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 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
6 number of units as polynomial.
8 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
9 Alan Hayward <alan.hayward@arm.com>
10 David Sherwood <david.sherwood@arm.com>
12 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
13 * target.def (autovectorize_vector_sizes): Return the vector sizes
14 by pointer, using vector_sizes rather than a bitmask.
15 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
16 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
17 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
19 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
20 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
21 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
22 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
23 * omp-general.c (omp_max_vf): Likewise.
24 * omp-low.c (omp_clause_aligned_alignment): Likewise.
25 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
26 * tree-vect-loop.c (vect_analyze_loop): Likewise.
27 * tree-vect-slp.c (vect_slp_bb): Likewise.
28 * doc/tm.texi: Regenerate.
29 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
31 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
32 the vector size as a poly_uint64 rather than an unsigned int.
33 (current_vector_size): Change from an unsigned int to a poly_uint64.
34 (get_vectype_for_scalar_type): Update accordingly.
35 * tree.h (build_truth_vector_type): Take the size and number of
36 units as a poly_uint64 rather than an unsigned int.
37 (build_vector_type): Add a temporary overload that takes
38 the number of units as a poly_uint64 rather than an unsigned int.
39 * tree.c (make_vector_type): Likewise.
40 (build_truth_vector_type): Take the number of units as a poly_uint64
41 rather than an unsigned int.
43 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
44 Alan Hayward <alan.hayward@arm.com>
45 David Sherwood <david.sherwood@arm.com>
47 * target.def (get_mask_mode): Take the number of units and length
48 as poly_uint64s rather than unsigned ints.
49 * targhooks.h (default_get_mask_mode): Update accordingly.
50 * targhooks.c (default_get_mask_mode): Likewise.
51 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
52 * doc/tm.texi: Regenerate.
54 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
55 Alan Hayward <alan.hayward@arm.com>
56 David Sherwood <david.sherwood@arm.com>
58 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
59 * omp-general.c (omp_max_vf): Likewise.
60 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
61 (expand_omp_simd): Handle polynomial safelen.
62 * omp-low.c (omplow_simd_context): Add a default constructor.
63 (omplow_simd_context::max_vf): Change from int to poly_uint64.
64 (lower_rec_simd_input_clauses): Update accordingly.
65 (lower_rec_input_clauses): Likewise.
67 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
68 Alan Hayward <alan.hayward@arm.com>
69 David Sherwood <david.sherwood@arm.com>
71 * tree-vectorizer.h (vect_nunits_for_cost): New function.
72 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
73 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
74 (vect_analyze_slp_cost): Likewise.
75 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
76 (vect_model_load_cost): Likewise.
78 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
79 Alan Hayward <alan.hayward@arm.com>
80 David Sherwood <david.sherwood@arm.com>
82 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
83 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
84 from an unsigned int * to a poly_uint64_pod *.
85 (calculate_unrolling_factor): New function.
86 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
88 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
89 Alan Hayward <alan.hayward@arm.com>
90 David Sherwood <david.sherwood@arm.com>
92 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
93 from an unsigned int to a poly_uint64.
94 (_loop_vec_info::slp_unrolling_factor): Likewise.
95 (_loop_vec_info::vectorization_factor): Change from an int
97 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
98 (vect_get_num_vectors): New function.
99 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
100 (vect_get_num_copies): Use vect_get_num_vectors.
101 (vect_analyze_data_ref_dependences): Change max_vf from an int *
102 to an unsigned int *.
103 (vect_analyze_data_refs): Change min_vf from an int * to a
105 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
106 than an unsigned HOST_WIDE_INT.
107 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
108 (vect_analyze_data_ref_dependence): Change max_vf from an int *
109 to an unsigned int *.
110 (vect_analyze_data_ref_dependences): Likewise.
111 (vect_compute_data_ref_alignment): Handle polynomial vf.
112 (vect_enhance_data_refs_alignment): Likewise.
113 (vect_prune_runtime_alias_test_list): Likewise.
114 (vect_shift_permute_load_chain): Likewise.
115 (vect_supportable_dr_alignment): Likewise.
116 (dependence_distance_ge_vf): Take the vectorization factor as a
117 poly_uint64 rather than an unsigned HOST_WIDE_INT.
118 (vect_analyze_data_refs): Change min_vf from an int * to a
120 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
121 vfm1 as a poly_uint64 rather than an int. Make the same change
122 for the returned bound_scalar.
123 (vect_gen_vector_loop_niters): Handle polynomial vf.
124 (vect_do_peeling): Likewise. Update call to
125 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
126 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
128 * tree-vect-loop.c (vect_determine_vectorization_factor)
129 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
130 (vect_get_known_peeling_cost): Likewise.
131 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
132 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
133 (vect_transform_loop): Likewise. Use the lowest possible VF when
134 updating the upper bounds of the loop.
135 (vect_min_worthwhile_factor): Make static. Return an unsigned int
137 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
138 polynomial unroll factors.
139 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
140 (vect_make_slp_decision): Likewise.
141 (vect_supported_load_permutation_p): Likewise, and polynomial
143 (vect_analyze_slp_cost): Handle polynomial vf.
144 (vect_slp_analyze_node_operations): Likewise.
