mliska@suse.cz>
36477 PR middle-end/78339
36478 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36479 is a CHKP clone, use original declaration.
36481 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36483 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36484 (arc_conditional_register_usage): Use a different allocation order
36485 when optimizing for size.
36486 * common/config/arc/arc-common.c (arc_option_optimization_table):
36487 Section anchors default on when optimizing for size.
36489 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36491 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36493 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36495 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36496 * config/arc/arc.md (cpu_facility): Add cd variant.
36497 (*movqi_insn): Add code density variant.
36498 (*movhi_insn): Likewise.
36499 (*movqi_insn): Likewise.
36500 (*addsi3_mixed): Likewise.
36501 (subsi3_insn): Likewise.
36503 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36505 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36507 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36509 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36510 expressions with MINUS and UNARY ops.
36512 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36515 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36517 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36518 between vec_select and vector argument.
36519 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36520 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36521 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36522 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36523 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36524 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36526 2017-03-13 Richard Biener <rguenther@suse.de>
36529 * params.def (vect-max-peeling-for-alignment): Fix typo.
36531 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36533 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36534 issue that only occurred with binutils below 2.18.
36536 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36538 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36539 refer to binutils 2.11/2.12 minimum.
36541 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36543 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36544 ftp.kernel.org and simplify binutils requirement.
36546 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36548 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36550 (Optimize Options): Ditto. Also remove redundancy.
36552 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36554 PR translation/79848
36555 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36557 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36558 to G_ to avoid double translation.
36560 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36562 PR translation/79923
36563 * auto-profile.c (get_combined_location): Convert leading
36564 character of diagnostics to lower case and remove trailing period.
36565 (read_profile): Likewise for various diagnostics.
36566 * config/arm/arm.c (arm_option_override): Remove trailing period
36567 from various diagnostics.
36568 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36569 (msp430_expand_delay_cycles): Likewise.
36571 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36574 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36575 full command-line argument, rather than just "str".
36576 (aarch64_validate_march): Likewise.
36577 (aarch64_validate_mtune): Likewise.
36579 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36581 PR rtl-optimization/78911
36582 * lra-assigns.c (must_not_spill_p): New function.
36583 (spill_for): Use it.
36585 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36587 PR tree-optimization/79981
36588 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36589 ATOMIC_COMPARE_EXCHANGE ifn result.
36590 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36591 IFN_ATOMIC_COMPARE_EXCHANGE.
36593 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36596 * opts.c (parse_sanitizer_options): Add missing question mark to
36597 "did you mean" message.
36599 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36601 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36603 (VMULEUH_UNS): Likewise.
36604 (VMULOUB_UNS): Likewise.
36605 (VMULOUH_UNS): Likewise.
36606 * config/rs6000/rs6000.c (builtin_function_type): Remove
36607 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36609 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36612 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36613 x with result of extra_parsing_for_operand_code_0.
36614 (function_reader::extra_parsing_for_operand_code_0): Convert
36615 return type from void to rtx, returning x. When reading
36616 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36617 larger size containing struct block_symbol.
36619 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36621 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36622 -mfloat128-hardware without -m64.
36624 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36627 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36628 entries to the case statement that marks unsigned arguments to
36629 overloaded functions.
36631 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36633 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36634 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36636 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36639 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36640 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36642 2017-03-10 Martin Liska <mliska@suse.cz>
36646 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36647 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36650 2017-03-10 Olivier Hainque <hainque@adacore.com>
36652 * tree-switch-conversion (array_value_type): Start by resetting
36653 candidate type to it's main variant.
36655 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36657 PR rtl-optimization/79909
36658 * combine.c (try_combine): Use simplify_replace_rtx on individual
36659 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36660 of the whole CALL_INSN_FUNCTION_USAGE.
36662 PR tree-optimization/79972
36663 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36664 get_range_info on SSA_NAMEs. Formatting fixes.
36666 2017-03-10 Richard Biener <rguenther@suse.de>
36667 Jakub Jelinek <jakub@redhat.com>
36669 PR tree-optimization/77975
36670 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36671 edge to be constant.
36672 (get_val_for): For constant x return it. Formatting fix.
36673 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36674 would use the same bases as the current one.
36676 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36678 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36679 instead of vec_select for V1TImode.
36680 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36682 (VSX_LE_128): Add V1TI to this mode iterator.
36683 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36684 (*vsx_le_perm_store_<mode>): Likewise.
36685 (pre-reload splitter for VSX stores): Likewise.
36686 (post-reload splitter for VSX stores): Likewise.
36687 (*vsx_xxpermdi2_le_<mode>): Likewise.
36688 (*vsx_lxvd2x2_le_<mode>): Likewise.
36689 (*vsx_stxvd2x2_le_<mode>): Likewise.
36691 2017-03-09 Michael Eager <eager@eagercon.com>
36693 Correct failures with --enable-checking=yes,rtl.
36695 * config/microblaze/microblaze.c (microblaze_expand_shift):
36696 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36697 test for const0_rtx.
36698 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36699 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36701 2017-03-09 Richard Biener <rguenther@suse.de>
36703 PR tree-optimization/79977
36704 * graphite-scop-detection.c (scop_detection::merge_sese):
36705 Handle the case of extra exits to blocks dominating the entry.
36707 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36709 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36712 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36714 PR rtl-optimization/79949
36715 * lra-constraints.c (process_alt_operands): Check memory when
36716 trying to predict a cycle. Print about the overall increase.
36718 2017-03-09 Richard Biener <rguenther@suse.de>
36720 PR middle-end/79971
36721 * gimple-expr.c (useless_type_conversion_p): Preserve
36722 TYPE_SATURATING for fixed-point types.
36724 2017-03-09 Richard Biener <rguenther@suse.de>
36727 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36728 alignment of BLKmode params.
36730 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36733 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36734 (VALL_NO_V2Q): Likewise.
36736 * config/aarch64/aarch64-simd.md
36737 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36739 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36740 VALL_NO_V2Q mode iterator.
36741 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36743 2017-03-09 Martin Liska <mliska@suse.cz>
36745 PR tree-optimization/79631
36746 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36747 tree_int_cst_sign_bit just for INTEGER constants.
36749 2017-03-09 Martin Liska <mliska@suse.cz>
36753 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36756 2017-03-09 Marek Polacek <polacek@redhat.com>
36759 * tree.c (inchash::add_expr): Handle TREE_VEC.
36761 2017-03-09 Martin Liska <mliska@suse.cz>
36764 (chkp_narrow_size_and_offset): New function.
36765 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36766 (void chkp_parse_bit_field_ref): New function.
36767 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36768 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36770 2017-03-09 Martin Liska <mliska@suse.cz>
36773 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36774 (chkp_find_bounds_1): Remove gcc_unreachable.
36776 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36779 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36780 BUILT_IN_SYNC*, determine the access type from the size suffix and
36781 always build a MEM_REF with that type. Handle forgotten
36782 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36785 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36786 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36787 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36788 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36789 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36790 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36791 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36792 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36793 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36794 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36795 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36796 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36797 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36798 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36799 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36800 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36801 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36802 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36803 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36804 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36805 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36806 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36807 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36808 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36809 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36810 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36811 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36812 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36813 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36814 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36815 definitions outside of __OPTIMIZE__ guarded section.
36818 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36819 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36820 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36821 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36824 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36826 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36827 ("vfenez<mode>"): Add missing constraints.
36829 2017-03-08 Martin Sebor <msebor@redhat.com>
36832 * config/nds32/nds32.c (nds32_option_override):
36833 Fix misspelled diagnostic.
36835 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36838 * gimplify.c (gimplify_omp_for): Replace index var in outer
36839 taskloop statement with an artificial variable and add
36840 OMP_CLAUSE_PRIVATE clause for it.
36842 2017-03-08 Richard Biener <rguenther@suse.de>
36844 PR tree-optimization/79955
36845 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36846 for accesses that are completely outside of the variable.
36848 2017-03-08 Andrew Haley <aph@redhat.com>
36850 PR tree-optimization/79943
36851 * tree-ssa-loop-split.c (compute_new_first_bound): When
36852 calculating the new upper bound, (END-BEG) should be added, not
36855 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36857 * config/avr/avr.md (setmemhi): Make sure match_dup
36858 operand number comes before match_scratch.
36860 2017-03-08 Richard Biener <rguenther@suse.de>
36862 PR tree-optimization/79920
36863 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36864 with ncopies == 1 to ...
36865 (vect_transform_slp_perm_load): ... here. Properly compute
36866 all element loads by iterating VF times over the group. Do
36867 not handle ncopies (computed in a broken way) in
36868 vect_create_mask_and_perm.
36870 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36873 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36874 is a uniform vector, use uniform_vector_p return value instead of
36875 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36877 2017-03-07 Marek Polacek <polacek@redhat.com>
36879 PR middle-end/79809
36880 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36881 (alloca_call_type): Likewise.
36883 2017-03-07 Martin Liska <mliska@suse.cz>
36885 * gcov.c (process_args): Put comment to correct location.
36887 2017-03-07 Martin Liska <mliska@suse.cz>
36889 PR middle-end/68270
36890 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36891 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36892 (chkp_narrow_bounds_for_field): Likewise.
36893 (chkp_parse_array_and_component_ref): Pass one more argument to
36896 2017-03-07 Richard Biener <rguenther@suse.de>
36898 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36901 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36903 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36904 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36906 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36909 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36910 to end of description.
36911 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36913 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36915 PR rtl-optimization/79901
36916 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36918 (*avx512f_<code><mode>3<mask_name>): ... this.
36919 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36920 iterator instead of VI8_AVX2_AVX512BW.
36922 PR rtl-optimization/79901
36923 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36924 min/max expander, expand it using expand_vec_cond_expr.
36927 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36930 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36933 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36934 to void * for PCH reasons.
36935 * dwarf2out.c (output_loc_operands, output_die): Cast
36936 v.val_vec.array to unsigned char *.
36938 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36941 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36944 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36946 PR rtl-optimization/79571
36947 * lra-constraints.c (process_alt_operands): Calculate static
36948 reject and subtract it from overall when only addresses will be
36951 2017-03-06 Julia Koval <julia.koval@intel.com>
36954 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36955 incoming stack boundary to 128 for 64-bit targets.
36957 2017-03-06 Richard Biener <rguenther@suse.de>
36959 PR tree-optimization/79894
36960 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36961 to NULL after folding it.
36963 2017-03-06 Richard Biener <rguenther@suse.de>
36965 PR tree-optimization/79824
36966 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36967 check disabling peeling for gaps.
36969 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36971 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36972 attributes): Document gettimeofday.
36974 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36976 * config/s390/s390.c (s390_option_override_internal): Set
36977 PARAM_MIN_VECT_LOOP_BOUND
36979 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36981 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36982 * config/s390/s390.md: Likewise.
36984 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36987 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36988 (<avx2_avx512>_perm<mode>): Rename to ...
36989 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36991 (<avx512>_perm<mode>_mask): Rename to ...
36992 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36994 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36995 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36996 instead of VI8F_256_512.
36997 (avx512f_perm<mode>): New define_expand.
36998 (avx512f_perm<mode>_mask): Likewise.
36999 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37000 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37002 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37004 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37005 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37007 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37009 2017-03-06 Martin Liska <mliska@suse.cz>
37012 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37013 when having a SSA NAME w/o VAR_DECL assigned to it.
37015 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37017 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37018 msa_dpsub_<su>_d): Fix MODE for vec_select.
37020 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37022 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37024 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37026 2017-03-06 Richard Biener <rguenther@suse.de>
37028 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37029 * plugin.c (register_plugin_info): Likewise.
37030 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37032 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37034 * config/i386/sse.md (sse_storehps, sse_storelps,
37035 avx_<castmode><avxsizesuffix>_<castmode>,
37036 avx512f_<castmode><avxsizesuffix>_<castmode>,
37037 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37038 in condition that at least one operand is not a MEM.
37040 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37042 PR middle-end/79805
37043 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37044 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37046 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37047 gimple_call_nothrow_p flag based on whether original builtin can throw.
37048 If it can, emit following stmts on the fallthrough edge.
37049 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37050 don't create new bb if inserting just debug stmts on the edge, try to
37051 insert them on the fallthru bb or just reset debug stmts.
37053 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37056 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37057 restore recog_data (including the operand rtxes inside it) around
37058 the call to get_insn_template.
37060 2017-03-03 Martin Sebor <msebor@redhat.com>
37062 PR tree-optimization/79699
37063 * context.c (context::~context): Free MPFR caches to avoid
37064 a memory leak on program exit.
37066 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37068 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37069 Use wide_int::ulow () instead of .elt (0).
37071 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37073 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37074 (*pushxf): Limit oF constraint to 32bit targets and add oC
37075 constraint for 64bit targets.
37076 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37077 (*pushdf): Change rmF constraint to rmC.
37079 2017-03-03 Martin Liska <mliska@suse.cz>
37081 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37082 Remove unused variable.
37084 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37087 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37088 is a memory operand, increase num_memory.
37089 (ix86_expand_args_builtin): Likewise.
37091 2017-03-03 Jan Hubicka <jh@suse.cz>
37094 * ipa-devirt.c (maybe_record_node): Properly handle
37095 __cxa_pure_virtual visibility.
37097 2017-03-03 Martin Liska <mliska@suse.cz>
37099 PR tree-optimization/79803
37100 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37102 (pass_loop_prefetch::execute): Disabled optimization if an
37103 assumption about L1 cache size is not met.
37105 2017-03-03 Martin Liska <mliska@suse.cz>
37107 PR rtl-optimization/79574
37108 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37109 (hash_scan_set): Likewise.
37110 (dump_hash_table): Likewise.
37111 (hoist_code): Likewise.
37113 2017-03-03 Richard Biener <rguenther@suse.de>
37115 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37116 in place of uhigh ().
37117 (fixed_convert_from_real): Likewise.
37119 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37122 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37124 2017-03-03 Richard Biener <rguenther@suse.de>
37126 PR middle-end/79818
37127 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37128 TYPE_OVERFLOW_UNDEFINED check.
37130 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37132 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37134 (vector_ae_<mode>_p): Likewise.
37135 (vector_nez_<mode>_p): Likewise.
37136 (vector_ne_v2di_p): Likewise.
37137 (vector_ae_v2di_p): Likewise.
37138 (vector_ne_<mode>_p): Likewise.
37139 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37141 (vsx_tsqrt<mode>2_fe): Likewise.
37143 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37146 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37148 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37150 PR rtl-optimization/79780
37151 * cprop.c (one_cprop_pass): When second and further conditional trap
37152 in a single basic block is turned into an unconditional trap, turn it
37153 into a deleted note to avoid RTL verification failures.
37155 2017-03-02 Richard Biener <rguenther@suse.de>
37157 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37159 2017-03-02 Richard Biener <rguenther@suse.de>
37161 PR tree-optimization/79345
37163 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37164 param and abort the walk, returning -1 if it is hit.
37165 (walk_aliased_vdefs): Take a limit param and pass it on.
37166 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37167 defaulting to 0 and return a signed int.
37168 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37169 (check_defs): New helper.
37170 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37171 about uninitialized memory.
37172 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37173 bogus uninitialized warning.
37174 (fixed_convert_from_real): Likewise.
37176 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37178 PR tree-optimization/66768
37179 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37180 iv_use if base object can't be determined.
37182 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37184 PR tree-optimization/79345
37185 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37186 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37187 (get_pattern_stats): Initialize it.
37188 * genemit.c (gen_expand): Verify match_scratch numbers come after
37189 match_operand/match_dup numbers.
37190 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37191 match_scratch numbers.
37192 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37194 * config/s390/s390.md (trunctdsd2): Likewise.
37196 2017-03-02 Richard Biener <rguenther@suse.de>
37198 * wide-int.h (wide_int_storage::operator=): Implement in terms
37201 2017-03-02 Richard Biener <rguenther@suse.de>
37203 PR tree-optimization/79777
37204 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37205 the to insert expression to sth existing.
37207 2017-03-01 Martin Sebor <msebor@redhat.com>
37209 PR middle-end/79692
37210 * gimple-ssa-sprintf.c
37211 (directive::known_width_and_precision): New function.
37212 (format_integer): Use it.
37213 (get_mpfr_format_length): Consider the full range of precision
37214 when computing %g output with the # flag. Set the likely byte
37215 count to 3 rather than 1 when precision is indeterminate.
37216 (format_floating): Correct the lower bound of precision.
37218 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37220 * doc/invoke.texi: Document default code model for 64-bit Linux.
37222 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37225 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37226 udiv rather than div since input pattern is unsigned.
37228 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37230 * config/i386/i386.c (print_reg): Warn for values of
37231 unsupported size in integer register.
37233 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37236 * config/rs6000/predicates.md (current_file_function_operand): Do
37237 not allow self calls to be local if the function is replaceable.
37239 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37242 * config/rs6000/altivec.h (vec_ctz and others): Change the
37243 preprocessor macro that controls conditional compilation from
37244 _ARCH_PWR9 to __POWER9_VECTOR__.
37245 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37246 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37247 control (instead of _ARCH_PWR9 control) so that template
37248 definition uses power9-specific function.
37249 (vec_any_eq): Likewise.
37250 (vec_all_ne): Change macro definition to use a power9-specific
37251 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37252 _ARCH_PWR9 control).
37253 (vec_any_eq) Likewise.
37254 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37255 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37256 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37257 support for xvcmpnedp instruction.
37258 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37259 macro expansion so that Power9 implementation of vec_all_ne does
37260 not use the AltiVec predicate framework.
37261 (VCMPNEH_P): Likewise.
37262 (VCMPNEW_P): Likewise.
37263 (VCMPNED_P): Likewise.
37264 (VCMPNEFP_P): Likewise.
37265 (VCMPNEDP_P): Likewise.
37266 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37267 implementation of vec_any_eq to not use AltiVec predicate
37269 (VCMPAEH_P): Likewise.
37270 (VCMPAEW_P): Likewise.
37271 (VCMPAED_P): Likewise.
37272 (VCMPAEFP_P): Likewise.
37273 (VCMPAEDP_P): Likewise.
37274 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37275 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37276 not use the AltiVec predicate framework.
37277 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37278 of vec_any_eq to not use AltiVec predicate framework.
37279 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37280 support for predefined __POWER9_VECTOR__ macro to indicate that
37281 Power9 instruction selection is enabled.
37282 (altivec_overloaded_builtins): Remove extraneous
37283 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37284 function argument types RS6000_BTI_bool_V16QI and
37285 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37286 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37287 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37288 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37289 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37290 Power9 for implementations of vec_cmpne. Change the signature for
37291 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37292 (representing vec_all_ne) to remove the previously described first
37293 argument of type RS6000_BTI_INTSI, as this was an artifact of
37294 reliance on the AltiVec predicate framework, which is no longer
37295 used in the implementation of these functions. Add
37296 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37297 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37298 since, unlike the AltiVec predicate framework implementation, we
37299 do not share function descriptors between vec_alle and vec_anyeq.
37300 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37301 set of modes that receive special treatment even when
37302 TARGET_P9_VECTOR is true. The special treatment emits code that
37303 does not depend on Power9 instructions.