145 (vect_slp_analyze_bb_1): Likewise.
146 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
147 than an unsigned HOST_WIDE_INT.
148 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
149 (vectorizable_load): Handle polynomial vf.
150 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
152 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
154 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
155 Alan Hayward <alan.hayward@arm.com>
156 David Sherwood <david.sherwood@arm.com>
158 * match.pd: Handle bit operations involving three constants
159 and try to fold one pair.
161 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
163 * tree-vect-loop-manip.c: Include gimple-fold.h.
164 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
165 niters_maybe_zero parameters. Handle other cases besides a step of 1.
166 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
167 Add a path that uses a step of VF instead of 1, but disable it
169 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
170 and niters_no_overflow parameters. Update calls to
171 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
172 Create a new SSA name if the latter choses to use a ste other
173 than zero, and return it via niters_vector_mult_vf_var.
174 * tree-vect-loop.c (vect_transform_loop): Update calls to
175 vect_do_peeling, vect_gen_vector_loop_niters and
176 slpeel_make_loop_iterate_ntimes.
177 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
178 (vect_gen_vector_loop_niters): Update declarations after above changes.
180 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
182 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
183 128-bit round to integer instructions.
184 (ceil<mode>2): Likewise.
185 (btrunc<mode>2): Likewise.
186 (round<mode>2): Likewise.
188 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
190 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
191 unaligned VSX load/store on P8/P9.
192 (expand_block_clear): Allow the use of unaligned VSX
195 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
197 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
199 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
200 swap associated with both a load and a store.
202 2018-01-02 Andrew Waterman <andrew@sifive.com>
204 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
205 * config/riscv/riscv.md (clear_cache): Use it.
207 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
209 * web.c: Remove out-of-date comment.
211 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
213 * expr.c (fixup_args_size_notes): Check that any existing
214 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
215 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
216 (emit_single_push_insn): ...here.
218 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
220 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
221 (const_vector_encoded_nelts): New function.
222 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
223 (const_vector_int_elt, const_vector_elt): Declare.
224 * emit-rtl.c (const_vector_int_elt_1): New function.
225 (const_vector_elt): Likewise.
226 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
229 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
231 * expr.c: Include rtx-vector-builder.h.
232 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
233 directly on the tree encoding.
234 (const_vector_from_tree): Likewise.
235 * optabs.c: Include rtx-vector-builder.h.
236 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
237 sequence of "u" values.
238 * vec-perm-indices.c: Include rtx-vector-builder.h.
239 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
240 directly on the vec_perm_indices encoding.
242 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
244 * doc/rtl.texi (const_vector): Describe new encoding scheme.
245 * Makefile.in (OBJS): Add rtx-vector-builder.o.
246 * rtx-vector-builder.h: New file.
247 * rtx-vector-builder.c: Likewise.
248 * rtl.h (rtx_def::u2): Add a const_vector field.
249 (CONST_VECTOR_NPATTERNS): New macro.
250 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
251 (CONST_VECTOR_DUPLICATE_P): Likewise.
252 (CONST_VECTOR_STEPPED_P): Likewise.
253 (CONST_VECTOR_ENCODED_ELT): Likewise.
254 (const_vec_duplicate_p): Check for a duplicated vector encoding.
255 (unwrap_const_vec_duplicate): Likewise.
256 (const_vec_series_p): Check for a non-duplicated vector encoding.
257 Say that the function only returns true for integer vectors.
258 * emit-rtl.c: Include rtx-vector-builder.h.
259 (gen_const_vec_duplicate_1): Delete.
260 (gen_const_vector): Call gen_const_vec_duplicate instead of
261 gen_const_vec_duplicate_1.
262 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
263 (gen_const_vec_duplicate): Use rtx_vector_builder.
264 (gen_const_vec_series): Likewise.
265 (gen_rtx_CONST_VECTOR): Likewise.
266 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
267 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
268 Build a new vector rather than modifying a CONST_VECTOR in-place.
269 (handle_special_swappables): Update call accordingly.
270 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
271 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
272 Build a new vector rather than modifying a CONST_VECTOR in-place.
273 (handle_special_swappables): Update call accordingly.
275 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
277 * simplify-rtx.c (simplify_const_binary_operation): Use
278 CONST_VECTOR_ELT instead of XVECEXP.
280 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
282 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
283 the selector elements to be different from the data elements
284 if the selector is a VECTOR_CST.
285 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
286 ssizetype for the selector.
288 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
290 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
291 before testing each element individually.
292 * tree-vect-generic.c (lower_vec_perm): Likewise.
294 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
296 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
297 * selftest-run-tests.c (selftest::run_tests): Call it.
298 * vector-builder.h (vector_builder::operator ==): New function.
299 (vector_builder::operator !=): Likewise.
300 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
301 (vec_perm_indices::all_from_input_p): New function.
302 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
303 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
304 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
305 instead of reading the VECTOR_CST directly. Detect whether both
306 vector inputs are the same before constructing the vec_perm_indices,
307 and update the number of inputs argument accordingly. Use the
308 utility functions added above. Only construct sel2 if we need to.
310 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
312 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
313 the broadcast of the low byte.
314 (expand_mult_highpart): Use an explicit encoding for the permutes.
315 * optabs-query.c (can_mult_highpart_p): Likewise.
316 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
317 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
318 (vectorizable_bswap): Likewise.
319 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
320 explicit encoding for the power-of-2 permutes.
321 (vect_permute_store_chain): Likewise.
322 (vect_grouped_load_supported): Likewise.
323 (vect_permute_load_chain): Likewise.
325 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
327 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
328 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
329 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
330 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
331 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
332 (vect_gen_perm_mask_any): Likewise.
334 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
336 * int-vector-builder.h: New file.
337 * vec-perm-indices.h: Include int-vector-builder.h.
338 (vec_perm_indices): Redefine as an int_vector_builder.
339 (auto_vec_perm_indices): Delete.
340 (vec_perm_builder): Redefine as a stand-alone class.
341 (vec_perm_indices::vec_perm_indices): New function.
342 (vec_perm_indices::clamp): Likewise.
343 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
344 (vec_perm_indices::new_vector): New function.
345 (vec_perm_indices::new_expanded_vector): Update for new
346 vec_perm_indices class.
347 (vec_perm_indices::rotate_inputs): New function.
348 (vec_perm_indices::all_in_range_p): Operate directly on the
349 encoded form, without computing elided elements.
350 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
351 encoding. Update for new vec_perm_indices class.
352 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
353 the given vec_perm_builder.
354 (expand_vec_perm_var): Update vec_perm_builder constructor.
355 (expand_mult_highpart): Use vec_perm_builder instead of
356 auto_vec_perm_indices.
357 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
358 vec_perm_indices instead of auto_vec_perm_indices. Use a single
359 or double series encoding as appropriate.
360 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
361 vec_perm_indices instead of auto_vec_perm_indices.
362 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
363 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
364 (vect_permute_store_chain): Likewise.
365 (vect_grouped_load_supported): Likewise.
366 (vect_permute_load_chain): Likewise.
367 (vect_shift_permute_load_chain): Likewise.
368 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
369 (vect_transform_slp_perm_load): Likewise.
370 (vect_schedule_slp_instance): Likewise.
371 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
372 (vectorizable_mask_load_store): Likewise.
373 (vectorizable_bswap): Likewise.
374 (vectorizable_store): Likewise.
375 (vectorizable_load): Likewise.
376 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
377 vec_perm_indices instead of auto_vec_perm_indices. Use
378 tree_to_vec_perm_builder to read the vector from a tree.
379 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
380 vec_perm_builder instead of a vec_perm_indices.
381 (have_whole_vector_shift): Use vec_perm_builder and
382 vec_perm_indices instead of auto_vec_perm_indices. Leave the
383 truncation to calc_vec_perm_mask_for_shift.
384 (vect_create_epilog_for_reduction): Likewise.
385 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
386 from auto_vec_perm_indices to vec_perm_indices.
387 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
388 instead of changing individual elements.
389 (aarch64_vectorize_vec_perm_const): Use new_vector to install
390 the vector in d.perm.
391 * config/arm/arm.c (expand_vec_perm_d::perm): Change
392 from auto_vec_perm_indices to vec_perm_indices.
393 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
394 instead of changing individual elements.
395 (arm_vectorize_vec_perm_const): Use new_vector to install
396 the vector in d.perm.
397 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
398 Update vec_perm_builder constructor.
399 (rs6000_expand_interleave): Likewise.
400 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
401 (rs6000_expand_interleave): Likewise.
403 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
405 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
406 to qimode could truncate the indices.
407 * optabs.c (expand_vec_perm_var): Likewise.
409 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
411 * Makefile.in (OBJS): Add vec-perm-indices.o.
412 * vec-perm-indices.h: New file.
413 * vec-perm-indices.c: Likewise.
414 * target.h (vec_perm_indices): Replace with a forward class
416 (auto_vec_perm_indices): Move to vec-perm-indices.h.
417 * optabs.h: Include vec-perm-indices.h.
418 (expand_vec_perm): Delete.
419 (selector_fits_mode_p, expand_vec_perm_var): Declare.
420 (expand_vec_perm_const): Declare.
421 * target.def (vec_perm_const_ok): Replace with...
422 (vec_perm_const): ...this new hook.
423 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
424 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
425 * doc/tm.texi: Regenerate.
426 * optabs.def (vec_perm_const): Delete.
427 * doc/md.texi (vec_perm_const): Likewise.
428 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
429 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
430 expand_vec_perm for constant permutation vectors. Assert that
431 the mode of variable permutation vectors is the integer equivalent
432 of the mode that is being permuted.
433 * optabs-query.h (selector_fits_mode_p): Declare.
434 * optabs-query.c: Include vec-perm-indices.h.
435 (selector_fits_mode_p): New function.
436 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
437 is defined, instead of checking whether the vec_perm_const_optab
438 exists. Use targetm.vectorize.vec_perm_const instead of
439 targetm.vectorize.vec_perm_const_ok. Check whether the indices
440 fit in the vector mode before using a variable permute.
441 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
442 vec_perm_indices instead of an rtx.
443 (expand_vec_perm): Replace with...
444 (expand_vec_perm_const): ...this new function. Take the selector
445 as a vec_perm_indices rather than an rtx. Also take the mode of
446 the selector. Update call to shift_amt_for_vec_perm_mask.