37304 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37305 define_expand to not rely on AltiVec predicate framework.
37306 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37308 (vector_ne_v2di_p): Change this define_expand to not rely on
37309 AltiVec predicate framework.
37310 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37312 (vector_ne_<mode>_p): Change this define_expand to not rely on
37313 AltiVec predicate framework.
37314 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37316 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37317 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37318 define_insn pattern.
37319 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37320 define_insn pattern because the xvcmpne<VSs>. instruction is not
37322 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37323 instruction is not supported.
37325 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37327 * config/nvptx/nvptx.c: Include intl.h.
37329 2017-03-01 Martin Jambor <mjambor@suse.cz>
37332 * ipa-prop.h (ipa_bits): Removed field known.
37333 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37334 to pointers. Adjusted their comments to warn about their sharing.
37335 (ipcp_transformation_summary): Change bits to a vector of pointers.
37336 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37337 (ipa_get_ipa_bits_for_value): Declare.
37338 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37339 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37340 (ipa_bits_hash_table): Likewise.
37341 (ipa_vr_ggc_hash_traits): Likewise.
37342 (ipa_vr_hash_table): Likewise.
37343 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37344 being pointers and vr_known being removed.
37345 (ipa_set_jf_unknown): Likewise.
37346 (ipa_get_ipa_bits_for_value): New function.
37347 (ipa_set_jfunc_bits): Likewise.
37348 (ipa_get_value_range): New overloaded functions.
37349 (ipa_set_jfunc_vr): Likewise.
37350 (ipa_compute_jump_functions_for_edge): Use the above functions to
37351 construct bits and vr parts of jump functions.
37352 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37353 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37355 (ipcp_grow_transformations_if_necessary): Also allocate
37356 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37358 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37359 them. Fix too long lines.
37360 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37361 vr_known being removed.
37362 (ipa_read_jump_function): Use new setter functions to construct bits
37363 and vr parts of jump functions or set them to NULL.
37364 (write_ipcp_transformation_info): Adjust for bits being pointers.
37365 (read_ipcp_transformation_info): Likewise.
37366 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37368 Include gt-ipa-prop.h.
37369 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37371 (ipcp_store_bits_results): Likewise.
37372 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37373 Do not write to existing jump functions but use a temporary instead.
37375 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37378 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37379 attempt to use its first operand as BIT_FIELD_REF base.
37381 2017-03-01 Richard Biener <rguenther@suse.de>
37383 PR middle-end/79721
37384 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37385 interpolating formula in wrapping arithmetic.
37386 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37388 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37390 PR tree-optimization/79734
37391 * tree-vect-generic.c (expand_vector_condition): Optimize
37392 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37393 Handle VEC_COND_EXPR where comparison has different inner width from
37394 type's inner width.
37396 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37398 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37399 markup, and similar issues. Remove @opindex entries for things
37400 that aren't options. Add missing -mmpy-option entries.
37402 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37404 PR tree-optimization/79737
37405 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37406 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37407 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37408 instead of byte_size. Formatting fix.
37409 (shift_bytes_in_array_right): Formatting fix.
37411 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37414 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37415 condition on optimize for the leaf function test.
37417 2017-02-28 Martin Liska <mliska@suse.cz>
37420 * read-rtl-function.c (function_reader::handle_unknown_directive):
37421 Bail out when one uses -flto.
37423 2017-02-28 Martin Liska <mliska@suse.cz>
37425 * common.opt: Replace space with tabular for options of <number>
37427 * config/i386/i386.opt: Show <number> value for
37428 -mlarge-data-threshold.
37429 * opts.c (print_filtered_help): Do not display number in hexadecimal
37432 2017-02-28 Martin Liska <mliska@suse.cz>
37434 * common.opt: Fix --help=option -Q for options which are of
37437 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37439 * config/i386/i386.c (print_reg): Error out for values
37440 of 8-bit size in invalid integer register.
37442 2017-02-28 Martin Sebor <msebor@redhat.com>
37444 PR tree-optimization/79691
37445 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37447 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37450 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37451 gcc_unreachable with output_operand_lossage.
37453 2017-02-28 Richard Biener <rguenther@suse.de>
37455 PR tree-optimization/79740
37456 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37458 (visit_nary_op): Insert the nary into the hashtable if we
37459 pattern-matched sth.
37460 * tree-ssa-pre.c (eliminate_insert): Robustify.
37462 2017-02-28 Richard Biener <rguenther@suse.de>
37464 PR middle-end/79731
37465 * fold-const.c (decode_field_reference): Reject out-of-bound
37468 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37470 * config/i386/i386.c: Include intl.h.
37471 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37472 instead of just cond ? "..." : "...".
37473 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37474 * coverage.c (read_counts_file): Likewise.
37475 * omp-offload.c: Include intl.h.
37476 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37477 of just cond ? "..." : "...".
37478 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37479 of just cond ? "..." : "...".
37481 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37484 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37486 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37487 'tune for' CPU name.
37488 * config/arm/arm-cpu-data.h: Regenerated.
37490 2017-02-28 Richard Biener <rguenther@suse.de>
37492 PR tree-optimization/79732
37493 * tree-inline.c (expand_call_inline): Do not shadow var.
37495 2017-02-28 Richard Biener <rguenther@suse.de>
37497 PR tree-optimization/79723
37498 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37499 address-space properly.
37501 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37503 * doc/optinfo.texi (Optimization groups): Fix option used for
37505 * doc/invoke.texi (-fopt-info): Document "omp".
37506 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37507 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37508 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37509 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37510 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37512 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37514 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37515 OPTGROUP_OMP with "omp".
37517 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37520 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37521 for arithmetic shift of unsigned V2DI.
37523 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37525 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37526 arc/linux.h headers.
37527 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37528 (LINK_SPEC): Likewise.
37529 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37530 (EXTRA_SPECS): Likewise.
37531 (STARTFILE_SPEC): Likewise.
37532 (ENDFILE_SPEC): Likewise.
37533 (LIB_SPEC): Likewise.
37534 (TARGET_SDATA_DEFAULT): Likewise.
37535 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37536 (MULTILIB_DEFAULTS): Likewise.
37537 (DWARF2_UNWIND_INFO): Likewise.
37538 * config/arc/big.h: New file.
37539 * config/arc/elf.h: Likewise.
37540 * config/arc/linux.h: Likewise.
37541 * config/arc/t-uClibc: Remove.
37543 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37545 PR tree-optimization/77536
37546 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37547 (tree_transform_and_unroll_loop): Use above function to compute the
37548 estimated niter of unrolled loop and use it when scaling profile.
37549 Also use count info rather than frequency if it's non-zero.
37550 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37551 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37552 (vect_transform_loop): Call above function.
37554 2017-02-27 Richard Biener <rguenther@suse.de>
37556 PR tree-optimization/45397
37557 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37558 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37559 (visit_nary_op): Add pattern matching for CSEing sign-changed
37560 or truncated operations with wider ones.
37562 2017-02-27 Richard Biener <rguenther@suse.de>
37564 PR tree-optimization/79690
37565 * tree-vect-stmts.c (vectorizable_store): Use vector type
37566 built from the DR with address-space.
37568 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37570 * doc/invoke.texi (Optimize Options): Refine the description
37571 of asan-use-after-return.
37573 2017-02-25 Alan Modra <amodra@gmail.com>
37575 PR rtl-optimization/79584
37576 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37577 base, not ad->base_term, the reg within base. Remove assertion
37578 that ad->base == ad->base_term. Replace gen_int_mode using
37579 bogus mode with const0_rtx.
37581 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37583 PR middle-end/79396
37584 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37585 FMA_EXPR like tcc_binary or tcc_unary.
37587 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37590 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37592 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37593 (output_loc_operands): Handle DW_OP_call_ref and
37594 DW_OP_GNU_variable_value.
37595 (struct variable_value_struct): New type.
37596 (struct variable_value_hasher): Likewise.
37597 (variable_value_hash): New variable.
37598 (string_types): Remove.
37599 (copy_loc_descr): New function.
37600 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37601 (prepend_loc_descr_to_each): New function.
37602 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37603 instead of add_loc_descr_to_each if the first argument is single
37604 location list and the second has multiple.
37605 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37606 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37607 when looking for variable value which doesn't have other location info.
37608 (loc_list_from_tree): Formatting fix.
37609 (gen_array_type_die): Simplify DW_AT_string_length handling.
37610 (adjust_string_types): Remove.
37611 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37612 string_types. Call resolve_variable_values.
37613 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37614 (resolve_addr_in_expr): Likewise. Add A argument.
37615 (copy_deref_exprloc): Remove deref argument. Adjust for the
37616 original expression being DW_OP_GNU_variable_value with optionally
37617 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37618 optionally after it.
37619 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37620 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37621 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37622 (variable_value_hasher::hash, variable_value_hasher::equal): New
37624 (resolve_variable_value_in_expr, resolve_variable_value,
37625 resolve_variable_values, note_variable_value_in_expr,
37626 note_variable_value): New functions.
37627 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37630 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37633 * opts.h (handle_generated_option): Add GENERATED_P argument.
37634 * opts-common.c (handle_option): Adjust function comment.
37635 (handle_generated_option): Add GENERATED_P argument, pass it to
37637 (control_warning_option): Pass false to handle_generated_option
37639 * opts.c (maybe_default_option): Pass true to handle_generated_option
37641 * optc-gen.awk: Likewise.
37643 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37645 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37646 a REG, look at the REG it is a SUBREG of.
37647 (splitter for cmpeqsi_t): Ditto.
37649 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37651 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37652 the special USEs with the pattern of the insn, not the insn itself.
37654 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37657 * doc/invoke.texi: Document -mload-store-pairs.
37659 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37660 Sandra Loosemore <sandra@codesourcery.com>
37662 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37663 argument isn't a CONST_INT.
37664 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37665 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37666 (nios2_validate_compare): Bypass alternate compare logic if *op2
37667 is not a CONST_INT.
37668 (ldstwm_operation_p): Return false if first_base is not a REG or
37669 if first_offset is not a CONST_INT.
37671 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37673 * config/cris/cris.md: Use correct operand in a define_peephole2.
37675 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37677 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37679 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37681 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37682 this_insn if it is an INSN or JUMP_INSN.
37683 (force_offsettable): Look at base, not at addr.
37684 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37685 on things that aren't necessarily CONST_INTs.
37687 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37689 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37690 -mfpmath=sse is the default also for x86-32 targets with SSE2
37691 instruction set when @option{-ffast-math} is enabled
37693 2017-02-24 Jeff Law <law@redhat.com>
37695 PR rtl-optimizatoin/79286
37696 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37699 2017-02-24 Richard Biener <rguenther@suse.de>
37701 PR tree-optimization/79389
37702 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37705 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37707 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37708 function comment to reflect reality.
37709 (loop_exits_before_overflow): Fix typo in function description.
37711 2017-02-24 Richard Biener <rguenther@suse.de>
37713 PR tree-optimization/79389
37714 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37715 properly that a threading opportunity exists. Detect conditional
37716 copy/constant propagation opportunities.
37718 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37720 * config/visium/visium.md (type): Add trap.
37721 (b): New mode attribute.
37722 (*btst): Rename into...
37723 (*btst<mode>): ...this and adjust.
37724 (*cbranchsi4_btst_insn): Rename into...
37725 (*cbranch<mode>4_btst_insn): ...this and adjust.
37726 (trap): New define_insn.
37728 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37730 PR tree-optimization/79389
37731 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37732 (noce_reversed_cond_code): New function.
37733 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37734 reversed_comparison_code. Formatting fix.
37735 (noce_try_store_flag): Test rev_cond != NULL in addition to
37736 reversed_comparison_code.
37737 (noce_try_store_flag_constants): Likewise.
37738 (noce_try_store_flag_mask): Likewise.
37739 (noce_try_addcc): Use rev_cond if non-NULL instead of
37740 reversed_comparison_code.
37741 (noce_try_cmove_arith): Likewise. Formatting fixes.
37742 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37743 (noce_find_if_block): Initialize rev_cond.
37744 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37745 instead of false as last argument never attempt to reverse it
37748 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37750 PR tree-optimization/79663
37751 * tree-predcom.c (combine_chains): Process refs in reverse order
37752 only for ZERO length chains, and add explaining comment.
37754 2017-02-23 Jeff Law <law@redhat.com>
37756 PR tree-optimization/79578
37757 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37758 in call to operand_equal_p.
37760 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37763 * config/i386/cpuid.h: Fix another undefined behavior.
37765 2017-02-23 Richard Biener <rguenther@suse.de>
37767 PR tree-optimization/79683
37768 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37769 vector types for data-refs.
37771 2017-02-23 Martin Liska <mliska@suse.cz>
37773 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37775 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37777 PR middle-end/79665
37778 * internal-fn.c (get_range_pos_neg): Moved to ...
37779 * tree.c (get_range_pos_neg): ... here. No longer static.
37780 * tree.h (get_range_pos_neg): New prototype.
37781 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37782 are known to be in between 0 and signed maximum inclusive, try to
37783 expand both unsigned and signed divmod and use the cheaper one from
37786 2017-02-22 Jeff Law <law@redhat.com>
37788 PR tree-optimization/79578
37789 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37790 to compare base operands.
37792 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37795 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37796 gpc_reg_operand instead of fpr_reg_operand.
37798 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37800 * config/mips/mips.c (mips_return_in_memory): Force FP
37801 vector types to be returned in memory for o32 ABI.
37803 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37805 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37806 instead of DW_TAG_member for static data member declarations and don't
37807 set no_linkage_name for static inline data members.
37808 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37811 2017-02-22 Martin Liska <mliska@suse.cz>
37813 * doc/invoke.texi: Replace inequality signs with square brackets
37816 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37819 * lra-constraints.c (simplify_operand_subreg): Handle
37820 WORD_REGISTER_OPERATIONS targets.
37822 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37825 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37826 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37827 elimination by swapping fld*.
37829 2017-02-22 Richard Biener <rguenther@suse.de>
37831 PR tree-optimization/79673
37832 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37833 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37834 irrelevant address-space qualifiers and avoiding a
37835 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37837 2017-02-22 Richard Biener <rguenther@suse.de>
37839 PR tree-optimization/79666
37840 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37841 to not symbolically negate if that may introduce undefined
37844 2017-02-22 Martin Liska <mliska@suse.cz>
37847 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37848 * data-streamer-out.c (streamer_write_gcov_count_stream):
37850 * value-prof.c (stream_out_histogram_value): Make assert more
37851 precise based on type of counter.
37853 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37856 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37857 Use nonimmediate_operand instead of memory_operand for operand 1.
37858 (float-extend standard_x87sse_constant_load splitter): Ditto.
37860 2017-02-21 Jeff Law <law@redhat.com>
37862 PR tree-optimization/79621
37863 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37864 blocks with edges to themselves.
37866 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37869 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37870 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37871 Use gimple_call_builtin_p.
37874 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37875 on temporarily removed DEBUG_INSNs.
37877 PR tree-optimization/79649
37878 * tree-loop-distribution.c (classify_partition): Give up on
37879 non-generic address space loads/stores.
37881 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37883 * doc/loop.texi (Loop manipulation): Remove nonexistent
37884 tree_ssa_loop_version from the documentation.
37885 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37887 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37890 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37891 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37892 * config/rs6000/rs6000.c: Include except.h.
37893 (rs6000_expand_split_stack_prologue): Call
37894 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37896 2017-02-21 Martin Jambor <mjambor@suse.cz>
37899 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37900 have been analyzed.
37902 2017-02-21 Martin Jambor <mjambor@suse.cz>
37904 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37905 for backward compatibility only.
37906 * doc/invoke.texi (Option Summary): Remove all references to
37907 -fipa-cp-alignment.
37909 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37913 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37915 * lra-constraints.c (curr_insn_transform): Handle
37916 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37918 2017-02-21 Martin Liska <mliska@suse.cz>
37920 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37922 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37924 PR translation/79638
37925 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37927 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37930 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37931 (arm_function_ok_for_sibcall): Return false for an indirect call by
37932 descriptor if all the argument registers are used.
37933 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37934 alignment of the function.
37936 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37938 PR tree-optimization/61441
37939 * simplify-rtx.c (simplify_const_unary_operation): For
37940 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37941 the sNaN unmodified.
37943 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37945 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37946 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37947 instead of SYSTEM_HEADER_DIR.
37949 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37950 Martin Liška <mliska@suse.cz>
37952 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37953 Fix typos and grammar, use active voice, and clarify.
37955 2017-02-20 Marek Polacek <polacek@redhat.com>
37957 PR middle-end/79537
37958 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37961 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37963 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37966 * config/i386/i386.c (ix86_expand_builtin): Handle
37967 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37968 ix86_builtins_isa[fcode].isa as a requirement of those
37969 flags and any other flag in the bitmask.
37970 (ix86_init_mmx_sse_builtins): Use 0 instead of
37971 ~OPTION_MASK_ISA_64BIT as mask.
37972 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37973 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37974 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37975 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37977 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37980 * lra-constraints.c (split_reg): Check requested split mode
37981 is supported by the register.
37983 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37985 * lra-constraints.c (simplify_operand_subreg): Remove early
37988 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37991 * lra-constraints.c (curr_insn_transform): Tighten condition
37992 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37994 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37997 * lra-constraints.c (curr_insn_transform): Handle
37998 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38000 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38003 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38005 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38007 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38010 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38013 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38015 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38018 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38020 * final.c (last_columnnum, override_columnnum): New variables.
38021 (final_start_function): Set last_columnnum, pass it to begin_prologue
38022 hook and pass 0 to dwarf2out_begin_prologue.
38023 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38024 to source_line debug hook.
38025 (notice_source_line): Compute last_columnnum and for debug_column_info
38026 return true on column changes.
38027 * debug.h (struct gcc_debug_hooks): Add column argument to
38028 source_line and begin_prologue hooks.
38029 (debug_nothing_int_charstar_int_bool): Remove prototype.
38030 (debug_nothing_int_int_charstar,
38031 debug_nothing_int_int_charstar_int_bool): New prototypes.
38032 (dwarf2out_begin_prologue): Add column argument.
38033 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38034 begin_prologue hooks.
38035 (debug_nothing_int_charstar_int_bool): Remove.
38036 (debug_nothing_int_int_charstar,
38037 debug_nothing_int_int_charstar_int_bool): New functions.
38038 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38039 through to dwarf2out_source_line.
38040 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38041 (dwarf2out_source_line): Add column argument, emit it if requested.
38042 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38044 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38045 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38046 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38047 through to dwarf2out_begin_prologue.
38048 (vmsdbgout_source_line): Add column argument, pass it through to
38049 dwarf2out_source_line.
38050 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38051 dbxout_source_line caller.
38052 (dbxout_source_line): Add column argument.
38054 * common.opt (gno-column-info, gcolumn-info): New options.
38055 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38056 (check_die): Also test for multiple DW_AT_decl_column attributes.
38057 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38058 DW_AT_decl_column if requested.
38059 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38061 (gen_variable_die): Likewise.