447 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
448 Use vec_perm_indices::new_expanded_vector to expand the original
449 selector into bytes. Check whether the indices fit in the vector
450 mode before using a variable permute.
451 (expand_vec_perm_var): Make global.
452 (expand_mult_highpart): Use expand_vec_perm_const.
453 * fold-const.c: Includes vec-perm-indices.h.
454 * tree-ssa-forwprop.c: Likewise.
455 * tree-vect-data-refs.c: Likewise.
456 * tree-vect-generic.c: Likewise.
457 * tree-vect-loop.c: Likewise.
458 * tree-vect-slp.c: Likewise.
459 * tree-vect-stmts.c: Likewise.
460 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
462 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
463 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
464 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
465 (aarch64_vectorize_vec_perm_const): ...this new function.
466 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
467 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
468 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
469 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
470 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
471 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
472 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
474 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
475 check for NEON modes.
476 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
477 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
478 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
479 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
481 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
482 the old VEC_PERM_CONST conditions.
483 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
484 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
485 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
486 (ia64_vectorize_vec_perm_const_ok): Merge into...
487 (ia64_vectorize_vec_perm_const): ...this new function.
488 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
489 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
490 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
491 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
492 * config/mips/mips.c (mips_expand_vec_perm_const)
493 (mips_vectorize_vec_perm_const_ok): Merge into...
494 (mips_vectorize_vec_perm_const): ...this new function.
495 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
496 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
497 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
498 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
499 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
500 (rs6000_expand_vec_perm_const): Delete.
501 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
503 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
504 (altivec_expand_vec_perm_const_le): Take each operand individually.
505 Operate on constant selectors rather than rtxes.
506 (altivec_expand_vec_perm_const): Likewise. Update call to
507 altivec_expand_vec_perm_const_le.
508 (rs6000_expand_vec_perm_const): Delete.
509 (rs6000_vectorize_vec_perm_const_ok): Delete.
510 (rs6000_vectorize_vec_perm_const): New function.
511 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
512 an element count and rtx array.
513 (rs6000_expand_extract_even): Update call accordingly.
514 (rs6000_expand_interleave): Likewise.
515 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
516 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
517 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
518 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
519 (rs6000_expand_vec_perm_const): Delete.
520 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
521 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
522 (altivec_expand_vec_perm_const_le): Take each operand individually.
523 Operate on constant selectors rather than rtxes.
524 (altivec_expand_vec_perm_const): Likewise. Update call to
525 altivec_expand_vec_perm_const_le.
526 (rs6000_expand_vec_perm_const): Delete.
527 (rs6000_vectorize_vec_perm_const_ok): Delete.
528 (rs6000_vectorize_vec_perm_const): New function. Remove stray
529 reference to the SPE evmerge intructions.
530 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
531 an element count and rtx array.
532 (rs6000_expand_extract_even): Update call accordingly.
533 (rs6000_expand_interleave): Likewise.
534 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
535 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
537 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
539 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
541 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
542 vector mode and that that mode matches the mode of the data
544 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
545 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
546 directly using expand_vec_perm_1 when forcing selectors into
548 (expand_vec_perm_var): New function, split out from expand_vec_perm.
550 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
552 * optabs-query.h (can_vec_perm_p): Delete.
553 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
554 * optabs-query.c (can_vec_perm_p): Split into...
555 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
556 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
557 particular selector is valid.
558 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
559 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
560 (vect_grouped_load_supported): Likewise.
561 (vect_shift_permute_load_chain): Likewise.
562 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
563 (vect_transform_slp_perm_load): Likewise.
564 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
565 (vectorizable_bswap): Likewise.
566 (vect_gen_perm_mask_checked): Likewise.
567 * fold-const.c (fold_ternary_loc): Likewise. Don't take
568 implementations of variable permutation vectors into account
569 when deciding which selector to use.
570 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
571 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
572 with a false third argument.
573 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
574 to test whether the constant selector is valid and can_vec_perm_var_p
575 to test whether a variable selector is valid.
577 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
579 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
580 * optabs-query.c (can_vec_perm_p): Likewise.
581 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
582 instead of vec_perm_indices.
583 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
584 (vect_gen_perm_mask_checked): Likewise,
585 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
586 (vect_gen_perm_mask_checked): Likewise,
588 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
590 * optabs-query.h (qimode_for_vec_perm): Declare.
591 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
592 (qimode_for_vec_perm): ...this new function.
593 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
595 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
597 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
598 does not have a conditional at the top.
600 2018-01-02 Richard Biener <rguenther@suse.de>
602 * ipa-inline.c (big_speedup_p): Fix expression.
604 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
607 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
610 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
614 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
615 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
616 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
617 cond_taken_branch_cost 3->4.
619 2018-01-01 Jakub Jelinek <jakub@redhat.com>
621 PR tree-optimization/83581
622 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
623 TODO_cleanup_cfg if any changes have been made.
626 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
627 convert_modes if target mode has the right side, but different mode
631 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
632 last argument when extracting from CONCAT. If either from_real or
633 from_imag is NULL, use expansion through memory. If result is not
634 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
635 the parts directly to inner mode, if even that fails, use expansion
639 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
640 check for bswap in mode rather than HImode and use that in expand_unop
643 2017-12-31 Jakub Jelinek <jakub@redhat.com>
646 * config/i386/i386.c (ix86_attribute_table): Remove excess
647 initializer for "shared" attribute.
649 2017-12-30 Tom de Vries <tom@codesourcery.com>
652 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
654 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
655 offload_funcs entries that no longer have a corresponding cgraph_node.
656 Mark the remaining ones as DECL_PRESERVE_P.
657 (output_lto): Call prune_offload_funcs.
659 2017-12-30 Jakub Jelinek <jakub@redhat.com>
661 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
662 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
663 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
664 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
665 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
666 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
667 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
668 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
669 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
670 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
671 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
672 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
673 vaesenc_<mode>, vpclmulqdq_<mode>,
674 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
676 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
678 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
679 _Float<N> and _Float<N>X variants for rounding built-in
681 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
682 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
683 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
684 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
685 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
686 * builtins.c (mathfn_built_in_2): Likewise.
687 * internal-fn.def (CEIL): Likewise.
689 (NEARBYINT): Likewise.
693 * convert.c (convert_to_integer_1): Likewise.
694 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
695 (integer_valued_real_call_p): Likewise.
696 * fold-const-call.c (fold_const_call_ss): Likewise.
697 * gencfn-macros.c (print_case_cfn): Change CFN and operator
698 printers to take a const char * suffix instead of a bool.
699 (print_define_operator_list): Likewise.
700 (fltall_suffixes): New list of suffixes, that include the
701 traditional suffixes as well as all of the _Float<N> and
703 (main): For _Float<N> and _Float<N>X functions, emit both
704 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
705 has the _Float<N> and _Float<N>X case names or operators. The
706 <name>_ALL variant has both the traditional and the
707 _Float<N>/_Float<N>X case names or operators.
708 * match.pd (COPYSIGN optimizations): Provide optimizations for
709 _Float<N> and _Float<N>X types where possible.
710 (MIN/MAX optimizations): Likewise.
711 (sqrt optimizations): Likewise.
712 (rounding optimizations): Likewise.
714 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
716 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
717 gen_int_mode rather than GEN_INT.
718 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
720 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
722 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
723 instead of CONSTANT_P.
724 (gen_vec_series): Likewise.
725 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
727 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
729 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
732 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
734 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
735 (aarch64_ld1x2<VDC:mode>): Likewise.
736 (aarch64_simd_ld1<mode>_x2): Likewise.
737 (aarch64_simd_ld1<mode>_x2): Likewise.
738 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
739 (vld1_s8_x2): Likewise.
740 (vld1_u16_x2): Likewise.
741 (vld1_s16_x2): Likewise.
742 (vld1_u32_x2): Likewise.
743 (vld1_s32_x2): Likewise.
744 (vld1_u64_x2): Likewise.
745 (vld1_s64_x2): Likewise.
746 (vld1_f16_x2): Likewise.
747 (vld1_f32_x2): Likewise.
748 (vld1_f64_x2): Likewise.
749 (vld1_p8_x2): Likewise.
750 (vld1_p16_x2): Likewise.
751 (vld1_p64_x2): Likewise.
752 (vld1q_u8_x2): Likewise.
753 (vld1q_s8_x2): Likewise.
754 (vld1q_u16_x2): Likewise.
755 (vld1q_s16_x2): Likewise.
756 (vld1q_u32_x2): Likewise.
757 (vld1q_s32_x2): Likewise.
758 (vld1q_u64_x2): Likewise.
759 (vld1q_s64_x2): Likewise.
760 (vld1q_f16_x2): Likewise.
761 (vld1q_f32_x2): Likewise.
762 (vld1q_f64_x2): Likewise.
763 (vld1q_p8_x2): Likewise.
764 (vld1q_p16_x2): Likewise.
765 (vld1q_p64_x2): Likewise.
767 2017-12-27 Martin Liska <mliska@suse.cz>
769 PR tree-optimization/83552
770 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
771 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
773 2017-12-27 Tom de Vries <tom@codesourcery.com>
775 * config/nvptx/nvptx.c (nvptx_option_override): Disable
776 -gstatement-frontiers.
778 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
780 PR rtl-optimization/83513
781 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
782 before priority comparison.
784 2017-12-25 Jakub Jelinek <jakub@redhat.com>
787 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
788 ix86_isa_flags2 to ix86_isa_flags.
789 * config/i386/i386-c.c (ix86_target_macros_internal): Test
790 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
791 isa_flags rather than isa_flags2.
792 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
793 and -mavx512bitalg from isa2_opts to isa_opts.
794 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
795 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
796 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
798 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
799 OPTION_MASK_ISA_AVX512F ored with another option similarly to
800 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
801 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
802 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
803 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
804 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
805 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
806 builtins from bdesc_args2 to bdesc_args section.