38062 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38063 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38066 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38067 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38068 (ix86_handle_option): Handle OPT_m3dnowa.
38069 * doc/invoke.texi (-m3dnowa): Document.
38070 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38071 -m3dnowa instead of -m3dnow -march=athlon.
38074 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38075 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38077 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38080 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38081 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38082 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38083 generator for vsx_xxpermdi_<mode>_be.
38084 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38085 force big-endian semantics.
38086 (vsx_xxpermdi_<mode>_be): New define_expand with same
38087 implementation as previous version of vsx_xxpermdi_<mode>.
38089 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38091 PR tree-optimization/79327
38092 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38093 variable, its initialization and use.
38095 2017-02-17 Julia Koval <julia.koval@intel.com>
38097 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38098 (OPTION_MASK_ISA_PKU_UNSET): New.
38099 (ix86_handle_option): Handle -mrdpid.
38100 * config/i386/cpuid.h (bit_RDPID): New.
38101 * config/i386/driver-i386.c (host_detect_local_cpu):
38102 Detect RDPID feature.
38103 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38104 * config/i386/i386-c.c (ix86_target_macros_internal):
38106 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38107 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38108 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38109 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38110 * config/i386/i386.md (define_insn "rdpid"): New.
38111 * config/i386/i386.opt Add -mrdpid.
38112 * config/i386/immintrin.h (_rdpid_u32): New.
38114 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38116 PR rtl-optimization/79541
38117 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38118 instead of transforming it into USE.
38120 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38122 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38123 If HONOR_SNANS (SFmode) force the input to a register.
38124 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38125 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38126 an frsp or similar insn.
38128 2017-02-17 Martin Liska <mliska@suse.cz>
38130 PR rtl-optimization/79577
38131 * params.def (selsched-max-sched-times): Increase minimum to 1.
38133 2017-02-17 Martin Liska <mliska@suse.cz>
38135 PR rtl-optimization/79574
38136 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38138 2017-02-17 Martin Liska <mliska@suse.cz>
38140 PR tree-optimization/79529
38141 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38142 ssa_defined_default_def_p to handle cases which are implicitly
38144 * tree-ssa.c (ssa_defined_default_def_p): New function.
38145 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38146 which are implicitly defined.
38147 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38149 2017-02-17 Richard Biener <rguenther@suse.de>
38151 PR middle-end/79576
38152 * params.def (max-ssa-name-query-depth): Limit to 10.
38154 2017-02-17 Richard Biener <rguenther@suse.de>
38156 PR tree-optimization/79552
38157 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38160 2017-02-17 Richard Biener <rguenther@suse.de>
38163 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38165 2017-02-17 Marek Polacek <polacek@redhat.com>
38167 PR middle-end/79536
38168 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38169 (fold_negate_expr): New wrapper.
38171 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38173 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38174 Correct terminology and de-emphasize pre-standard behavior.
38176 2017-02-16 Alan Modra <amodra@gmail.com>
38178 PR rtl-optimization/79286
38179 * ira.c (def_dominates_uses): New function.
38180 (update_equiv_regs): Don't create an equivalence for insns that
38181 may trap where the register def does not dominate the use.
38183 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38185 PR rtl-optimization/78127
38186 * lra.c (lra): Call lra_eliminate before finish the loop after
38189 2017-02-16 Richard Biener <rguenther@suse.de>
38191 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38193 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38194 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38195 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38196 (isl_val_int_from_wi): New function.
38197 (extract_affine_gmp): Rename to ...
38198 (extract_affine_wi): ... this, take a widest_int.
38199 (extract_affine_int): Just wrap extract_affine_wi.
38200 (add_param_constraints): Use isl_val_int_from_wi.
38201 (add_loop_constraints): Likewise, and extract_affine_wi.
38203 2017-02-15 Jeff Law <law@redhat.com>
38205 PR middle-end/79521
38206 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38207 ira_init_register_move_cost_if_necessary.
38209 2017-02-15 Martin Sebor <msebor@redhat.com>
38211 PR middle-end/32003
38212 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38213 removed in a prior commit.
38215 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38217 PR tree-optimization/79347
38218 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38219 counters during peeling.
38221 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38223 * Makefile.in (site.exp): Remove "set ISLVER".
38225 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38228 * real.c (real_from_integer): Call real_convert even for decimal.
38230 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38233 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38235 2017-02-14 Andrew Pinski <apinski@cavium.com>
38237 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38238 cores and change the partno/implementer to be correct.
38239 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38240 the 'B" as the implementer.
38241 * config/aarch64/aarch64-tune.md: Regenerate.
38243 2017-02-14 Carl Love <cel@us.ibm.com>
38245 * config/rs6000/rs6000.c: Add case statement entry to make the
38246 xvcvuxdsp built-in argument unsigned.
38247 * config/rs6000/vsx.md: Fix the source and return operand types so they
38248 match the instruction definitions from the ISA document. Fix typo
38249 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38252 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38255 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38256 member early_clobber_alts.
38257 * lra-lives.c (reg_early_clobber_p): New.
38258 (process_bb_lives): Use it.
38259 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38260 (debug_operand_data): Initialize early_clobber_alts.
38261 (setup_operand_alternative): Set up early_clobber_alts.
38262 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38263 alternatives to new_insn_reg.
38264 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38266 (lra_update_insn_regno_info): Pass the new arg.
38268 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38270 PR middle-end/79505
38271 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38272 (new_oacc_loop_raw): Don't clear already cleared fields.
38275 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38276 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38277 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38279 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38282 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38284 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38287 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38288 the extra instruction to the right place to store 128-bit constant
38291 2017-02-14 Martin Sebor <msebor@redhat.com>
38293 PR middle-end/79448
38294 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38295 warning for strings of unknown length.
38297 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38299 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38301 2017-02-14 Jeff Law <law@redhat.com>
38304 * ira-costs.c (scan_one_insn): Initialize register move costs
38305 for pseudos seen in USE/CLOBBER insns.
38307 PR tree-optimization/79095
38308 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38309 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38310 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38311 if the operands are known to be not equal, then the resulting range
38313 (intersect_ranges): If the new range is ~[0,0] and the old range is
38314 wide, then prefer ~[0,0].
38315 * tree-vrp.c (overflow_comparison_p_1): New function.
38316 (overflow_comparison_p): New function.
38317 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38318 if NAME is used in an overflow test.
38319 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38320 overflow check that can be expressed as an equality test, then adjust
38321 ops to be that equality test.
38323 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38325 * config/s390/s390-builtin-types.def: Remove flags argument.
38326 * config/s390/s390.c (s390_init_builtins): Likewise.
38328 2017-02-14 Martin Liska <mliska@suse.cz>
38330 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38331 vector. Fix trailing white spaces.
38333 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38335 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38338 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38340 PR rtl-optimization/68664
38341 * config/arm/arm.c (arm_sched_can_speculate_insn):
38342 New function. Declare prototype.
38343 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38345 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38347 PR rtl-optimization/68664
38348 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38350 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38352 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38354 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38355 max skip bytes for function, loop and jump.
38357 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38359 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38360 ABS_EXPR for gimple dump.
38362 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38365 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38367 PR tree-optimization/79408
38368 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38369 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38370 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38371 also if rhs1 is INTEGER_CST.
38373 2017-02-14 Richard Biener <rguenther@suse.de>
38375 PR middle-end/79432
38376 * tree-into-ssa.c (insert_phi_nodes): When the function can
38377 have abnormal edges rewrite SSA names with broken use-def
38378 dominance out of SSA and register them for PHI insertion.
38380 2017-02-13 Martin Sebor <msebor@redhat.com>
38382 PR middle-end/79496
38383 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38384 clearing info.nowrite flag when snprintf size argument is a range.
38386 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38388 * cprop.c (cprop_jump): Add missing space in string literal.
38389 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38390 (get_constraint_for_component_ref): Likewise.
38391 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38392 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38393 * lra-constraints.c (process_alt_operands): Likewise.
38394 * ipa-inline.c (inline_small_functions): Likewise.
38395 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38396 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38397 * trans-mem.c (diagnose_tm_1_op): Likewise.
38398 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38399 (grid_parallel_clauses_gridifiable): Likewise.
38401 * config/nvptx/mkoffload.c (process): Add space in between
38404 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38405 "MOD4_SSE_REGS" and "ALL_REGS".
38407 * spellcheck.c (test_data): Add , in between "foo" and "food".
38409 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38412 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38413 boundary crossing check and subsequent code generation agree.
38415 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38417 * config/aarch64/aarch64.c (has_memory_op): Delete.
38418 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38421 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38423 PR rtl-optimization/79388
38424 PR rtl-optimization/79450
38425 * combine.c (distribute_notes): When removing TEM_INSN for which
38426 corresponding dest has last value recorded, invalidate that last
38429 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38431 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38432 of explicit '@'. Add missing assembly comment marker on branch costs
38435 2017-02-13 Nathan Sidwell <nathan@acm.org>
38437 * gengtype-lex.l (<in_struct>): Add '/'.
38439 2017-02-13 Martin Liska <mliska@suse.cz>
38442 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38444 2017-02-13 Richard Biener <rguenther@suse.de>
38446 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38448 * configure: Re-generate.
38449 * config.in: Likewise.
38450 * graphite-dependences.c: Simplify as if
38451 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38452 * graphite-isl-ast-to-gimple.c: Likewise.
38453 * graphite-optimize-isl.c: Likewise.
38454 * graphite-poly.c: Likewise.
38455 * graphite-sese-to-poly.c: Likewise.
38456 * graphite.h: Likewise.
38457 * toplev.c: Include isl/version.h and use isl_version () for
38458 printing the ISL version.
38459 * doc/install.texi: Update ISL requirement.
38461 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38463 * doc/standards.texi (Standards): Update reference to
38466 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38468 * doc/extend.texi (Named Address Spaces): sourceware.org now
38470 * doc/install.texi (Binaries): Ditto.
38473 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38475 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38476 standard terminology "stringize"/"stringizing" throughout.
38477 * doc/cppinternals.texi: Likewise.
38479 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38481 * doc/extend.texi: Fix some spelling mistakes and typos.
38482 * doc/invoke.texi: Likewise.
38484 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38487 * params.def (inline-min-speedup) Change from 10 to 8.
38489 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38491 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38494 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38497 * ipa-inline-analysis.c (get_minimal_bb): New function.
38498 (record_modified): Use it.
38499 (remap_edge_change_prob): Handle also ancestor functions.
38501 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38503 * doc/contrib.texi (Contributors): Remove broken link into
38504 the Mauve CVS repository.
38506 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38508 PR middle-end/79454
38509 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38510 result computation whenever lhs doesn't have vector mode, not
38511 just when it has BLKmode.
38513 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38515 * doc/makefile.texi (profiledbootstrap): Refer to the
38516 installation instructions only in textual form.
38518 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38521 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38523 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38525 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38526 (Specific): Update mingw-w64 reference.
38528 (Specific): Remove broken link to Renesas RX processor.
38530 2017-02-10 Richard Biener <rguenther@suse.de>
38532 * toplev.c (process_options): Do not mention obsolete graphite
38533 options when printing sorry message about missing graphite support.
38534 Mention -floop-nest-optimize.
38536 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38538 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38539 (vtst_p16): Likewise.
38540 (vtstq_p8): Likewise.
38541 (vtstq_p16): Likewise.
38543 (vtstq_p64): Likewise.
38544 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38545 (vset_lane_p64): New.
38546 (vsetq_lane_p64): New.
38548 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38550 PR tree-optimization/79411
38551 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38552 stmt operands are SSA_NAMEs used in abnormal phis.
38553 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38556 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38559 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38562 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38564 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38566 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38568 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38572 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38573 not arbitrary TREE_CONSTANT.
38576 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38577 "omp declare target link" attribute unless is_global_var.
38578 * omp-offload.c (find_link_var_op): Likewise.
38580 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38581 Chung-Lin Tang <cltang@codesourcery.com>
38583 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38585 (gimplify_adjust_omp_clauses): Don't delete TILE.
38586 (gimplify_omp_for): Deal with TILE.
38587 * internal-fn.c (expand_GOACC_TILE): New function.
38588 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38590 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38591 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38593 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38594 avoid DIV for outermost collapse var.
38595 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38596 Remove out of date comments, fix whitespace.
38597 * omp-general.c (omp_extract_for_data): Deal with tiling.
38598 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38599 adjust OLF_DIM_BASE value.
38600 (struct omp_for_data): Add tiling field.
38601 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38602 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38603 for auto loops. Remove default auto determining, moved to
38604 oacc_loop_fixed_partitions.
38605 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38606 stmts, add e_mask field.
38607 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38608 (oacc_thread_numbers): Use oacc_dim_call.
38609 (oacc_xform_tile): New.
38610 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38611 (finish_oacc_loop): Adjust for ifns vector.
38612 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38613 add case for GOACC_TILE fns.
38614 (oacc_loop_xform_loop): Delete.
38615 (oacc_loop_process): Iterate over call list directly, and add
38616 handling for GOACC_TILE fns.
38617 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38619 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38620 vector partitioning to outer loops. Assign 2 partitions to loops
38621 when available. Add TILE handling.
38622 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38623 (execite_oacc_device_lower): Process GOACC_TILE fns,
38624 ignore unknown specs.
38625 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38626 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38627 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38629 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38631 * configure.ac (ACX_BUGURL): Update.
38632 * configure: Regenerate.
38634 2017-02-09 Richard Biener <rguenther@suse.de>
38636 PR tree-optimization/69823
38637 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38638 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38640 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38642 * config/arc/arc-c.def: Add __NPS400__ definition.
38643 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38644 (TARGET_NPS400): Define.
38646 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38648 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38650 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38651 pointer, arch_info.
38652 (arc_cpu_types): Fill the arch_info field with a pointer into the
38653 arc_arch_types table.
38654 (arc_selected_cpu): Declare.
38655 * config/arc/arc.c (arc_selected_cpu): Make global.
38656 (arc_selected_arch): Delete.
38657 (arc_base_cpu): Delete.
38658 (arc_override_options): Remove references to deleted variables,
38659 update access to arch information.
38660 (ARC_OPT): Update access to arch information.
38661 (ARC_OPTX): Likewise.
38662 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38663 (TARGET_ARC600): Update access to arch information.
38664 (TARGET_ARC601): Likewise.
38665 (TARGET_ARC700): Likewise.
38666 (TARGET_EM): Likewise.
38667 (TARGET_HS): Likewise.
38668 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38671 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38674 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38675 condition/operands for integer GE/LE/GEU/LEU operations.
38677 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38679 PR translation/79397
38680 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38683 2017-02-08 Martin Jambor <mjambor@suse.cz>
38686 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38687 whether allocation happened.
38688 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38689 nothing was allocated.
38691 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38693 PR tree-optimization/79408
38694 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38695 constant, but SSA_NAME with a known integer range, use the minimum
38696 of that range instead of op1 to determine if modulo can be replaced
38697 with its first operand.
38699 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38701 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38703 2017-02-08 Richard Biener <rguenther@suse.de>
38705 PR tree-optimization/71824
38706 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38707 Check all loops contained in the merged region.
38709 2017-02-07 Andrew Pinski <apinski@cavium.com>
38711 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38713 2017-02-07 Andrew Pinski <apinski@cavium.com>
38715 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38716 (thunderxt88): Likewise.
38717 (thunderxt81): Disable LSE and change v8.1 to v8.
38718 (thunderxt83): Likewise.
38720 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38721 Richard Biener <rguenther@suse.de>
38723 PR middle-end/79399
38724 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38725 type from int to size_t.
38726 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38728 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38730 PR rtl-optimization/79386
38731 * cprop.c (bypass_conditional_jumps): Initialize
38732 bypass_last_basic_block already before splitting bbs after
38733 unconditional traps...
38734 (bypass_conditional_jumps): ... rather than here.
38737 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38738 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38739 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38740 fix -masm=intel patterns.
38742 2017-02-07 Richard Biener <rguenther@suse.de>
38744 PR tree-optimization/79256
38745 PR middle-end/79278
38746 * builtins.c (get_object_alignment_2): Use min_align_of_type
38747 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38748 and ADJUST_FIELD_ALIGN.
38750 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38752 * doc/tm.texi: Regenerate.
38753 * stor-layout.c (layout_decl): Adjust.
38754 (update_alignment_for_field): Likewise.
38755 (place_field): Likewise.
38756 (min_align_of_type): Likewise.
38757 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38758 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38759 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38760 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38761 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38762 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38763 * config/i386/i386.c (x86_field_alignment): Likewise.
38764 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38765 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38766 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38767 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38768 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38769 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38773 2017-01-30 Richard Biener <rguenther@suse.de>
38775 PR tree-optimization/79256
38776 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38777 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38780 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38782 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38783 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38784 builtins to SImode and emit a zero-extend, if necessary.
38786 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38788 * docs/invoke.texi (RISC-V Options): Alphabetize.
38790 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38792 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38795 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38797 * config/riscv/riscv.c: New file.
38798 * common/config/riscv/riscv-common.c: Likewise.
38799 * config.gcc: Likewise.
38800 * config/riscv/constraints.md: Likewise.
38801 * config/riscv/elf.h: Likewise.
38802 * config/riscv/generic.md: Likewise.
38803 * config/riscv/linux.h: Likewise.
38804 * config/riscv/multilib-generator: Likewise.
38805 * config/riscv/peephole.md: Likewise.
38806 * config/riscv/pic.md: Likewise.
38807 * config/riscv/predicates.md: Likewise.
38808 * config/riscv/riscv-builtins.c: Likewise.
38809 * config/riscv/riscv-c.c: Likewise.
38810 * config/riscv/riscv-ftypes.def: Likewise.
38811 * config/riscv/riscv-modes.def: Likewise.
38812 * config/riscv/riscv-opts.h: Likewise.
38813 * config/riscv/riscv-protos.h: Likewise.
38814 * config/riscv/riscv.h: Likewise.
38815 * config/riscv/riscv.md: Likewise.
38816 * config/riscv/riscv.opt: Likewise.
38817 * config/riscv/sync.md: Likewise.
38818 * config/riscv/t-elf-multilib: Likewise.
38819 * config/riscv/t-linux: Likewise.
38820 * config/riscv/t-linux-multilib: Likewise.
38821 * config/riscv/t-riscv: Likewise.
38822 * configure.ac: Likewise.
38823 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38824 Waterman as RISC-V maintainers.
38825 * doc/install.texi: Add RISC-V entries.
38826 * doc/invoke.texi: Add RISC-V options section.
38827 * doc/md.texi: Add RISC-V constraints section.
38828 * configure: Regenerated.
38830 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38833 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38834 false values to be constant vectors with all 0 or all 1 bits set.
38835 (vcondu<mode><mode>): Likewise.
38836 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38838 (fpmask_comparison_operator): Update comment.
38839 (vecint_comparison_operator): New predicate.
38840 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38841 vector conditionals when the true and false values are constant
38842 vectors with all 0 bits or all 1 bits set.
38844 2017-02-06 Martin Sebor <msebor@redhat.com>
38846 PR tree-optimization/79376
38847 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38849 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38851 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38852 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38853 to simplify split condition.