807 (__builtin_ia32_compressstoreuqi512_mask,
808 __builtin_ia32_compressstoreuhi512_mask,
809 __builtin_ia32_compressstoreuqi256_mask,
810 __builtin_ia32_expandloadqi512_mask,
811 __builtin_ia32_expandloadqi512_maskz,
812 __builtin_ia32_expandloadhi512_mask,
813 __builtin_ia32_expandloadhi512_maskz,
814 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
815 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
816 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
817 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
818 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
819 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
820 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
821 __builtin_ia32_vpshldv_v32hi_maskz,
822 __builtin_ia32_vpopcountb_v64qi_mask,
823 __builtin_ia32_vpopcountw_v32hi_mask,
824 __builtin_ia32_vpshufbitqmb512_mask,
825 __builtin_ia32_vpshufbitqmb256_mask): Add
826 " | OPTION_MASK_ISA_AVX512BW".
827 (__builtin_ia32_expandloadqi256_mask,
828 __builtin_ia32_expandloadqi256_maskz,
829 __builtin_ia32_vpopcountb_v32qi_mask): Add
830 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
831 (__builtin_ia32_expandloadhi256_mask,
832 __builtin_ia32_expandloadhi256_maskz,
833 __builtin_ia32_expandloadqi128_mask,
834 __builtin_ia32_expandloadqi128_maskz,
835 __builtin_ia32_expandloadhi128_mask,
836 __builtin_ia32_expandloadhi128_maskz,
837 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
838 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
839 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
840 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
841 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
842 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
843 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
844 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
845 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
846 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
847 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
848 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
849 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
850 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
851 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
852 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
853 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
854 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
855 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
856 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
857 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
858 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
859 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
860 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
861 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
862 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
863 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
864 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
865 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
866 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
867 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
868 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
869 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
870 __builtin_ia32_vpopcountw_v8hi_mask): Add
871 " | OPTION_MASK_ISA_AVX512VL".
872 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
873 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
874 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
875 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
876 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
877 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
878 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
879 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
880 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
881 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
882 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
883 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
884 _mm_mask_bitshuffle_epi64_mask): Likewise.
885 * common/config/i386/i386-common.c
886 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
887 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
888 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
889 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
890 OPTION_MASK_ISA_AVX512BITALG_UNSET.
891 (OPTION_MASK_ISA2_AVX512F_UNSET,
892 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
893 (ix86_handle_option): For -mno-general-regs-only, clear from
894 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
895 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
896 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
897 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
898 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
899 for moving from ix86_isa_flags2 to ix86_isa_flags.
901 2017-12-23 Jakub Jelinek <jakub@redhat.com>
904 * fold-const.c (struct contains_label_data): New type.
905 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
906 inside of a SWITCH_BODY seen during the walk.
907 (contains_label_p): Use walk_tree instead of
908 walk_tree_without_duplicates, prepare data for contains_label_1 and
911 2017-12-22 Jakub Jelinek <jakub@redhat.com>
914 * tree-iterator.c (alloc_stmt_list): Start with cleared
915 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
919 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
920 enable avx512f explicitly in #pragma GCC target.
921 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
922 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
923 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
924 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
925 __builtin_ia32_vpdpbusds_v8si_mask,
926 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
927 __builtin_ia32_vpdpbusds_v4si_mask,
928 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
929 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
930 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
931 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
932 __builtin_ia32_vpdpwssds_v8si_mask,
933 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
934 __builtin_ia32_vpdpwssds_v4si_mask,
935 __builtin_ia32_vpdpwssds_v4si_maskz): Use
936 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
937 just OPTION_MASK_ISA_AVX512VNNI.
939 2017-12-22 Martin Jambor <mjambor@suse.cz>
942 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
945 2017-12-22 Julia Koval <julia.koval@intel.com>
946 Sebastian Peryt <sebastian.peryt@intel.com>
948 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
949 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
950 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
951 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
952 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
953 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
954 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
955 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
956 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
957 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
958 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
959 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
960 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
961 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
962 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
963 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
964 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
965 _mm256_maskz_popcnt_epi64): New intrinsics.
966 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
967 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
968 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
969 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
970 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
971 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
972 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
973 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
974 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
975 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
976 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
977 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
978 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
979 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
980 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
981 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
982 __builtin_ia32_vpshufbitqmb256_mask,
983 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
984 * config/i386/i386-c.c (__AVX512BITALG__): New.
985 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
986 (ix86_valid_target_attribute_inner_p): Ditto.
987 (ix86_expand_args_builtin): Handle new types.
988 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
989 * config/i386/i386.opt: Add -mavx512bitalg.
990 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
991 avx512bitalgintrin.h.
992 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
993 (vpopcount<mode><mask_name>): Add more types.
994 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
995 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
997 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
999 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
1000 Or in OPTION_MASK_ISA_AVX512F_SET.
1001 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1002 OPTION_MASK_ISA_AVX512VNNI_UNSET.
1003 (ix86_handle_option): Adjust for
1004 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
1005 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1007 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1008 isa_flag instead of isa_flag2.
1009 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1010 isa_opts2 to isa_opts.
1011 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1014 2017-12-22 Mike Stump <mikestump@comcast.net>
1015 Eric Botcazou <ebotcazou@adacore.com>
1017 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1019 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1022 * config/i386/i386.c (ix86_function_arg_boundary): Return
1023 PARM_BOUNDARY for TYPE_EMPTY_P types.
1026 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1027 if navail is >= dir.len.
1029 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1031 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1034 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1036 PR rtl-optimization/80747
1037 PR rtl-optimization/83512
1038 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1039 succ edge from ENTRY, copy partition from e->dest to the newly
1041 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1042 ENTRY, use BB_PARTITION of its successor block as current_partition.
1043 Don't copy partition when splitting succ edge from ENTRY.
1045 PR tree-optimization/83523
1046 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1047 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1048 (convert_mult_to_fma): Likewise.
1050 PR tree-optimization/83521
1051 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1052 gimple_build_assign without code on result of
1053 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1054 a VIEW_CONVERT_EXPR.
1056 2017-12-21 Andrew Pinski <apinski@cavium.com>
1057 Steve Ellcey <sellcey@cavium.com>
1059 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1060 multi-arch for ilp32.
1062 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1065 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1066 constraints to operand 2.
1067 (*ashl<mode>3_mask_1): Ditto.
1068 (*<shift_insn><mode>3_mask): Ditto.
1069 (*<shift_insn><mode>3_mask_1): Ditto.
1070 (*<rotate_insn><mode>3_mask): Ditto.
1071 (*<rotate_insn><mode>3_mask_1): Ditto.
1073 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1075 * reorg.c (make_return_insns): Reemit each insn with its own location.
1077 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1079 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1080 generation for cases where splatting a value is not useful.
1081 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1082 across a vec_duplicate and a paradoxical subreg forming a vector
1083 mode to a vec_concat.
1085 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1087 * combine.c (simplify_set): Do not transform subregs to zero_extends
1088 if the destination is not a scalar int mode.
1090 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1093 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1094 shared zero if the input has overflowed.
1096 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1098 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1099 support for Cortex-A55 and Cortex-A75.
1101 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1103 * common/config/arm/arm-common.c (compare_opt_names): Add function
1104 comment. Use strcmp instead of manual loop.
1106 2017-12-21 Martin Liska <mliska@suse.cz>
1108 PR gcov-profile/83509
1109 * gcov-dump.c (dump_gcov_file): Do not read info about
1110 support_unexecuted_blocks for gcda files.
1112 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1114 PR rtl-optimization/82973
1115 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1116 (valid_for_const_vector_p): ... this.
1117 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1118 (valid_for_const_vector_p): ... this. Adjust function comment.
1119 (gen_vec_duplicate): Adjust caller.
1120 * optabs.c (expand_vector_broadcast): Likewise.
1121 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1122 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1124 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1125 instead of GET_CODE == CONST_FIXED.
1126 (simplify_subreg): Use CONST_FIXED_P macro instead of
1127 GET_CODE == CONST_FIXED.
1130 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1131 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1132 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1133 (ix86_option_override_internal): Adjust for
1134 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1135 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1136 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1137 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1138 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1139 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1140 def_builtin2 for CET builtins.
1141 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1142 up formatting in IX86_BUILTIN_RDPID code.
1143 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1144 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1145 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1146 ix86_isa_flags2 to ix86_isa_flags.
1147 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1149 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1150 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1151 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1153 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1154 Or in OPTION_MASK_ISA_AVX512F_SET.
1155 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1156 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1157 (ix86_handle_option): Adjust for
1158 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1159 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1161 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1162 Alan Hayward <alan.hayward@arm.com>
1163 David Sherwood <david.sherwood@arm.com>
1165 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1166 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1167 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1168 Track polynomial offsets.
1170 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1171 Alan Hayward <alan.hayward@arm.com>
1172 David Sherwood <david.sherwood@arm.com>
1174 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1175 Treat drb->init as a poly_int. Fail if its misalignment wrt
1176 vector_alignment isn't known.
1178 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1179 Alan Hayward <alan.hayward@arm.com>
1180 David Sherwood <david.sherwood@arm.com>
1182 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1184 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1185 (vect_loop_versioning): Take the loop versioning threshold as a
1187 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1188 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1189 versioning_threshold.
1190 (vect_analyze_loop_2): Compute the loop versioning threshold
1191 whenever loop versioning is needed, and store it in the new
1192 field rather than combining it with the cost model threshold.
1193 (vect_transform_loop): Update call to vect_loop_versioning.
1194 Try to combine the loop versioning and cost thresholds here.
1196 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1197 Alan Hayward <alan.hayward@arm.com>
1198 David Sherwood <david.sherwood@arm.com>
1200 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1201 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1202 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1203 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1204 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1205 (group_compare_offset): Update accordingly.
1206 (split_small_address_groups_p): Likewise.
1207 (record_use): Take addr_offset as a poly_uint64 rather than
1208 an unsigned HOST_WIDE_INT.
1209 (strip_offset): Return the offset as a poly_uint64 rather than
1210 an unsigned HOST_WIDE_INT.
1211 (record_group_use, split_address_groups): Track polynomial offsets.
1212 (add_iv_candidate_for_use): Likewise.
1213 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1214 than a HOST_WIDE_INT.
1215 (strip_offset_1): Return the offset as a poly_int64 rather than
1218 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1219 Alan Hayward <alan.hayward@arm.com>
1220 David Sherwood <david.sherwood@arm.com>
1222 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1223 rather than a HOST_WIDE_INT.
1224 * tree.c (get_binfo_at_offset): Likewise.
1226 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1227 Alan Hayward <alan.hayward@arm.com>
1228 David Sherwood <david.sherwood@arm.com>
1230 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1231 rather than a HOST_WIDE_INT.