38855 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38857 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38858 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38861 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38863 PR rtl-optimization/68664
38864 * target.def (can_speculate_insn): New hook.
38865 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38866 * doc/tm.texi: Regenerate.
38867 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38868 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38869 (rs6000_sched_can_speculate_insn): New function.
38871 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38873 PR tree-optimization/79284
38874 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38875 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38876 vectorizable_mask_load_store, vectorizable_operation,
38877 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38878 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38879 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38880 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38881 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38882 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38883 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38884 is_gimple_assign (stmt). Replace another such test with
38885 is_gimple_assign (stmt).
38887 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38890 * config/avr/avr.c (rtl-iter.h): Include it.
38891 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38892 (avr_legitimate_combined_insn): ...and implementation.
38894 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38896 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38897 * config/s390/s390.c (s390_const_operand_ok)
38898 (s390_canonicalize_comparison, s390_extract_part)
38899 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38900 (s390_contiguous_bitmask_p, s390_rtx_costs)
38901 (legitimize_pic_address): Likewise.
38902 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38903 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38904 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38905 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38906 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38908 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38910 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38911 REGNO($0) == REGNO($1).
38913 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38915 * config/s390/linux.h(SIZE_TYPE): Add comment.
38917 2017-02-06 Julian Brown <julian@codesourcery.com>
38918 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38919 Virendra Pathak <virendra.pathak@broadcom.com>
38921 * config/aarch64/aarch64-cores.def: Change the scheduler
38923 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38924 * config/aarch64/thunderx2t99.md: New file.
38926 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38928 * doc/standards.texi (Go Language): Update link to language
38931 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38933 * tree-eh.c (lower_resx): Sanitize profile.
38934 (cleanup_empty_eh_move_lp): Likewise.
38936 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38939 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38941 * cfgloopmanip.h (loop_version): Update prototype.
38942 * modulo-sched.c (sms_schedule): Update call of loop_version.
38943 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38944 * tree-parloops.c (gen_parallel_loop): Likewise.
38945 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38946 * tree-ssa-loop-split.c (split_loop): Likewise.
38947 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38948 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38950 2017-02-05 Martin Liska <mliska@suse.cz>
38953 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38955 (print_operand_address): Initialize a struct to zero.
38957 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38959 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38960 garbage collector only in textual form.
38962 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38964 * doc/extend.texi (x86 specific memory model extensions for
38965 transactional memory): Simplify a phrase.
38967 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38970 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38971 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38972 (atomic_storedi_1): Likewise.
38974 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38976 PR tree-optimization/79338
38977 * tree-parloops.c (gather_scalar_reductions): Don't call
38978 vect_analyze_loop_form for loop->inner before destroying loop's
38981 2017-02-03 Martin Sebor <msebor@redhat.com>
38983 PR tree-optimization/79327
38984 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38985 when precision has resulted in leading zeros.
38986 (format_integer): Adjust the likely counter to assume an unknown
38987 argument that may be zero is non-zero.
38989 2017-02-03 Jason Merrill <jason@redhat.com>
38992 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38993 avoid copying non-taken branch.
38995 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38997 PR tree-optimization/79340
38998 * tree-vect-loop.c (vectorizable_reduction): Release
38999 vec_defs elements after safe_splicing them into other vectors.
39002 PR tree-optimization/79327
39003 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39004 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39006 (format_integer): Use wide_int_to_tree instead of build_int_cst
39007 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39008 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39009 of shortest and longest sequence.
39011 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39013 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39014 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39016 2017-02-03 Walter Lee <walt@tilera.com>
39019 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39020 after initial stackframe link reg save.
39021 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39023 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39026 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39027 wu for stxssp alternative.
39029 2017-02-03 Martin Sebor <msebor@redhat.com>
39031 PR tree-optimization/79352
39032 * gimple-fold.c (get_range_strlen): Add argument.
39033 (get_range_strlen): Change return type to bool.
39034 (get_maxval_strlen): Pass in a dummy argument.
39035 * gimple-fold.h (get_range_strlen): Change return type to bool.
39036 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39037 * tree.h (array_at_struct_end_p): Add argument.
39038 * tree.c (array_at_struct_end_p): Handle it.
39040 2017-02-03 Martin Liska <mliska@suse.cz>
39043 * multiple_target.c (create_dispatcher_calls): Redirect edge
39044 from a caller of a dispatcher.
39045 (expand_target_clones): Make the clones local.
39046 (ipa_target_clone): Do both target clones and resolvers.
39047 (ipa_dispatcher_calls): Remove the pass.
39048 (pass_dispatcher_calls::gate): Likewise.
39049 (make_pass_dispatcher_calls): Likewise.
39050 * passes.def (pass_target_clone): Put as very first IPA early
39053 2017-02-03 Martin Liska <mliska@suse.cz>
39055 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39056 in case of a function with ifunc attribute.
39058 2017-02-03 Martin Liska <mliska@suse.cz>
39060 * cgraph.c (cgraph_node::dump): Dump function version info.
39061 * symtab.c (symtab_node::dump_base): Add missing new line.
39063 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39065 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39066 (ifcombine_ifandif): Use it.
39068 2017-02-03 Martin Liska <mliska@suse.cz>
39070 * doc/invoke.texi: Document default value for
39071 use-after-scope-direct-emission-threshold.
39073 2017-02-03 Martin Liska <mliska@suse.cz>
39075 PR tree-optimization/79339
39076 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39077 (format_floating): Likewise.
39079 2017-02-03 Martin Liska <mliska@suse.cz>
39082 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39084 (ipa_node_params_t::remove): Likewise.
39085 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39086 initialization from removed ipa_node_params_t::insert.
39087 (ipa_node_params::~ipa_node_params): Move from removed
39088 ipa_node_params_t::release.
39089 * symbol-summary.h (symbol_summary::m_released): New member.
39090 Do not release a summary twice. Do not allow to call finalizer
39091 for types of a summary that live in GGC memory.
39093 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39095 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39098 2017-02-02 Martin Sebor <msebor@redhat.com>
39100 PR middle-end/79275
39101 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39102 (format_string): Tighten up the range of output for non-constant
39103 strings and correct the expected range for wide non-constant strings.
39105 2017-02-02 Martin Sebor <msebor@redhat.com>
39107 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39109 PR middle-end/32003
39110 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39112 (-fdump-tree-@var): Add to index and document how to come up
39113 with pass-specific option and dump file names.
39114 (-fdump-passes): Clarify where to look for output.
39116 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39118 PR middle-end/77445
39119 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39120 statistics of the analyzed path; allow threading for speed when
39121 any of BBs along the path are optimized for speed.
39123 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39125 PR middle-end/78468
39126 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39127 settings of the virtual registers.
39130 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39132 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39133 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39136 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39138 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39139 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39141 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39143 * config/s390/s390.md: Add missing comments with the expanded
39145 * config/s390/vector.md: Likewise.
39146 * config/s390/vx-builtins.md: Likewise.
39148 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39151 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39152 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39153 conditions on a single line.
39155 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39157 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39158 __S390_VX__ to __VX__.
39160 2017-02-01 Andrew Pinski <apinski@cavium.com>
39162 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39163 stmt_info to record_stmt_cost.
39164 (vect_get_known_peeling_cost): Pass stmt_info if known to
39166 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39167 cpu_vector_cost field into
39168 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39169 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39170 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39171 splitting of scalar_stmt_cost and vec_stmt_cost.
39172 (thunderx_vector_cost): Likewise.
39173 (cortexa57_vector_cost): LIkewise.
39174 (exynosm1_vector_cost): Likewise.
39175 (xgene1_vector_cost): Likewise.
39176 (thunderx2t99_vector_cost): Improve after the splitting of the two
39178 (aarch64_builtin_vectorization_cost): Update for the splitting of
39179 scalar_stmt_cost and vec_stmt_cost.
39181 2017-02-01 Torvald Riegel <triegel@redhat.com>
39182 Richard Henderson <rth@redhat.com>
39184 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39185 conditional on existance of a fast atomic load.
39186 * optabs-query.c (can_atomic_load_p): New function.
39187 * optabs-query.h (can_atomic_load_p): Declare it.
39188 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39189 no fast atomic load is available for the particular size of access.
39190 (expand_atomic_compare_and_swap): Likewise.
39191 (expand_atomic_load): Likewise.
39192 (expand_atomic_store): Likewise.
39193 (expand_atomic_fetch_op): Likewise.
39194 * testsuite/lib/target-supports.exp
39195 (check_effective_target_sync_int_128): Remove x86 because it provides
39196 no fast atomic load.
39197 (check_effective_target_sync_int_128_runtime): Likewise.
39199 2017-02-01 Richard Biener <rguenther@suse.de>
39201 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39202 (graphite_transform_loops): Provide opt-info for optimized nests.
39203 * tree-parloop.c (parallelize_loops): Provide opt-info for
39204 parallelized loops.
39206 2017-02-01 Richard Biener <rguenther@suse.de>
39208 PR middle-end/79315
39209 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39210 was not set before.
39212 2017-02-01 Richard Biener <rguenther@suse.de>
39214 PR tree-optimization/71824
39215 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39216 Verify the loops are valid in the merged SESE region.
39217 (scop_detection::can_represent_loop_1): Check analyzing the
39218 evolution of the number of iterations in the region succeeds.
39220 2017-01-31 Ian Lance Taylor <iant@golang.org>
39222 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39223 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39225 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39227 PR preprocessor/79210
39228 * input.c (get_substring_ranges_for_loc): Replace line_width
39229 assertion with error-handling.
39231 2017-01-31 Richard Biener <rguenther@suse.de>
39233 PR tree-optimization/77318
39234 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39235 (create_pw_aff_from_tree): Take loop parameter.
39236 (add_condition_to_pbb): Pass loop of the condition to
39237 create_pw_aff_from_tree.
39239 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39241 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39242 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39244 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39248 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39250 (convert_int_to_float128): Likewise.
39251 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39252 (convert_int_to_float128): Likewise.
39253 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39254 (UNSPEC_IEEE128_CONVERT): Likewise.
39255 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39256 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39257 Use local variables for IBM extended format.
39258 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39259 (fix_trunc<mode>si2_fprs): Likewise.
39260 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39261 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39262 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39263 to know that we can now have integers of all sizes in vector
39265 (fix<uns>_<mode>di2_hw): Likewise.
39266 (float<uns>_<mode>si2_hw): Likewise.
39267 (fix_<mode>si2_hw): Likewise.
39268 (fixuns_<mode>si2_hw): Likewise.
39269 (float<uns>_<mode>di2_hw): Likewise.
39270 (float_<mode>di2_hw): Likewise.
39271 (float_<mode>si2_hw): Likewise.
39272 (floatuns_<mode>di2_hw): Likewise.
39273 (floatuns_<mode>si2_hw): Likewise.
39274 (xscvqp<su>wz_<mode>): Delete, no longer used.
39275 (xscvqp<su>dz_<mode>): Likewise.
39276 (xscv<su>dqp_<mode>): Likewise.
39277 (ieee128_mfvsrd_64bit): Likewise.
39278 (ieee128_mfvsrd_32bit): Likewise.
39279 (ieee128_mfvsrwz): Likewise.
39280 (ieee128_mtvsrw): Likewise.
39281 (ieee128_mtvsrd_64bit): Likewise.
39282 (ieee128_mtvsrd_32bit): Likewise.
39284 2017-01-31 Martin Liska <mliska@suse.cz>
39287 * ipa-prop.c (ipa_free_all_node_params): Call release method
39288 instead of ~sumbol_summary to not to trigger double times
39291 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39293 PR tree-optimization/71691
39294 * bitmap.h (class auto_bitmap): New.
39295 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39296 is_maybe_undefined instead of ssa_undefined_value_p.
39298 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39300 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39301 __S390_ARCH_LEVEL__ to __ARCH__.
39303 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39305 PR tree-optimization/79267
39306 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39307 if should_remove_lhs_p is true.
39309 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39312 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39313 (add_alignment_attribute): New.
39314 (base_type_die): Add alignment attribute.
39315 (subrange_type_die): Likewise.
39316 (modified_type_die): Likewise.
39317 (gen_array_type_die): Likewise.
39318 (gen_descr_array_type_die: Likewise.
39319 (gen_enumeration_type_die): Likewise.
39320 (gen_subprogram_die): Likewise.
39321 (gen_variable_die): Likewise.
39322 (gen_field_die): Likewise.
39323 (gen_ptr_to_mbr_type_die): Likewise.
39324 (gen_struct_or_union_type_die): Likewise.
39325 (gen_subroutine_type_die): Likewise.
39326 (gen_typedef_die): Likewise.
39327 (base_type_cmp): Compare alignment attribute.
39329 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39332 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39333 (setb_unsigned) New pattern for setb with CCUNS.
39334 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39335 subfc./subfe sequence to avoid overflow problems. Generate a
39336 shorter sequence with cmpld/setb for power9.
39337 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39338 for generating subfc. instruction.
39339 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39340 now uses this instruction.
39342 2017-01-30 Ian Lance Taylor <iant@google.com>
39345 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39346 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39348 2017-01-30 Martin Sebor <msebor@redhat.com>
39350 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39351 Move constant to the right of a relational operator.
39352 (get_mpfr_format_length, format_character, format_string): Ditto.
39353 (should_warn_p, maybe_warn): Same.
39355 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39357 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39360 * asan.c (get_translation_unit_decl): Remove function.
39361 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39363 2017-01-30 Martin Liska <mliska@suse.cz>
39365 PR gcov-profile/79259
39366 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39367 -fprofile-generate.
39369 2017-01-30 Martin Liska <mliska@suse.cz>
39372 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39373 Initialize variables with NULL value.
39375 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39378 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39380 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39382 2017-01-30 Richard Biener <rguenther@suse.de>
39384 PR tree-optimization/79276
39385 * tree-vrp.c (process_assert_insertions): Properly adjust common
39386 when removing a duplicate.
39388 2017-01-30 Richard Biener <rguenther@suse.de>
39390 PR tree-optimization/79256
39391 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39392 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39394 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39396 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39399 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39400 ("*r<noxa>sbg_<mode>_sll_bitmask")
39401 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39402 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39403 Use contiguous_bitmask_nowrap_operand.
39405 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39408 * config/rs6000/altivec.h (vec_xl): Revise #define.
39409 (vec_xst): Likewise.
39411 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39413 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39415 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39417 PR rtl-optimization/79194
39418 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39419 traps before call to bypass_conditional_jumps.
39421 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39423 PR tree-optimization/71374
39424 * lra-constraints.c (check_conflict_input_operands): New.
39425 (match_reload): Use it.
39427 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39430 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39431 account to calculate conflict_set.
39433 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39435 PR rtl-optimization/78559
39436 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39437 other_insn in combine.
39439 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39441 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39442 uint16_type_node for BT_UINT16.
39444 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39446 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39447 "RTL Tests" to menu.
39448 (GIMPLE Tests): New node.
39449 (RTL Tests): New node.
39451 2017-01-27 Richard Biener <rguenther@suse.de>
39453 PR tree-optimization/79245
39454 * tree-loop-distribution.c (distribute_loop): Apply cost
39455 modeling also to detected patterns.
39457 2017-01-27 Richard Biener <rguenther@suse.de>
39459 PR tree-optimization/71433
39460 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39461 (compare_assert_loc): New function.
39462 (process_assert_insertions): Sort and optimize assert locations
39463 to remove duplicates and push down identical assertions on
39464 edges to their destination block.
39466 2017-01-27 Richard Biener <rguenther@suse.de>
39468 PR tree-optimization/79244
39469 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39470 out SSA names even if abnormal.
39472 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39474 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39475 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39476 instead of MPFR_RNDN.
39478 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39481 * arm.c (arm_option_override): Don't call build_target_option_node
39482 until after doing all option overrides.
39483 (arm_valid_target_attribute_tree): Likewise.
39485 2017-01-27 Martin Liska <mliska@suse.cz>
39487 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39488 for {cd}tors and C++ {cd}tors.
39490 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39492 * config/s390/s390.md ("*setmem_long_and")
39493 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39495 2017-01-26 Martin Sebor <msebor@redhat.com>
39497 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39500 2017-01-26 Martin Sebor <msebor@redhat.com>
39502 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39503 HAVE_DFmode before using XFmode or DFmode.
39504 (parse_directive): Avoid using the z length modifier to avoid
39505 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39507 PR middle-end/78703
39508 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39509 to accept adjustment as an array.
39510 (get_int_range): New function.
39511 (struct directive): Make width and prec arrays.
39512 (directive::set_width, directive::set_precision): Call get_int_range.
39513 (format_integer, format_floating): Handle width and precision ranges.
39514 (format_string, parse_directive): Same.
39516 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39519 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39520 comdat_type_p set, just clone them, but keep the children in the
39524 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39525 which have direct callers with -fvar-tracking-assignments enabled
39527 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39528 inside of type units.
39530 2017-01-26 Martin Sebor <msebor@redhat.com>
39532 PR middle-end/78703
39533 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39535 (struct format_result): Replace number_chars, number_chars_min,
39536 and number_chars_max with a single member of struct result_range.
39538 (format_result::operator+=): Adjust.
39539 (struct fmtresult): Remove bounded. Handle likely and unlikely
39541 (fmtresult::adjust_for_width_or_precision): New function.
39542 (fmtresult:type_max_digits): New function.
39543 (bytes_remaining): Handle likely and unlikely counters.
39544 (min_bytes_remaining): Remove.
39545 (format_percent): Simplify.
39546 (format_integer, format_floating): Set likely and unlikely counters.
39547 (get_string_length, format_character, format_string): Same.
39548 (format_plain, should_warn_p): New function.
39549 (maybe_warn): Call should_warn_p. Update diagnostic messages
39550 and handle those for all directives, including plain strings.
39551 (format_directive): Handle likely and unlikely counters.
39552 Remove unnecessary quoting from diagnostics. Add an informational
39554 (add_bytes): Remove.
39555 (pass_sprintf_length::compute_format_length): Simplify.
39556 (try_substitute_return_value): Handle likely and unlikely counters.
39558 2017-01-26 Carl Love <cel@us.ibm.com>
39560 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39561 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39563 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39566 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39567 endianess for subregs into account.
39568 * lra-constraints.c (lra_constraints): Do risky transformations
39569 always on the first iteration.
39570 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39571 last_call_used_reg_set.
39572 (process_bb_lives): Define and use last_call_used_reg_set.
39573 * lra.c (lra): Always continue after lra_constraints on the first
39576 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39578 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39580 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39582 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39584 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39585 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39586 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39587 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39588 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39589 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39590 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39591 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39592 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39594 2017-01-26 Marek Polacek <polacek@redhat.com>
39597 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39598 for the third operand.
39600 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39602 PR middle-end/79236
39603 * omp-low.c (struct omp_context): Add simt_stmt field.
39604 (scan_omp_for): Return omp_context *.
39605 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39606 context to the _simt_ SIMD stmt.
39607 (lower_omp_for): For combined SIMD with sibling _simt_
39608 SIMD, make sure to use the same decls in _looptemp_
39609 clauses as in the sibling.
39611 2017-01-26 David Sherwood <david.sherwood@arm.com>
39613 PR middle-end/79212
39614 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39617 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39620 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39621 emit fld b; fld a; if possible.
39623 * brig-builtins.def: Update copyright years.
39624 * config/arm/arm_acle_builtins.def: Update copyright years.
39626 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39629 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39630 constraint instead of o for the stxsd instruction.
39632 2017-01-25 Carl Love <cel@us.ibm.com>
39634 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39635 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39637 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39639 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39641 2017-01-25 Richard Biener <rguenther@suse.de>
39643 PR tree-optimization/69264
39644 * target.def (vector_alignment_reachable): Improve documentation.
39645 * doc/tm.texi: Regenerate.
39646 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39648 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39649 earlier changes with respect to TYPE_USER_ALIGN.
39650 (vector_alignment_reachable_p): Likewise. Improve dumping.
39652 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39655 * config/arm/arm.md (xordi3): Force constant operand into a register
39658 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39660 * doc/invoke.texi (-fstore-merging): Correct default optimization
39661 levels at which it is enabled.
39662 (-O): Move -fstore-merging from list to...
39665 2017-01-25 Richard Biener <rguenther@suse.de>
39668 * omp-expand.c: Include debug.h.
39669 (expand_omp_taskreg): Make sure to generate early debug before
39670 outlining anything from a function.
39671 (expand_omp_target): Likewise.
39672 (grid_expand_target_grid_body): Likewise.
39674 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39677 * asan.c (get_translation_unit_decl): New function.
39678 (asan_add_global): Extract modules file name from globals
39679 TRANSLATION_UNIT_DECL name.
39681 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39684 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39685 for long calls with APCS frame and VFP.
39687 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39689 * cfg.c (original_copy_tables_initialized_p): New function.
39690 * cfg.h (original_copy_tables_initialized_p): New decl.
39691 * cfgrtl.c (relink_block_chain): Guard the call to
39692 free_original_copy_tables with a call to
39693 original_copy_tables_initialized_p.
39694 * cgraph.h (symtab_node::native_rtl_p): New decl.
39695 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39696 (symtab_node::needed_p): Don't assert for early assembly output
39697 for __RTL functions.
39698 (cgraph_node::finalize_function): Set "force_output" for __RTL
39700 (cgraph_node::analyze): Bail out early for __RTL functions.
39701 (analyze_functions): Update assertion to support __RTL functions.
39702 (cgraph_node::expand): Bail out early for __RTL functions.
39703 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39705 * function.h (struct function): Update comment for field
39707 * gimple-expr.c: Include "tree-pass.h".
39708 (gimple_has_body_p): Return false for __RTL functions.
39709 * Makefile.in (OBJS): Add run-rtl-passes.o.
39710 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39712 (gcc::pass_manager::get_clean_slate): New accessor.
39713 * passes.c: Include "insn-addr.h".
39714 (should_skip_pass_p): Add logging. Update logic for running
39715 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39716 property-provider override so it is only done for gimple passes.
39718 (skip_pass): New function.
39719 (execute_one_pass): Call skip_pass when skipping passes.
39720 * read-md.c (md_reader::read_char): Support filtering
39721 the input to a subset of line numbers.
39722 (md_reader::md_reader): Initialize fields
39723 m_first_line and m_last_line.
39724 (md_reader::read_file_fragment): New function.
39725 * read-md.h (md_reader::read_file_fragment): New decl.
39726 (md_reader::m_first_line): New field.
39727 (md_reader::m_last_line): New field.
39728 * read-rtl-function.c (function_reader::create_function): Only
39729 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39730 curr_properties. Set DECL_INITIAL to a dummy block.
39731 (read_rtl_function_body_from_file_range): New function.
39732 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39734 * run-rtl-passes.c: New file.
39735 * run-rtl-passes.h: New file.
39737 2017-01-24 Jeff Law <law@redhat.com>
39739 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39742 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39744 PR tree-optimization/79159
39745 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39746 (record_nonwrapping_iv): Improve boundary using above function if no
39747 value range information.
39749 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39750 Martin Jambor <mjambor@suse.cz>
39752 * brig-builtins.def: New file.
39753 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39754 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39755 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39756 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39757 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39758 * builtin-types.def (BT_INT8): New.
39759 (BT_INT16): Likewise.
39760 (BT_UINT8): Likewise.
39761 (BT_UINT16): Likewise.
39762 (BT_FN_ULONG): Likewise.
39763 (BT_FN_UINT_INT): Likewise.
39764 (BT_FN_UINT_ULONG): Likewise.
39765 (BT_FN_UINT_LONG): Likewise.
39766 (BT_FN_UINT_PTR): Likewise.
39767 (BT_FN_ULONG_PTR): Likewise.
39768 (BT_FN_INT8_FLOAT): Likewise.
39769 (BT_FN_INT16_FLOAT): Likewise.
39770 (BT_FN_UINT32_FLOAT): Likewise.
39771 (BT_FN_UINT16_FLOAT): Likewise.
39772 (BT_FN_UINT8_FLOAT): Likewise.
39773 (BT_FN_UINT64_FLOAT): Likewise.
39774 (BT_FN_UINT16_UINT32): Likewise.
39775 (BT_FN_UINT32_UINT16): Likewise.
39776 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39777 (BT_FN_INT_PTR_INT): Likewise.
39778 (BT_FN_UINT_PTR_UINT): Likewise.
39779 (BT_FN_LONG_PTR_LONG): Likewise.
39780 (BT_FN_ULONG_PTR_ULONG): Likewise.
39781 (BT_FN_VOID_UINT64_UINT64): Likewise.
39782 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39783 (BT_FN_INT8_INT8_INT8): Likewise.
39784 (BT_FN_INT16_INT16_INT16): Likewise.
39785 (BT_FN_INT_INT_INT): Likewise.
39786 (BT_FN_UINT_FLOAT_UINT): Likewise.
39787 (BT_FN_FLOAT_UINT_UINT): Likewise.
39788 (BT_FN_ULONG_UINT_UINT): Likewise.
39789 (BT_FN_ULONG_UINT_PTR): Likewise.
39790 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39791 (BT_FN_UINT_UINT_UINT): Likewise.
39792 (BT_FN_VOID_UINT_PTR): Likewise.
39793 (BT_FN_UINT_UINT_PTR: Likewise.
39794 (BT_FN_UINT32_UINT64_PTR): Likewise.
39795 (BT_FN_INT_INT_UINT_UINT): Likewise.
39796 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39797 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39798 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39799 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39800 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39801 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39802 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39803 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39804 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39805 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39806 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39807 * doc/frontends.texi: List BRIG FE.
39808 * doc/install.texi (Testing): Add BRIG tesring requirements.
39809 * doc/invoke.texi (Overall Options): Mention BRIG.
39810 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39812 2017-01-24 Richard Biener <rguenther@suse.de>
39814 PR translation/79208
39815 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39817 2017-01-24 Martin Jambor <mjambor@suse.cz>
39820 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39821 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39822 and known_contexts.
39824 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39826 PR middle-end/79123
39827 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39828 casts from signed to unsigned really don't have a range.
39830 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39832 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39833 GMP_RNDx for compatiblity.
39835 2017-01-24 Martin Liska <mliska@suse.cz>
39838 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39839 that would prevent us to call alloca with -1 as argument.
39841 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39843 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39844 Avoid -Wformat-security warning.
39846 2017-01-23 Andrew Pinski <apinski@cavium.com>
39848 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39851 2017-01-23 Martin Sebor <msebor@redhat.com>
39853 PR middle-end/78703
39854 * gimple-ssa-sprintf.c (warn_level): New global.
39855 (format_integer): Use it here and throughout the rest of the file.
39856 Use the same switch to compute sign as base.
39857 (maybe_warn): New function.
39858 (format_directive): Factor out warnings into maybe_warn.
39859 Add debugging output. Use warn_level.
39860 (add_bytes): Use warn_level.
39861 (pass_sprintf_length::compute_format_length): Add debugging output.
39862 (try_substitute_return_value): Same.
39863 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39865 PR middle-end/78703
39866 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39867 (struct fmtresult, format_integer, format_floating): Adjust.
39868 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39869 (get_string_length, format_string,format_directive): Same.
39870 (pass_sprintf_length::compute_format_length): Same.
39871 (try_substitute_return_value): Simplify slightly.
39873 PR middle-end/78703
39874 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39875 (fmtresult::operator+=): Outlined.
39876 (struct fmtresult): Add ctors.
39877 (struct conversion_spec): Rename...
39878 (struct directive): ...to this. Add and remove data members.
39879 (directive::set_width, directive::set_precision): New functions.
39880 (format_percent): Use fmtresult ctor.
39881 (get_width_and_precision): Remove.
39882 (format_integer): Make naming changes. Avoid computing width and
39884 (format_floating): Same. Adjust indentation.
39885 (format_character, format_none): New functions.
39886 (format_string): Moved character handling to format_character.
39887 (format_directive): Remove arguments, change return type.
39888 (parse_directive): New function.
39889 (pass_sprintf_length::compute_format_length): Move directive
39890 parsing to parse_directive.
39892 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39894 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39895 (assign_assembler_name_if_needed): ... this.
39896 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39897 (assign_assembler_name_if_needed): ... this.
39898 (free_lang_data_in_cgraph): Adjust callers.
39899 * cgraphunit.c (cgraph_node::analyze): Likewise.
39900 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39903 2017-01-23 Richard Biener <rguenther@suse.de>
39905 PR tree-optimization/79088
39906 PR tree-optimization/79188
39907 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39908 resetting loop bounds after last path deletion. Reset loop
39909 bounds of the target loop, make code match the comments.
39910 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39911 Make sure loops need no fixups.
39913 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39915 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39916 exponent support with double type for first argument.
39917 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39918 type returned by __builtin_vec_extract_sig,
39919 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39920 functions from "vector int" to "vector unsigned int" or from
39921 "vector long long int" to "vector unsigned long long int".
39922 Changed type returned by __builtin_vec_extract_exp,
39923 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39924 functions from "vector int" to "vector unsigned int" or from
39925 "vector long long int" to "vector unsigned long long int".
39926 Changed return type of __builtin_vec_test_data_class,
39927 __builtin_vec_test_data_class_sp, and
39928 __builtin_vec_test_data_class_dp from "vector int" to
39929 "vector bool int" or from "vector long long int" to "vector bool
39930 long long int" and changed second argument type from "unsigned
39931 int" to "int". Added new overloaded function forms "vector float
39932 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39933 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39934 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39935 double, vector unsigned long long int)" and "vector double
39936 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39937 long int)". Changed return type of
39938 __builtin_scalar_test_data_class and
39939 __builtin_scalar_test_data_class_sp and
39940 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39941 int" and changed second argument from "unsigned int" to "int".
39942 Changed type returned by __builtin_scalar_test_neg,
39943 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39944 from "int" to "bool int". Added new overloaded function form
39945 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39946 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39947 exponent double-precision with floating point first argument.
39948 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39949 documentation of scalar_test_data_class, scalar_test_neg,
39950 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39951 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39952 vec_test_data_class built-in functions to reflect refinements in
39953 their type signatures.
39955 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39957 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39959 (aarch64_elf_asm_destructor): Likewise.
39961 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39963 PR rtl-optimization/78634
39964 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39965 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39966 * ifcvt.c (noce_try_cmove): Add missing cost check.
39968 PR rtl-optimization/71724
39969 * combine.c (if_then_else_cond): Look for situations where it is
39970 beneficial to undo the work of one of the recursive calls.
39972 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39974 PR tree-optimization/70754
39975 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39976 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39977 combined stmt before it if not NULL.
39978 (combine_chains): Process refs reversely and compute dominance point
39981 2017-01-23 Martin Liska <mliska@suse.cz>
39983 PR tree-optimization/79196
39984 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39985 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39987 (strlen_optimize_stmt): Call the renamed function.
39989 2017-01-23 Michael Matz <matz@suse.de>
39991 PR tree-optimization/78384
39992 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39994 2017-01-23 Richard Biener <rguenther@suse.de>
39996 PR tree-optimization/79186
39997 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39998 both incoming edges before moving an assert.
40000 2017-01-23 Martin Jambor <mjambor@suse.cz>
40002 * ipa-prop.c (load_from_param_1): Removed.
40003 (load_from_unmodified_param): Bits from load_from_param_1 put back
40005 (load_from_param): Removed.
40006 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40007 with stmt. Reverted back to use of load_from_unmodified_param.
40009 2017-01-23 Martin Jambor <mjambor@suse.cz>
40012 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40013 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40014 field a pointer to garbage collected vector, mark lattices and
40015 ipcp_orig_node with GTY((skip)).
40016 (ipa_get_param_count): Adjust to descriptors being a pointer.
40017 (ipa_get_param): Likewise.
40018 (ipa_get_type): Likewise.
40019 (ipa_get_param_move_cost): Likewise.
40020 (ipa_set_param_used): Likewise.
40021 (ipa_get_controlled_uses): Likewise.
40022 (ipa_set_controlled_uses): Likewise.
40023 (ipa_is_param_used): Likewise.
40024 (ipa_node_params_t): Move into garbage collector. New methods insert
40026 (ipa_node_params_sum): Annotate wth GTY(()).
40027 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40029 (ipa_load_from_parm_agg): Adjust declaration.
40030 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40031 * ipa-profile.c (ipa_profile): Likewise.
40032 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40033 (ipa_populate_param_decls): Make descriptors parameter garbage
40035 (ipa_dump_param): Adjust to descriptors being a pointer.
40036 (ipa_alloc_node_params): Likewise.
40037 (ipa_initialize_node_params): Likewise.
40038 (load_from_param_1): Make descriptors parameter garbage collected.
40039 (load_from_unmodified_param): Likewise.
40040 (load_from_param): Likewise.
40041 (ipa_load_from_parm_agg): Likewise.
40042 (ipa_node_params::~ipa_node_params): Removed.
40043 (ipa_free_all_node_params): Remove call to delete operator.
40044 (ipa_node_params_t::insert): New.
40045 (ipa_node_params_t::remove): Likewise.
40046 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40047 copy known_csts and known_contexts vectors.
40048 (ipa_read_node_info): Adjust to descriptors being a pointer.
40049 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40051 (ipcp_transform_function): Make descriptors variable garbage
40054 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40056 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40057 * config/i386/avx512dqintrin.h: Ditto.
40058 * config/i386/avx512fintrin.h: Ditto.
40059 * config/i386/i386.c: Handle new builtins.
40060 * config/i386/i386-builtin.def: Add new builtins.
40061 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40062 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40064 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40065 Martin Liska <mliska@suse.cz>
40067 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40068 * asan.c (asan_expand_poison_ifn): Support stores and use
40069 appropriate ASAN report function.
40070 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40071 * internal-fn.def (ASAN_POISON_USE): Declare.
40072 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40073 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40074 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40075 ASAN_POISON calls w/o LHS.
40076 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40077 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40078 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40079 * gimplify.c (asan_poison_variables): Add attribute
40080 use_after_scope_memory to variables that really needs to live
40082 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40083 having the attribute.
40085 2017-01-23 Martin Liska <mliska@suse.cz>
40087 * asan.c (create_asan_shadow_var): New function.
40088 (asan_expand_poison_ifn): Likewise.
40089 * asan.h (asan_expand_poison_ifn): New declaration.
40090 * internal-fn.c (expand_ASAN_POISON): Likewise.
40091 * internal-fn.def (ASAN_POISON): New builtin.
40092 * sanopt.c (pass_sanopt::execute): Expand
40093 asan_expand_poison_ifn.
40094 * tree-inline.c (copy_decl_for_dup_finish): Make function
40096 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40097 * tree-ssa.c (is_asan_mark_p): New function.
40098 (execute_update_addresses_taken): Rewrite local variables
40099 (identified just by use-after-scope as addressable) into SSA.
40101 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40103 * doc/install.texi (Specific): opensource.apple.com uses https
40104 now. Remove trailing slash.
40106 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40108 * README.Portability: Remove note on an Irix compatibility issue.
40110 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40112 * gcov.c (INCLUDE_ALGORITHM): Define.
40113 (INCLUDE_VECTOR): Define.
40114 No longer include <vector> and <algorithm> directly.
40116 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40118 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40120 * doc/invoke.texi (Code Gen Options): Ditto.
40122 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40125 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40127 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40129 rtl-optimization/79125
40130 * cprop.c (local_cprop_pass): Handle cases where we make an
40131 unconditional trap.
40133 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40137 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40138 read from, for big endian.
40140 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40142 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40143 register pauth builtins for LP64 only.
40145 2017-01-20 Marek Polacek <polacek@redhat.com>
40148 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40151 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40153 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40155 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40156 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40157 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40159 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40162 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40163 in big-endian mode when they are not a single duplicated value.
40165 2017-01-20 Richard Biener <rguenther@suse.de>
40167 * BASE-VER: Bump to 7.0.1.
40169 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40171 * omp-low.c (omplow_simd_context): New struct. Use it...
40172 (lower_rec_simd_input_clauses): ...here and...
40173 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40174 references to idx, lane, max_vf, is_simt.
40176 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40178 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40181 2017-01-20 Martin Jambor <mjambor@suse.cz>
40183 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40184 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40186 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40187 (GTFILES): Rename hsa.c to hsa-common.c.
40188 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40189 * hsa-dump.c: Likewise.
40190 * hsa-gen.c: Likewise.
40191 * hsa-regalloc.c: Likewise.
40192 * ipa-hsa.c: Likewise.
40193 * omp-expand.c: Likewise.
40194 * omp-low.c: Likewise.
40195 * toplev.c: Likewise.
40197 2017-01-20 Marek Polacek <polacek@redhat.com>
40200 * doc/invoke.texi: Document -Wduplicated-branches.
40201 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40202 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40203 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40204 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40205 return 0 only when not OEP_LEXICOGRAPHIC.
40206 (fold_build_cleanup_point_expr): Use the expression
40207 location when building CLEANUP_POINT_EXPR.
40208 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40209 * tree.c (add_expr): Handle error_mark_node.
40211 2017-01-20 Martin Liska <mliska@suse.cz>
40214 * tree-profile.c (init_ic_make_global_vars): Do not call
40216 (gimple_init_gcov_profiler): Likewise.
40218 2017-01-20 Martin Liska <mliska@suse.cz>
40221 * cgraph.h (maybe_create_reference): Remove argument and
40223 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40225 * ipa-cp.c (create_specialized_node): Likewise.
40226 * symtab.c (symtab_node::maybe_create_reference): Handle
40227 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40229 2017-01-20 Martin Liska <mliska@suse.cz>
40231 * read-rtl-function.c (function_reader::create_function): Use
40232 build_decl instread of build_decl_stat.
40234 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40236 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40237 * config/i386/avx512dqintrin.h: Ditto.
40238 * config/i386/avx512fintrin.h: Ditto.
40239 * config/i386/i386-builtin-types.def: Add new types.
40240 * config/i386/i386.c: Handle new types.
40241 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40242 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40243 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40244 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40245 (__builtin_ia32_kshiftridi): New.
40246 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40248 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40252 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40253 define to rs6000_init_stack_protect_guard.
40254 (rs6000_init_stack_protect_guard): New function.
40256 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40257 Yunqiang Su <yunqiang.su@imgtec.com>
40259 * config.gcc (supported_defaults): Add madd4.
40260 (with_madd4): Add validation.
40261 (all_defaults): Add madd4.
40262 * config/mips/mips.opt (mmadd4): New option.
40263 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40265 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40267 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40268 (ISA_HAS_FUSED_MADD4): Likewise.
40269 * doc/invoke.texi (-mmadd4): Document the new option.
40270 * doc/install.texi (--with-madd4): Document the new option.
40272 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40274 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40275 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40276 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40277 (aarch64_init_pauth_hint_builtins): New.
40278 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40279 (aarch64_expand_builtin): Expand new builtins.
40281 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40283 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40284 * combine-stack-adj.c (no_unhandled_cfa): Handle
40285 REG_CFA_TOGGLE_RA_MANGLE.
40286 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40287 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40288 info for return address signing.
40289 (aarch64_expand_epilogue): Likewise.
40291 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40293 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40294 * config/aarch64/aarch64-protos.h
40295 (aarch64_return_address_signing_enabled): New declaration.
40296 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40298 (aarch64_expand_prologue): Sign return address before it's pushed onto
40300 (aarch64_expand_epilogue): Authenticate return address fetched from
40302 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40303 (aarch64_attributes): New function attributes for "sign-return-address".
40304 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40305 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40306 ("*do_return"): Generate combined instructions according to key index.
40307 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40308 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40310 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40311 * config/aarch64/aarch64.opt (msign-return-address=): New.
40312 * doc/extend.texi (AArch64 Function Attributes): Documents
40313 "sign-return-address=".
40314 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40316 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40318 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40319 overall option summary.
40321 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40323 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40324 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40325 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40326 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40328 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40330 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40331 -mpower9-minmax by default for -mcpu=power9.
40332 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40333 128-bit floating point.
40335 2017-01-20 Alan Modra <amodra@gmail.com>
40337 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40338 optimizing for size.
40340 2017-01-20 Alan Modra <amodra@gmail.com>
40343 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40344 for strcmp and strncmp from corresponding builtin decl.
40346 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40348 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40349 instead of i386/rtems-64.h.
40350 * config/i386/rtems-64.h: Remove.
40352 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40356 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40358 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40360 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40362 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40363 Change int to HOST_WIDE_INT.
40364 * config/aarch64/aarch64-protos.h
40365 (aarch64_simd_gen_const_vector_dup): Likewise.
40366 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40368 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40370 * langhooks-def.h (lhd_type_for_size): New decl.
40371 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40372 * langhooks.c (lhd_type_for_size): New function, taken from
40375 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40377 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40378 define_bypass for CR latency.
40379 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40380 (power9-alu2): Add define_bypass for CR latency.
40382 (power9-mul): Update insn latency.
40383 (power9-mul-compare): Update insn latency, bypass latency and remove
40386 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40388 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40390 * config/aarch64/aarch64.md
40391 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40392 aarch64_nopcrelative_literal_loads.
40393 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40395 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40397 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40398 TARGET_LOONGSON_3A.
40399 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40401 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40404 * config.gcc (supported_defaults): Add lxc1-sxc1.
40405 (with_lxc1_sxc1): Add validation.
40406 (all_defaults): Add lxc1-sxc1.
40407 * config/mips/mips.opt (mlxc1-sxc1): New option.
40408 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40410 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40411 __mips_no_lxc1_sxc1.
40412 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40413 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40414 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40416 2017-01-19 Richard Biener <rguenther@suse.de>
40418 PR tree-optimization/72488
40419 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40420 sure to restore SSA info.
40421 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40423 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40425 PR rtl-optimization/79121
40426 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40427 of the inner type when shifting an extended value.
40429 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40432 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40433 interposable aliases.
40435 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40438 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40439 Use the evmergelohi instruction.
40440 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40441 (mov_sitf_e500_subreg8_2_be): Likewise.
40442 (mov_sitf_e500_subreg12_2_le): Likewise.
40443 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40444 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40445 (mov_sitf_e500_subreg8_2_le): Likewise.
40446 (mov_sitf_e500_subreg12_2_be): Likewise.
40448 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40450 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40451 attribute from vecsimple to vecperm.
40452 (altivec_vbpermq2): Likewise.
40454 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40457 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40459 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40460 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40461 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40462 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40463 case where N arg is SIZE_MAX.
40464 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40465 (cmpstrsi): Add pattern.
40467 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40469 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40470 __builtin_vec_revb builtins.
40471 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40472 built-in functions to support generation of the ISA 3.0 XXBR<x>
40473 vector byte reverse instructions.
40474 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40475 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40476 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40477 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40478 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40479 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40480 (P9V_BUILTIN_VEC_REVB): Likewise.
40481 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40482 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40483 (p9_xxbrq_v16qi): Likewise.
40484 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40485 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40486 (p9_xxbrh_v8hi): Likewise.
40487 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40488 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40489 vec_revb built-in functions.
40491 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40493 PR rtl-optimization/78952
40494 * config/i386/i386.md (any_extract): New code iterator.
40495 (*insvqi_2): Use any_extract for source operand.
40496 (*insvqi_3): Use any_shiftrt for source operand.
40498 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40500 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40502 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40504 2017-01-18 Matthias Klose <doko@ubuntu.com>
40506 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40508 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40510 * config/rs6000/altivec.h (vec_bperm): Change #define.
40511 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40512 (altivec_vbpermq2): New define_insn.
40513 (altivec_vbpermd): Likewise.
40514 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40515 function interface.
40516 (VBPERMD): Likewise.
40517 (VBPERM): New polymorphic function interface.
40518 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40519 Add entries for P9V_BUILTIN_VEC_VBPERM.
40520 * doc/extend.texi: Add interfaces for vec_bperm.
40522 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40524 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40525 first letter of error messages.
40526 (s390_resolve_overloaded_builtin): Likewise.
40527 * config/s390/s390.c (s390_expand_builtin): Likewise.
40528 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40529 (s390_valid_target_attribute_inner_p): Likewise.
40530 * config/s390/s390.md ("tabort"): Likewise.
40532 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40534 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40535 (ISA_AVOID_DIV_HILO): New macro.
40536 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40537 (ISA_HAS_DDIV): Likewise.
40539 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40541 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40543 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40545 * doc/invoke.texi (fabi-version): Spelling fix.
40547 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40550 * doc/invoke.texi (fabi-version): Mention mangling fix for
40553 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40556 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40558 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40561 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40562 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40564 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40566 (stack_protect_test): Ditto.
40567 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40568 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40570 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40571 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40572 -mstack-protector-guard-offset=.
40573 (RS/6000 and PowerPC Options): Ditto.
40575 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40577 * config/i386/i386.h (MASK_CLASS_P): New define.
40578 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40579 there are no registers from different register sets also when
40580 mask registers are used. Update function comment.
40581 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40582 to (*k/*r) and (*k/*km) alternatives.
40584 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40586 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40587 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40588 (EH_RETURN_HANDLER_RTX): New define.
40589 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40590 Force frame pointer in EH return functions.
40591 (aarch64_expand_epilogue): Add barrier for eh_return.
40592 (aarch64_final_eh_return_addr): Remove.
40593 (aarch64_eh_return_handler_rtx): New function.
40594 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40596 (aarch64_eh_return_handler_rtx): New prototype.
40598 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40600 * config/rs6000/altivec.h (vec_rlmi): New #define.
40601 (vec_vrlnm): Likewise.
40602 (vec_rlnm): Likewise.
40603 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40604 (UNSPEC_VRLNM): Likewise.
40605 (VIlong): New mode iterator.
40606 (altivec_vrl<VI_char>mi): New define_insn.
40607 (altivec_vrl<VI_char>nm): Likewise.
40608 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40610 (VRLDNM): Likewise.
40611 (RLNM): New polymorphic function entry.
40612 (VRLWMI): New monomorphic function entry.
40613 (VRLDMI): Likewise.
40614 (RLMI): New polymorphic function entry.
40615 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40616 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40617 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40620 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40623 * dwarf2out.c (field_byte_offset): Restore the
40624 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40625 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40626 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40629 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40632 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40634 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40637 * dwarf2out.c (add_data_member_location_attribute): For constant
40638 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40639 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40640 DW_AT_byte_size attributes.
40642 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40644 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40645 after forcing to constant memory when the code model is medium.
40647 2017-01-17 Julia Koval <julia.koval@intel.com>
40650 * config/i386/avx512fintrin.h
40651 (_mm512_i32gather_ps): Change __addr type to void const*.
40652 (_mm512_mask_i32gather_ps): Ditto.
40653 (_mm512_i32gather_pd): Ditto.
40654 (_mm512_mask_i32gather_pd): Ditto.
40655 (_mm512_i64gather_ps): Ditto.
40656 (_mm512_mask_i64gather_ps): Ditto.
40657 (_mm512_i64gather_pd): Ditto.
40658 (_mm512_mask_i64gather_pd): Ditto.
40659 (_mm512_i32gather_epi32): Ditto.
40660 (_mm512_mask_i32gather_epi32): Ditto.
40661 (_mm512_i32gather_epi64): Ditto.
40662 (_mm512_mask_i32gather_epi64): Ditto.
40663 (_mm512_i64gather_epi32): Ditto.
40664 (_mm512_mask_i64gather_epi32): Ditto.
40665 (_mm512_i64gather_epi64): Ditto.
40666 (_mm512_mask_i64gather_epi64): Ditto.
40667 (_mm512_i32scatter_ps): Change __addr type to void*.
40668 (_mm512_mask_i32scatter_ps): Ditto.
40669 (_mm512_i32scatter_pd): Ditto.
40670 (_mm512_mask_i32scatter_pd): Ditto.
40671 (_mm512_i64scatter_ps): Ditto.
40672 (_mm512_mask_i64scatter_ps): Ditto.
40673 (_mm512_i64scatter_pd): Ditto.
40674 (_mm512_mask_i64scatter_pd): Ditto.
40675 (_mm512_i32scatter_epi32): Ditto.
40676 (_mm512_mask_i32scatter_epi32): Ditto.
40677 (_mm512_i32scatter_epi64): Ditto.
40678 (_mm512_mask_i32scatter_epi64): Ditto.
40679 (_mm512_i64scatter_epi32): Ditto.
40680 (_mm512_mask_i64scatter_epi32): Ditto.
40681 (_mm512_i64scatter_epi64): Ditto.
40682 (_mm512_mask_i64scatter_epi64): Ditto.
40683 * config/i386/avx512pfintrin.h
40684 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40685 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40686 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40687 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40688 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40689 (_mm512_prefetch_i32scatter_ps): Ditto.
40690 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40691 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40692 (_mm512_prefetch_i64scatter_pd): Ditto.
40693 (_mm512_prefetch_i64scatter_ps): Ditto.
40694 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40695 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40696 * config/i386/avx512vlintrin.h
40697 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40698 (_mm_mmask_i32gather_ps): Ditto.
40699 (_mm256_mmask_i32gather_pd): Ditto.
40700 (_mm_mmask_i32gather_pd): Ditto.
40701 (_mm256_mmask_i64gather_ps): Ditto.
40702 (_mm_mmask_i64gather_ps): Ditto.
40703 (_mm256_mmask_i64gather_pd): Ditto.
40704 (_mm_mmask_i64gather_pd): Ditto.
40705 (_mm256_mmask_i32gather_epi32): Ditto.
40706 (_mm_mmask_i32gather_epi32): Ditto.
40707 (_mm256_mmask_i32gather_epi64): Ditto.
40708 (_mm_mmask_i32gather_epi64): Ditto.
40709 (_mm256_mmask_i64gather_epi32): Ditto.
40710 (_mm_mmask_i64gather_epi32): Ditto.
40711 (_mm256_mmask_i64gather_epi64): Ditto.
40712 (_mm_mmask_i64gather_epi64): Ditto.
40713 (_mm256_i32scatter_ps): Change __addr type to void*.
40714 (_mm256_mask_i32scatter_ps): Ditto.
40715 (_mm_i32scatter_ps): Ditto.
40716 (_mm_mask_i32scatter_ps): Ditto.
40717 (_mm256_i32scatter_pd): Ditto.
40718 (_mm256_mask_i32scatter_pd): Ditto.
40719 (_mm_i32scatter_pd): Ditto.
40720 (_mm_mask_i32scatter_pd): Ditto.
40721 (_mm256_i64scatter_ps): Ditto.
40722 (_mm256_mask_i64scatter_ps): Ditto.
40723 (_mm_i64scatter_ps): Ditto.
40724 (_mm_mask_i64scatter_ps): Ditto.
40725 (_mm256_i64scatter_pd): Ditto.
40726 (_mm256_mask_i64scatter_pd): Ditto.
40727 (_mm_i64scatter_pd): Ditto.
40728 (_mm_mask_i64scatter_pd): Ditto.
40729 (_mm256_i32scatter_epi32): Ditto.
40730 (_mm256_mask_i32scatter_epi32): Ditto.
40731 (_mm_i32scatter_epi32): Ditto.
40732 (_mm_mask_i32scatter_epi32): Ditto.
40733 (_mm256_i32scatter_epi64): Ditto.
40734 (_mm256_mask_i32scatter_epi64): Ditto.
40735 (_mm_i32scatter_epi64): Ditto.
40736 (_mm_mask_i32scatter_epi64): Ditto.
40737 (_mm256_i64scatter_epi32): Ditto.
40738 (_mm256_mask_i64scatter_epi32): Ditto.
40739 (_mm_i64scatter_epi32): Ditto.
40740 (_mm_mask_i64scatter_epi32): Ditto.
40741 (_mm256_i64scatter_epi64): Ditto.
40742 (_mm256_mask_i64scatter_epi64): Ditto.
40743 (_mm_i64scatter_epi64): Ditto.
40744 (_mm_mask_i64scatter_epi64): Ditto.
40745 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40746 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40747 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40748 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40749 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40750 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40751 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40752 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40753 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40754 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40755 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40756 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40757 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40758 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40759 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40760 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40761 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40762 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40763 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40764 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40765 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40766 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40767 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40768 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40769 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40770 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40771 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40772 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40773 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40774 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40775 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40776 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40777 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40778 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40779 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40780 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40781 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40782 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40783 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40784 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40785 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40786 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40787 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40788 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40789 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40790 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40791 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40792 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40793 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40794 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40795 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40796 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40797 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40798 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40799 definitions accordingly.
40801 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40802 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40805 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40808 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40811 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40812 hard regs for inner regno.
40814 2017-01-17 Martin Liska <mliska@suse.cz>
40817 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40818 assumption and add comment.
40820 2017-01-17 Nathan Sidwell <nathan@acm.org>
40822 * ipa-visibility.c (localize_node): New function, broken out of ...
40823 (function_and_variable_visibility): ... here. Call it.
40825 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40827 PR middle-end/77445
40828 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40829 correctly set frequency of oudgoing edge.
40830 (duplicate_thread_path): Fix profile updating.
40832 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40835 * configure.ac: Add GCC_BASE_VER.
40836 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40837 version from BASE-VER file.
40838 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40839 (gcc.o): Depend on $(BASEVER).
40840 * common.opt (dumpfullversion): New option.
40841 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40842 * doc/invoke.texi: Document -dumpfullversion.
40843 * doc/install.texi: Document --with-gcc-major-version-only.
40844 * configure: Regenerated.
40846 2017-01-17 Richard Biener <rguenther@suse.de>
40848 PR tree-optimization/71433
40849 * tree-vrp.c (register_new_assert_for): Merge same asserts
40850 on all incoming edges.
40851 (process_assert_insertions_for): Handle insertions at the
40854 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40856 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40857 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40859 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40862 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40865 2017-01-17 Alan Modra <amodra@gmail.com>
40868 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40869 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40870 symbolic stack limit when pic.
40872 2017-01-16 Martin Sebor <msebor@redhat.com>
40874 PR tree-optimization/78608
40875 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40877 2017-01-16 Jeff Law <law@redhat.com>
40880 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40881 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40882 for several include directories that may be relative to sysroot.
40883 * config/i386/x-mingw32 (gplus_includedir): Define.
40884 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40885 (native_system_includedir): Likewise.
40886 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40887 override if TARGET_SYSTEM_ROOT is defined.
40888 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40890 PR tree-optimization/79090
40891 PR tree-optimization/33562
40892 PR tree-optimization/61912
40893 PR tree-optimization/77485
40894 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40895 and computed trims into the dump file.
40897 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40899 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40901 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40904 * gimplify.c (gimplify_init_constructor): If want_value and
40905 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40909 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40910 sequence. Formatting fixes.
40911 (doloop_optimize): Formatting fixes.
40914 * gcc.c (debug_level_greater_than_spec_func): New function.
40915 (static_spec_functions): Add debug-level-gt spec function.
40916 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40918 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40919 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40920 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40921 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40922 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40923 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40925 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40927 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40928 QImode fixups to general and mask registers only.
40930 2017-01-16 Carl Love <cel@us.ibm.com>
40932 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40933 for built-in functions
40934 vector signed char vec_nabs (vector signed char)
40935 vector signed short vec_nabs (vector signed short)
40936 vector signed int vec_nabs (vector signed int)
40937 vector signed long long vec_nabs (vector signed long long)
40938 vector float vec_nabs (vector float)
40939 vector double vec_nabs (vector double)
40940 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40942 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40943 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40944 * doc/extend.texi: Update the documentation file for the new built-in
40947 2017-01-16 Martin Sebor <msebor@redhat.com>
40949 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40952 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40954 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40955 UNSPEC_VSX__XXSPLTD to require special splat handling.
40957 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40960 * system.h: Poison strndup.
40962 2017-01-16 Alan Modra <amodra@gmail.com>
40965 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40968 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40970 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40972 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40974 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40975 call recog here. Assert that INSN_CODE (insn) is non-negative.
40977 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40980 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40982 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40983 in the currently scheduled RTL fragment.
40985 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40987 PR rtl-optimization/78751
40988 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40991 2017-01-14 Jeff Law <law@redhat.com>
40993 PR tree-optimization/79090
40994 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40995 variable length stores.
40996 (compute_trims): Delete dead assignment to *trim_tail.
40997 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41000 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41002 PR rtl-optimization/78626
41003 PR rtl-optimization/78727
41004 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41005 of a block, and split such blocks after everything else is finished.
41007 2017-01-14 Alan Modra <amodra@gmail.com>
41010 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41011 target legitimate_combined_insn.
41012 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41013 (rs6000_legitimate_combined_insn): New function.
41014 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41016 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41017 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41018 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41020 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41022 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41024 2017-01-13 Jeff Law <law@redhat.com>
41026 PR tree-optimization/33562
41027 PR tree-optimization/61912
41028 PR tree-optimization/77485
41029 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41031 (delete_dead_assignment): Likewise.
41032 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41033 statement to delete_dead_call and delete_dead_assignment.
41035 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41038 * substring-locations.c (format_warning_va): Strengthen case 1 so
41039 that both endpoints of the substring must be within the format
41040 range for just the substring to be printed.
41042 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41044 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41045 * config/i386/i386.c (ix86_target_string): Add missing options
41046 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41047 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41048 flags_other and ix86_target_other to flags2_other. Display unknown
41050 (ix86_valid_target_attribute_inner_p): Add missing options and
41051 reorder options by implied ISAs, as in ix86_target_string.
41053 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41055 * hash-table.h (hash_table::too_empty_p): New function.
41056 (hash_table::expand): Use it.
41057 (hash_table::traverse): Likewise.
41058 (hash_table::empty_slot): Use sizeof (value_type) instead of
41059 sizeof (PTR) to convert bytes to elements. Shrink the table
41060 if the current size is excessive for the current number of
41063 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41065 * ira-costs.c (record_reg_classes): Break from the inner loop
41066 early once alt_fail is known to be true. Update outer loop
41067 handling accordingly.
41069 2017-01-13 Jeff Law <law@redhat.com>
41071 * tree-ssa-dse.c (decrement_count): New function.
41072 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41073 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41074 when we know the partially dead statement is a mem* function.
41076 PR tree-optimization/61912
41077 PR tree-optimization/77485
41078 * tree-ssa-dse.c: Include expr.h.
41079 (maybe_trim_constructor_store): New function.
41080 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41082 PR tree-optimization/33562
41083 PR tree-optimization/61912
41084 PR tree-optimization/77485
41085 * doc/invoke.texi: Document new dse-max-object-size param.
41086 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41087 * tree-ssa-dse.c: Include params.h.
41088 (dse_store_status): New enum.
41089 (initialize_ao_ref_for_dse): New, partially extracted from
41091 (valid_ao_ref_for_dse, normalize_ref): New.
41092 (setup_live_bytes_from_ref, compute_trims): Likewise.
41093 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41094 (maybe_trim_partially_dead_store): Likewise.
41095 (maybe_trim_complex_store): Likewise.
41096 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41097 Track what bytes live from the original store. Return tri-state
41098 for dead, partially dead or live.
41099 (dse_dom_walker): Add constructor, destructor and new private members.
41100 (delete_dead_call, delete_dead_assignment): New extracted from
41102 (dse_optimize_stmt): Make a member of dse_dom_walker.
41103 Use initialize_ao_ref_for_dse.
41105 PR tree-optimization/33562
41106 PR tree-optimization/61912
41107 PR tree-optimization/77485
41108 * sbitmap.h (bitmap_count_bits): Prototype.
41109 (bitmap_clear_range, bitmap_set_range): Likewise.
41110 * sbitmap.c (bitmap_clear_range): New function.
41111 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41113 2017-01-13 Martin Liska <mliska@suse.cz>
41116 * function.c (set_cfun): Add new argument force.
41117 * function.h (set_cfun): Likewise.
41118 * ipa-inline-transform.c (inline_call): Use the function when
41119 strict alising from is dropped for function we inline to.
41121 2017-01-13 Richard Biener <rguenther@suse.de>
41123 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41124 for dumping GIMPLE INTEGER_CSTs.
41126 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41128 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41129 to 201112L since C++17.
41131 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41134 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41135 if -fsanitize=kernel-address is present.
41137 2017-01-13 Richard Biener <rguenther@suse.de>
41139 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41140 as _Literal ( type ) number in case usual suffixes do not
41141 preserve all information.
41143 2017-01-13 Richard Biener <rguenther@suse.de>
41145 PR tree-optimization/77283
41146 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41147 and ssa-iterators.h.
41148 (is_feasible_trace): Implement a cost model based on joiner
41151 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41154 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41155 char or short to __float128/_Float128 directly.
41157 2017-01-12 Martin Sebor <msebor@redhat.com>
41159 to -Wformat-overflow.
41160 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41161 (min_bytes_remaining): Same.
41162 (get_string_length): Same.
41163 (format_string): Same.
41164 (format_directive): Same.
41166 (pass_sprintf_length::handle_gimple_call): Same.
41168 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41170 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41171 info.nowrite calls with no lhs that can't throw. Return bool
41172 whether gsi_remove has been called or not.
41173 (pass_sprintf_length::handle_gimple_call): Return bool whether
41174 try_substitute_return_value called gsi_remove. Formatting fix.
41175 (pass_sprintf_length::execute): Don't use gsi_remove if
41176 handle_gimple_call returned true.
41179 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41180 be removed due to side-effects, don't remove following barrier nor
41181 turn the successor edge into fallthru edge.
41183 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41186 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41187 element-reversing loads and stores as not swappable.
41189 2017-01-12 Nathan Sidwell <nathan@acm.org>
41190 Nicolai Stange <nicstange@gmail.com>
41192 * combine.c (try_combine): Don't ignore result of overlap checking
41193 loop. Combine overlap & asm check into single loop.
41195 2017-01-12 Richard Biener <rguenther@suse.de>
41197 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41198 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41200 2017-01-12 Richard Biener <rguenther@suse.de>
41202 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41203 and TS_TARGET_OPTION directly derive from TS_BASE.
41204 * tree-core.h (tree_optimization_option): Derive from tree_base.
41205 (tree_target_option): Likewise.
41207 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41209 * config/i386/i386.c (memory_address_length): Increase len
41210 only when rip_relative_addr_p returns false.
41212 2017-01-11 Julia Koval <julia.koval@intel.com>
41214 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41215 (OPTION_MASK_ISA_SGX_SET): New.
41216 (ix86_handle_option): Handle OPT_msgx.
41217 * config.gcc: Added sgxintrin.h.
41218 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41219 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41220 * config/i386/i386.c (ix86_target_string): Add -msgx.
41222 (ix86_option_override_internal): Handle new options.
41223 (ix86_valid_target_attribute_inner_p): Add sgx.
41224 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41225 * config/i386/i386.opt: Add msgx.
41226 * config/i386/sgxintrin.h: New file.
41227 * config/i386/x86intrin.h: Add sgxintrin.h.
41229 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41232 * fold-const.c (maybe_nonzero_address): Return 1 for function
41234 (tree_single_nonzero_warnv_p): Don't handle function local objects
41238 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41241 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41244 * opts.c: Include "spellcheck.h"
41245 (struct string_fragment): New struct.
41246 (struct edit_distance_traits<const string_fragment &>): New
41248 (get_closest_sanitizer_option): New function.
41249 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41251 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41253 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41255 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41256 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41257 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41258 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41259 for initial die_offset if dwarf_split_debug_info.
41260 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41261 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41263 (output_skeleton_debug_sections): Formatting fix. Use
41264 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41265 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41267 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41269 * config/arm/cortex-a53.md: Add bypasses for
41270 cortex_a53_r2f_cvt.
41271 (cortex_a53_r2f): Only use for transfers.
41272 (cortex_a53_f2r): Likewise.
41273 (cortex_a53_r2f_cvt): Add reservation for conversions.
41274 (cortex_a53_f2r_cvt): Likewise.
41276 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41278 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41279 to all inlined functions, change static to extern.
41281 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41284 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41286 (arm_assemble_integer): Likewise.
41288 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41290 * config.gcc: Use new awk script to check CPU, FPU and architecture
41291 parameters for --with-... options.
41292 * config/arm/parsecpu.awk: New file
41293 * config/arm/arm-cpus.in: New file.
41294 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41296 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41298 * config/arm/t-arm: Update dependency rules.
41299 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41300 of processing .def files.
41301 * config/arm/genopt.sh: Deleted.
41302 * config/arm/gentune.sh: Deleted.
41303 * config/arm/arm-cores.def: Deleted.
41304 * config/arm/arm-arches.def: Deleted.
41305 * config/arm/arm-fpus.def: Deleted.
41306 * config/arm/arm-tune.md: Regenerated.
41307 * config/arm/arm-tables.opt: Regenerated.
41308 * config/arm/arm-cpu.h: New generated file.
41309 * config/arm/arm-cpu-data.h: New generated file.
41310 * config/arm/arm-cpu-cdata.h: New generated file.
41312 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41315 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41317 (input_varpool_node): Unpack dynamically_initialized bit.
41319 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41321 PR rtl-optimization/79032
41322 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41323 the alignment of the adjusted memory reference against that of MODE,
41324 instead of the alignment of the original memory reference.
41326 2017-01-11 Martin Jambor <mjambor@suse.cz>
41328 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41330 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41331 decorated functions.
41333 2017-01-11 Richard Biener <rguenther@suse.de>
41335 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41336 set range/nonnull info for PHI results. Do not set it on
41337 stmts marked for removal.
41339 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41341 * expr.c (store_field): In the bitfield case, fetch the return value
41342 from the registers before applying a single big-endian adjustment.
41343 Always do a final load for a BLKmode value not larger than a word.
41345 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41348 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41349 that we correctly handle column numbers greater than
41350 LINE_MAP_MAX_COLUMN_NUMBER.
41352 2017-01-10 Martin Sebor <msebor@redhat.com>
41354 PR middle-end/78245
41355 * gimple-ssa-sprintf.c (get_destination_size): Call
41356 {init,fini}object_sizes.
41357 * tree-object-size.c (addr_object_size): Adjust.
41358 (pass_through_call): Adjust.
41359 (pass_object_sizes::execute): Adjust.
41360 * tree-object-size.h (fini_object_sizes): Declare.
41362 2017-01-10 Martin Sebor <msebor@redhat.com>
41364 PR tree-optimization/78775
41365 * builtins.c (get_size_range): Move...
41366 * calls.c: ...to here.
41367 (alloc_max_size): Accept zero argument.
41368 (operand_signed_p): Remove.
41369 (maybe_warn_alloc_args_overflow): Call get_size_range.
41370 * calls.h (get_size_range): Declare.
41372 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41374 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41375 from TI's devices.csv file as of September 2016.
41376 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41378 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41380 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41381 * doc/invoke.texi: Likewise.
41382 * doc/md.texi: Likewise.
41383 * doc/objc.texi: Likewise.
41385 2017-01-10 Joshua Conner <joshconner@google.com>
41387 * config/arm/fuchsia-elf.h: New file.
41388 * config/fuchsia.h: New file.
41389 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41390 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41392 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41394 2016-01-10 Richard Biener <rguenther@suse.de>
41396 PR tree-optimization/79034
41397 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41398 Propagate out degenerate PHIs in the joiner.
41400 2017-01-10 Martin Liska <mliska@suse.cz>
41402 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41403 (sort_congruence_classes_by_decl_uid): Likewise.
41404 (sort_congruence_class_groups_by_decl_uid): Likewise.
41405 (sem_item_optimizer::merge_classes): Sort class, groups in these
41406 classes and members in the groups by DECL_UID of declarations.
41407 This would make merge operations stable.
41409 2017-01-10 Martin Liska <mliska@suse.cz>
41411 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41412 usage of m_classes_vec.
41413 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41414 (sem_item_optimizer::get_group_by_hash): Likewise.
41415 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41416 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41417 (sem_item_optimizer::verify_classes): Likewise.
41418 (sem_item_optimizer::process_cong_reduction): Likewise.
41419 (sem_item_optimizer::dump_cong_classes): Likewise.
41420 (sem_item_optimizer::merge_classes): Likewise.
41421 * ipa-icf.h (congruence_class_hash): Rename from
41422 congruence_class_group_hash. Remove declaration of m_classes_vec.
41424 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41426 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41427 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41428 * config.gcc: Add avx512vpopcntdqintrin.h.
41429 * config/i386/avx512vpopcntdqintrin.h: New.
41430 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41431 * config/i386/i386-builtin-types.def: Add new types.
41432 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41433 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41434 __builtin_ia32_vpopcountq_v8di_mask): New.
41435 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41436 __AVX512VPOPCNTDQ__.
41437 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41438 (PTA_AVX512VPOPCNTDQ): Define.
41439 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41440 TARGET_AVX512VPOPCNTDQ_P): Define.
41441 * config/i386/i386.opt: Add mavx512vpopcntdq.
41442 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41443 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41445 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41447 PR middle-end/77484
41448 * predict.def (PRED_CALL): Set to 67.
41450 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41452 * expr.c (store_field): In the bitfield case, if the value comes from
41453 a function call and is of an aggregate type returned in registers, do
41454 not modify the field mode; extract the value in all cases if the mode
41455 is BLKmode and the size is not larger than a word.
41457 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41460 * config/i386/cpuid.h: Fix undefined behavior.
41462 2017-01-04 Jeff Law <law@redhat.com>
41464 PR tree-optimization/79007
41465 PR tree-optimization/67955
41466 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41467 conservative for pt.null when flag_non_call_exceptions is on.
41469 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41471 PR translation/79019
41472 PR translation/79020
41473 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41474 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41475 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41477 * config/avr/avr.opt (maccumulate-args): Likewise.
41478 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41479 * common.opt (freport-bug): Likewise.
41480 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41481 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41482 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41483 translatable string.
41484 * config/i386/i386.c (function_value_32): Likewise.
41485 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41486 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41488 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41489 * common/config/msp430/msp430-common.c (msp430_handle_option):
41491 * symtab.c (symtab_node::verify_base): Likewise.
41492 * opts.c (set_debug_level): Likewise.
41493 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41494 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41495 missing whitespace to translatable strings.
41496 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41497 * config/sh/superh.h: Likewise.
41498 * config/i386/xopintrin.h: Likewise.
41499 * config/i386/znver1.md: Likewise.
41500 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41501 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41502 * double-int.h (struct double_int): Likewise.
41503 * double-int.c (div_and_round_double): Likewise.
41504 * wide-int.cc: Likewise.
41505 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41506 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41507 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41508 (crossjumps_occurred): ... this.
41509 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41512 PR tree-optimization/78899
41513 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41514 returning bool return struct loop *, NULL for failure and the new
41516 (versionable_outer_loop_p): Don't version outer loop if it has
41517 dont_vectorized bit set.
41518 (tree_if_conversion): When versioning outer loop, ensure
41519 tree_if_conversion is performed also on the inner loop of the
41520 non-vectorizable outer loop copy.
41521 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41522 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41523 prevent vectorization of it.
41524 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41525 the outer loop vectorization of the non-scalar version is attempted
41526 before vectorization of the inner loop in scalar version. If
41527 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41528 vectorization of its inner loop.
41529 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41530 has 2 inner loops, rename also on edges from bb whose single pred
41531 is outer_loop->header. Fix typo in function comment.
41533 2017-01-09 Martin Sebor <msebor@redhat.com>
41536 * asan.c (asan_emit_stack_protection): Increase local buffer size
41537 to avoid snprintf truncation warning.
41539 2017-01-09 Andrew Pinski <apinski@cavium.com>
41541 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41542 to reference thunderx2t99 for the tuning structure
41543 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41545 (thunderx2t99_extra_costs): This.
41546 * config/aarch64/aarch64-tune.md: Regenerate.
41547 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41548 (vulcan_addrcost_table): This.
41549 (vulcan_regmove_cost): Rename to ...
41550 (thunderx2t99_regmove_cost): This.
41551 (vulcan_vector_cost): Rename to ...
41552 (thunderx2t99_vector_cost): this.
41553 (vulcan_branch_cost): Rename to ...
41554 (thunderx2t99_branch_cost): This.
41555 (vulcan_tunings): Rename to ...
41556 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41557 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41559 2017-01-09 Martin Jambor <mjambor@suse.cz>
41563 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41564 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41565 (propagate_vr_accross_jump_function): Use the above function for all
41566 value range computations for pass-through jump functions and type
41567 converasion from explicit value range values.
41568 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41569 parameters from TYPE_ARG_TYPES.
41570 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41571 (ipa_write_node_info): Stream type of the actual argument.
41572 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41574 2017-01-09 Martin Liska <mliska@suse.cz>
41577 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41578 (lookup_compiler): Do not show error message with have_E.
41580 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41582 PR tree-optimization/78938
41583 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41584 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41585 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41586 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41589 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41591 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41594 2017-01-09 Richard Biener <rguenther@suse.de>
41596 PR tree-optimization/78997
41597 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41598 name condition properly.
41600 2017-01-09 Richard Biener <rguenther@suse.de>
41603 * dwarf2out.c (is_cxx): New overload with context.
41604 (is_naming_typedef_decl): Use it.
41606 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41608 * invoke.texi (Option Summary): Correct spacing in option lists
41609 and add line breaks to fix over-long lines.
41611 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41613 PR middle-end/17660
41615 * extend.texi (Common Variable Attributes): Add xref to GCC
41616 Internals manual to explain mode attribute keywords.
41618 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41621 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41622 and Preprocessor Options.
41623 (Options for Linking): Document -pthread here....
41624 (RS/6000 and PowerPC Options): ...not here.
41625 (Solaris 2 Options): ...or here.
41626 * doc/cppopts.texi: Document -pthread.
41628 2017-01-08 Martin Sebor <msebor@redhat.com>
41630 PR middle-end/77708
41631 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41632 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41633 New member functions.
41634 (format_directive): Used them.
41636 (pass_sprintf_length::handle_gimple_call): Same.
41637 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41638 to avoid truncation for any argument.
41639 (extract_affine_mul): Same.
41640 * tree.c (get_file_function_name): Same.
41642 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41644 PR middle-end/77484
41645 * predict.def (PRED_INDIR_CALL): Set to 86.
41647 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41649 PR preprocessor/54124
41650 * doc/cppopts.texi: Reformat -d subtable to list the full name
41651 of the options. Add cross-reference to the docs for the general
41652 compiler -d options.
41653 * doc/invoke.texi (Developer Options): Add cross-reference to the
41654 preprocessor-specific -d option documentation.
41656 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41658 PR preprocessor/13498
41659 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41660 redudant material, and reflect new command-line options.
41661 (System Headers): Likewise.
41663 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41665 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41666 -isystem, and -idirafter. Copy-edit.
41667 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41668 default for -ftrack-macro-expansion. Delete obsolete and
41669 badly-formatted implementation details about -fdebug-cpp output.
41670 * doc/cppwarnopts.texi: Copy-edit.
41672 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41675 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41676 that the transition from a max line width >= 1<<10 to narrower
41677 lines works correctly.
41679 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41681 * doc/options.texi (PerFunction): New.
41682 * opt-functions.awk (switch_flags): Map both Optimization and
41683 PerFunction to CL_OPTIMIZATION.
41684 * opth-gen.awk: Test for PerFunction flag along with
41686 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41687 it only when the latter is present. Skip those that don't in
41688 the hash function generator.
41689 * common.opt (fvar-tracking): Mark as PerFunction instead of
41691 (fvar-tracking-assignments): Likewise.
41692 (fvar-tracking-assignments-toggle): Likewise.
41693 (fvar-tracking-uninit): Likewise.
41695 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41697 PR translation/79018
41698 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41701 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41704 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41705 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41706 TARGET_LONG_JUMP_TABLE_OFFSETS.
41707 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41708 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41709 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41710 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41711 * config/m68k/m68k.md (tablejump expander): Likewise.
41712 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41713 TARGET_LONG_JUMP_TABLE_OFFSETS.
41714 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41715 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41717 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41718 David Holsgrove <david.holsgrove@xilinx.com>
41720 * common/config/microblaze/microblaze-common.c
41721 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41722 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41724 * config/microblaze/microblaze.c (microblaze_must_save_register)
41725 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41727 (microblaze_eh_return): New function.
41728 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41729 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41730 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41731 * config/microblaze/microblaze.md (eh_return): New pattern.
41733 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41735 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41736 GCC_DIAGNOSTIC_STRINGIFY): Define.
41738 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41740 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41742 * config/arm/arm.md (<mcrr>): New.
41744 * config/arm/arm.c (arm_arch5te): New.
41745 (arm_option_override): Set arm_arch5te.
41746 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41748 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41749 (arm_mcrr_qualifiers): ... this. New.
41750 (MRRC_QUALIFIERS): Define to...
41751 (arm_mrrc_qualifiers): ... this. New.
41752 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41754 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41755 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41756 (MRRCI, mrrc, MRRC): New.
41757 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41758 VUNSPEC_MRRC2): New.
41760 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41762 * config/arm/arm.md (<mcr>): New.
41764 * config/arm/arm.c (arm_coproc_builtin_available): Add
41765 support for mcr, mrc, mcr2 and mrc2.
41766 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41767 (arm_mcr_qualifiers): ... this. New.
41768 (MRC_QUALIFIERS): Define to ...
41769 (arm_mrc_qualifiers): ... this. New.
41770 (MCR_QUALIFIERS): Define to ...
41771 (arm_mcr_qualifiers): ... this. New.
41772 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41774 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41775 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41776 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41777 VUNSPEC_MRC2): New.
41779 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41781 * config/arm/arm.md (*ldc): New.
41785 * config/arm/arm.c (arm_coproc_builtin_available): Add
41786 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41787 (arm_coproc_ldc_stc_legitimate_address): New.
41788 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41789 'qualifier_const_pointer'.
41790 (LDC_QUALIFIERS): Define to...
41791 (arm_ldc_qualifiers): ... this. New.
41792 (STC_QUALIFIERS): Define to...
41793 (arm_stc_qualifiers): ... this. New.
41794 * config/arm/arm-protos.h
41795 (arm_coproc_ldc_stc_legitimate_address): New.
41796 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41797 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41798 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41799 stc2, stcl, stc2l): New.
41800 * config/arm/constraints.md (Uz): New.
41801 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41802 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41803 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41804 VUNSPEC_STC2L): New.
41806 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41808 * config/arm/arm.md (<cdp>): New.
41809 * config/arm/arm.c (neon_const_bounds): Rename this ...
41810 (arm_const_bounds): ... this.
41811 (arm_coproc_builtin_available): New.
41812 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41813 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41814 (CDP_QUALIFIERS): Define to...
41815 (arm_cdp_qualifiers): ... this. New.
41817 (arm_expand_builtin_args): Add case for 6 arguments.
41818 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41819 (arm_const_bounds): ... this.
41820 (arm_coproc_builtin_available): New.
41821 * config/arm/arm_acle.h (__arm_cdp): New.
41823 * config/arm/arm_acle_builtins.def (cdp): New.
41825 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41826 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41827 'arm_const_bounds'.
41828 * config/arm/types.md (coproc): New.
41829 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41830 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41831 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41832 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41834 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41836 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41837 (UBINOP_QUALIFIERS): New.
41839 (acle_builtin_data): New. Change comment.
41840 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41841 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41842 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41843 arm_acle_builtins.def.
41844 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41845 (arm_init_acle_builtins): New.
41846 (CRC32_BUILTIN): Remove.
41847 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41848 crc32cb, crc32ch and crc32cw.
41849 (arm_init_crc32_builtins): Remove.
41850 (arm_init_builtins): Use arm_init_acle_builtins rather
41851 than arm_init_crc32_builtins.
41852 (arm_expand_acle_builtin): New.
41853 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41854 * config/arm/arm_acle_builtins.def: New.
41856 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41858 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41859 (arm_builtin_datum): ... this.
41860 (arm_init_neon_builtin): Rename to ...
41861 (arm_init_builtin): ... this. Add a new parameters PREFIX
41862 and USE_SIG_IN_NAME.
41863 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41864 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41865 'arm_builtin_datum'.
41866 (arm_init_vfp_builtins): Likewise.
41867 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41868 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41869 (arm_expand_neon_args): Rename to ...
41870 (arm_expand_builtin_args): ... this. Rename builtin_arg
41871 enum values and differentiate between ARG_BUILTIN_MEMORY
41872 and ARG_BUILTIN_NEON_MEMORY.
41873 (arm_expand_neon_builtin_1): Rename to ...
41874 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41875 values, arm_expand_builtin_args and add bool parameter NEON.
41876 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41877 (arm_expand_vfp_builtin): Likewise.
41878 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41880 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41882 PR middle-end/77484
41883 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41884 * predict.c (tree_estimate_probability_bb): Reverse direction of
41885 polymorphic call predictor.
41887 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41889 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41890 (determine_pass_name_match): ...this new function and...
41891 (should_skip_pass_p): ...this new function.
41893 2017-01-06 Nathan Sidwell <nathan@acm.org>
41895 * ipa-visibility.c (function_and_variable_visibility): Reformat
41896 comments and long lines. Remove extrneous if.
41897 * symtab.c (symtab_node::make_decl_local): Fix code format.
41898 (symtab_node::set_section_for_node): Fix comment typo.
41900 2017-01-06 Martin Liska <mliska@suse.cz>
41903 * lra-constraints.c: Rename invariant to lra_invariant.
41904 * predict.c (set_even_probabilities): Initialize e to NULL.
41906 2017-01-05 Martin Sebor <msebor@redhat.com>
41908 PR tree-optimization/78910
41909 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41910 (format_integer): Correct off-by-one error in the handling
41911 of precision with negative numbers in signed conversions..
41913 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41915 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41917 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41919 PR tree-optimization/71016
41920 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41921 factor_out_conditional_conversion. Formatting fix.
41922 (factor_out_conditional_conversion): Add cond_stmt argument.
41923 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41924 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41927 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41929 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41930 read-rtl-function.o, and selftest-rtl.o.
41931 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41932 (selftest::aarch64_test_loading_full_dump): New function.
41933 (selftest::aarch64_run_selftests): New function.
41934 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41935 selftest::aarch64_run_selftests.
41936 * config/i386/i386.c
41937 (selftest::ix86_test_loading_dump_fragment_1): New function.
41938 (selftest::ix86_test_loading_call_insn): New function.
41939 (selftest::ix86_test_loading_full_dump): New function.
41940 (selftest::ix86_test_loading_unspec): New function.
41941 (selftest::ix86_run_selftests): Call the new functions.
41942 * emit-rtl.c (maybe_set_max_label_num): New function.
41943 * emit-rtl.h (maybe_set_max_label_num): New decl.
41944 * function.c (instantiate_decls): Guard call to
41945 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41946 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41948 * gensupport.c (gen_reader::gen_reader): Pass "false"
41949 for new "compact" param of rtx_reader.
41950 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41951 rather than an empty string for NULL strings.
41952 * read-md.c: Potentially include config.h rather than bconfig.h.
41953 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41954 (have_error): New global, copied from errors.c.
41955 (md_reader::read_name): Rename to...
41956 (md_reader::read_name_1): ...this, adding "out_loc" param,
41957 and converting "missing name or number" to returning false, rather
41959 (md_reader::read_name): Reimplement in terms of read_name_1.
41960 (md_reader::read_name_or_nil): New function.
41961 (md_reader::read_string): Handle "(nil)" by returning NULL.
41962 (md_reader::md_reader): Add new param "compact".
41963 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41964 (md_reader::read_file): New method.
41965 * read-md.h (md_reader::md_reader): Add new param "compact".
41966 (md_reader::read_file): New method.
41967 (md_reader::is_compact): New accessor.
41968 (md_reader::read_name): Convert return type from void to file_location.
41969 (md_reader::read_name_or_nil): New decl.
41970 (md_reader::read_name_1): New decl.
41971 (md_reader::m_compact): New field.
41972 (noop_reader::noop_reader): Pass "false" for new "compact" param
41974 (rtx_reader::rtx_reader): Add new "compact" param.
41975 (rtx_reader::read_rtx_operand): Make virtual and convert return
41976 type from void to rtx.
41977 (rtx_reader::read_until): New decl.
41978 (rtx_reader::handle_any_trailing_information): New virtual function.
41979 (rtx_reader::postprocess): New virtual function.
41980 (rtx_reader::finalize_string): New virtual function.
41981 (rtx_reader::m_in_call_function_usage): New field.
41982 (rtx_reader::m_reuse_rtx_by_id): New field.
41983 * read-rtl-function.c: New file.
41984 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41985 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41986 (selftest::verify_three_block_rtl_cfg): New decl.
41987 * read-rtl-function.h: New file.
41988 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41989 For host, include function.h, memmodel.h, and emit-rtl.h.
41990 (one_time_initialization): New function.
41991 (struct compact_insn_name): New struct.
41992 (compact_insn_names): New array.
41993 (find_code): Handle insn codes in compact dumps.
41994 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41995 (bind_subst_iter_and_attr): Likewise.
41996 (add_condition_to_string): Likewise.
41997 (add_condition_to_rtx): Likewise.
41998 (apply_attribute_uses): Likewise.
41999 (add_current_iterators): Likewise.
42000 (apply_iterators): Likewise.
42001 (initialize_iterators): Guard usage of apply_subst_iterator with
42002 #ifdef GENERATOR_FILE.
42003 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42004 (md_reader::read_mapping): Likewise.
42005 (add_define_attr_for_define_subst): Likewise.
42006 (add_define_subst_attr): Likewise.
42007 (read_subst_mapping): Likewise.
42008 (check_code_iterator): Likewise.
42009 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42011 (one_time_initialization): New function.
42012 (rtx_reader::read_until): New method.
42013 (read_flags): New function.
42014 (parse_reg_note_name): New function.
42015 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42016 Handle reuse_rtx ids.
42017 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42018 Add parsing support for RTL dumps, mirroring the special-cases in
42019 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42020 values, and calling handle_any_trailing_information.
42021 (rtx_reader::read_rtx_operand): Convert return type from void
42022 to rtx, returning return_rtx. Handle case 'e'. Call
42023 finalize_string on XSTR and XTMPL fields.
42024 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42025 "(nil)" values were omitted. Call the postprocess vfunc on the
42027 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42028 class ctor. Initialize m_in_call_function_usage. Call
42029 one_time_initialization.
42030 * rtl-tests.c (selftest::test_uncond_jump): Call
42031 set_new_first_and_last_insn.
42032 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42033 * selftest-rtl.c: New file.
42034 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42035 (selftest::get_insn_by_uid): New decl.
42036 * selftest-run-tests.c (selftest::run_tests): Call
42037 read_rtl_function_c_tests.
42038 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42039 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42042 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42044 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42045 operands in a special way. Assert that pos+len <= mode precision.
42047 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42049 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42050 3 argument Alias with unlimited for the negative form.
42051 (fno-vect-cost-model): Removed.
42053 2017-01-05 Martin Liska <mliska@suse.cz>
42055 * hsa-gen.c (gen_hsa_divmod): New function.
42056 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42058 2017-01-05 Martin Liska <mliska@suse.cz>
42061 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42064 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42066 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42067 small constant length operands.
42069 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42071 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42072 between loop iterations.
42074 2017-01-05 Martin Liska <mliska@suse.cz>
42077 * gimplify.c (gimplify_decl_expr): Compare to
42078 asan_poisoned_variables instread of checking flags.
42079 (gimplify_target_expr): Likewise.
42080 (gimplify_expr): Likewise.
42081 (gimplify_function_tree): Conditionally initialize
42082 asan_poisoned_variables.
42084 2017-01-04 Jeff Law <law@redhat.com>
42086 PR tree-optimizatin/78812
42087 * rtl.h (contains_mem_rtx_p): Prototype.
42088 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42089 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42090 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42091 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42092 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42094 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42096 * input.c (assert_char_at_range): Default-initialize actual_range.
42098 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42100 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42101 to match the caller.
42103 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42105 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42106 insns after final jump in test to emit dummy move.
42108 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42110 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42111 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42113 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42115 * multiple_target.c (create_dispatcher_calls): Init e_next.
42116 * tree-ssa-loop-split.c (split_loop): Init border.
42117 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42120 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42125 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42126 (altivec_register_operand): Do not return true if the operand
42127 contains a SUBREG mixing SImode and SFmode.
42128 (vsx_register_operand): Likewise.
42129 (vsx_reg_sfsubreg_ok): New predicate.
42130 (vfloat_operand): Do not return true if the operand contains a
42131 SUBREG mixing SImode and SFmode.
42132 (vint_operand): Likewise.
42133 (vlogical_operand): Likewise.
42134 (gpc_reg_operand): Likewise.
42135 (int_reg_operand): Likewise.
42136 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42137 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42138 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42140 (rs6000_emit_move_si_sf_subreg): New helper function.
42141 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42142 fixup SUBREGs involving SImode and SFmode.
42143 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42144 numbers for the new peephole2 optimization.
42145 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42146 the GLIBC math library that do AND/IOR/XOR operations on single
42147 precision floating point.
42148 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42149 target macros to say whether we need to avoid SUBREGs mixing
42151 (TARGET_ALLOW_SF_SUBREG): Likewise.
42152 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42153 (UNSPEC_SI_FROM_SF): Likewise.
42154 (iorxor): Change spacing.
42155 (and_ior_xor): New iterator for AND, IOR, and XOR.
42156 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42157 (movdi_from_sf_zero_ext): Likewise.
42158 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42159 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42160 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42161 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42162 (fms<mode>4): Likewise.
42163 (fnma<mode>4): Likewise.
42164 (fnms<mode>4): Likewise.
42165 (nfma<mode>4): Likewise.
42166 (nfms<mode>4): Likewise.
42168 2017-01-04 Marek Polacek <polacek@redhat.com>
42171 * doc/invoke.texi: Document -Wpointer-compare.
42173 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42175 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42178 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42179 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42182 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42185 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42186 documentation of the powerpc_popcntb_ok attribute.
42187 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42188 code to issue warning messages if a requested CPU configuration is
42189 not supported by the binary (assembler and loader) toolchain.
42190 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42191 made to define a built-in function that has been disabled.
42192 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42193 made to define a built-in function that has been disabled.
42194 (altivec_init_builtins): Add comment explaining why definition
42195 of the DST built-in functions is not preceded by an assertion
42196 check. Add assertions to prevent ICE if attempts are made to
42197 define an altivec predicate or an abs* built-in function that has
42199 (htm_init_builtins): Add comment explaining why definition of the
42200 htm built-in functions is not preceded by an assertion check.
42202 2017-01-04 Jeff Law <law@redhat.com>
42204 PR tree-optimizatin/67955
42205 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42206 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42207 the points-to solution does not include pt_null. Use DECL_PT_UID
42210 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42212 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42213 Use gen_int_mode instead of gen_lopwart for const_int operands.
42215 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42217 PR tree-optimization/71563
42218 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42219 out of range value - has low bits known to be zero.
42221 2017-01-04 Alan Modra <amodra@gmail.com>
42223 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42224 * configure: Regenerate.
42225 * config.in: Regenerate.
42227 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42230 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42231 a substring of the message, but strcmp with the whole message. Ifdef
42232 ENABLE_NLS, translate the message first using dgettext.
42234 2017-01-03 Jeff Law <law@redhat.com>
42236 PR tree-optimizatin/78856
42237 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42238 (mark_threaded_blocks): Remove code to truncate thread paths that
42239 cross multiple loop headers. Instead invalidate the cached loop
42240 iteration information and handle case of a thread path walking
42241 into an irreducible region.
42243 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42246 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42247 assertions. Add support for doing the signbit if the IEEE 128-bit
42248 floating point value is in a GPR.
42249 * config/rs6000/rs6000.md (Fsignbit): Delete.
42250 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42251 Update the length attribute if the value is in a GPR.
42252 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42253 the sign or zero extension instruction, since the value is always 0/1.
42254 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42257 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42258 extracting SImode to a GPR register so that we can generate a
42259 store, limit the vector to be in a traditional Altivec register
42260 for the vextuwrx instruction.
42262 2017-01-03 Ian Lance Taylor <iant@google.com>
42264 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42266 2017-01-03 Martin Sebor <msebor@redhat.com>
42268 PR tree-optimization/78696
42269 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42270 precision. Use MPFR for %f for greater fidelity. Correct handling
42272 (pass_sprintf_length::compute_format_length): Set width and precision
42273 specified by asrerisk to void_node for vararg functions.
42274 (try_substitute_return_value): Adjust dump output.
42276 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42278 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42280 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42282 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42283 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42284 -mlra/-mno-lra was passed to the compiler.
42286 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42288 PR rtl-optimization/65618
42289 * emit-rtl.c (try_split): Move initialization of "before" and
42290 "after" to just before the call to emit_insn_after_setloc.
42292 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42294 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42296 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42298 * dwarf2out.c (gen_enumeration_type_die): When
42299 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42301 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42303 PR tree-optimization/78965
42304 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42305 Change first argument from const call_info & to call_info &. For %n
42306 set info.nowrite to false.
42308 PR middle-end/78901
42309 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42310 possibly throwing calls.
42312 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42313 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42314 and fns handling, rather than in a separate case SSA_NAME.
42316 2017-01-02 Jeff Law <law@redhat.com>
42318 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42319 fixes for first_period and second_period variables.
42321 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42324 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42325 (*insvqi_1): New insn pattern.
42326 (*insvqi_1_mem_rex64): Ditto.
42327 (*insvqi_2): Ditto.
42328 (*insvqi_3): Rename from *insvqi.
42330 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42332 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42334 * doc/cfg.texi (Edges): Remove reference to Java.
42335 (Maintaining the CFG): Ditto.
42337 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42339 PR middle-end/77674
42340 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42341 transparent aliases.
42343 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42345 PR middle-end/77484
42346 * predict.def (PRED_CALL): Update hitrate.
42347 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42348 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42349 into direct/indirect/polymorphic variants.
42351 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42353 Update copyright years.
42355 * gcc.c (process_command): Update copyright notice dates.
42356 * gcov-dump.c (print_version): Ditto.
42357 * gcov.c (print_version): Ditto.
42358 * gcov-tool.c (print_version): Ditto.
42359 * gengtype.c (create_file): Ditto.
42360 * doc/cpp.texi: Bump @copying's copyright year.
42361 * doc/cppinternals.texi: Ditto.
42362 * doc/gcc.texi: Ditto.
42363 * doc/gccint.texi: Ditto.
42364 * doc/gcov.texi: Ditto.
42365 * doc/install.texi: Ditto.
42366 * doc/invoke.texi: Ditto.
42368 Copyright (C) 2017 Free Software Foundation, Inc.
42370 Copying and distribution of this file, with or without modification,
42371 are permitted in any medium without royalty provided the copyright
42372 notice and this notice are preserved.