1232 * tree-sra.c (build_ref_for_offset): Likewise.
1234 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1235 Alan Hayward <alan.hayward@arm.com>
1236 David Sherwood <david.sherwood@arm.com>
1238 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1240 * tree.c (mem_ref_offset): Likewise.
1241 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1242 * builtins.c (get_object_alignment_2): Likewise.
1243 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1244 * gimple-fold.c (get_base_constructor): Likewise.
1245 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1246 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1248 * ipa-polymorphic-call.c
1249 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1250 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1251 (get_ancestor_addr_info): Likewise.
1252 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1253 * match.pd: Likewise.
1254 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1255 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1256 * tree-eh.c (tree_could_trap_p): Likewise.
1257 * tree-object-size.c (addr_object_size): Likewise.
1258 * tree-ssa-address.c (copy_ref_info): Likewise.
1259 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1260 (indirect_refs_may_alias_p): Likewise.
1261 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1262 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1263 (non_rewritable_mem_ref_base): Likewise.
1264 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1265 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1266 * varasm.c (decode_addr_const): Likewise.
1268 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1269 Alan Hayward <alan.hayward@arm.com>
1270 David Sherwood <david.sherwood@arm.com>
1272 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1273 offsets for MEM_REFs.
1275 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1276 Alan Hayward <alan.hayward@arm.com>
1277 David Sherwood <david.sherwood@arm.com>
1279 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1280 as a poly_uint64 rather than an unsigned int.
1281 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1283 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1284 Alan Hayward <alan.hayward@arm.com>
1285 David Sherwood <david.sherwood@arm.com>
1287 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1289 (decode_addr_const): Update accordingly.
1291 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1292 Alan Hayward <alan.hayward@arm.com>
1293 David Sherwood <david.sherwood@arm.com>
1295 * tree.h (bit_field_size, bit_field_offset): New functions.
1296 * hsa-gen.c (gen_hsa_addr): Use them.
1297 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1298 (simplify_vector_constructor): Likewise.
1299 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1300 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1301 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1302 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1305 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1306 Alan Hayward <alan.hayward@arm.com>
1307 David Sherwood <david.sherwood@arm.com>
1309 * expr.h (emit_group_load, emit_group_load_into_temps)
1310 (emit_group_store): Take the size as a poly_int64 rather than an int.
1311 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1312 (emit_group_load_into_temp, emit_group_store): Likewise.
1314 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1315 Alan Hayward <alan.hayward@arm.com>
1316 David Sherwood <david.sherwood@arm.com>
1318 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1319 an unsigned int to a poly_uint64.
1320 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1321 sizes as poly_uint64s rather than unsigned ints.
1322 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1325 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1326 Alan Hayward <alan.hayward@arm.com>
1327 David Sherwood <david.sherwood@arm.com>
1329 * lra-constraints.c (emit_inc): Change inc_amount from an int
1332 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1333 Alan Hayward <alan.hayward@arm.com>
1334 David Sherwood <david.sherwood@arm.com>
1336 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1338 (add_stack_var, stack_var_cmp, partition_stack_vars)
1339 (dump_stack_var_partition): Update accordingly.
1340 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1341 than a HOST_WIDE_INT.
1342 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1343 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1344 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1345 rather than a HOST_WIDE_INT.
1347 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1348 Alan Hayward <alan.hayward@arm.com>
1349 David Sherwood <david.sherwood@arm.com>
1351 * target.def (return_pops_args): Treat both the input and output
1352 sizes as poly_int64s rather than HOST_WIDE_INTS.
1353 * targhooks.h (default_return_pops_args): Update accordingly.
1354 * targhooks.c (default_return_pops_args): Likewise.
1355 * doc/tm.texi: Regenerate.
1356 * emit-rtl.h (incoming_args): Change pops_args, size and
1357 outgoing_args_size from int to poly_int64_pod.
1358 * function.h (expr_status): Change x_pending_stack_adjust and
1359 x_stack_pointer_delta from int to poly_int64.
1360 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1361 (ARGS_SIZE_RTX): Update accordingly.
1362 * calls.c (highest_outgoing_arg_in_use): Change from int to
1364 (stack_usage_watermark, stored_args_watermark): New variables.
1365 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1366 (emit_call_1): Change the stack_size and rounded_stack_size
1367 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1369 (save_fixed_argument_area): Check stack_usage_watermark.
1370 (initialize_argument_information): Change old_pending_adj from
1371 a HOST_WIDE_INT * to a poly_int64_pod *.
1372 (compute_argument_block_size): Return the size as a poly_int64
1374 (finalize_must_preallocate): Track polynomial argument sizes.
1375 (compute_argument_addresses): Likewise.
1376 (internal_arg_pointer_based_exp): Track polynomial offsets.
1377 (mem_overlaps_already_clobbered_arg_p): Rename to...
1378 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1379 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1380 Check stored_args_used_watermark.
1381 (load_register_parameters): Update accordingly.
1382 (check_sibcall_argument_overlap_1): Likewise.
1383 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1384 args size as a poly_int64 rather than an int. Return a bool
1385 indicating whether the optimization was possible and return
1386 the new adjustment by reference.
1387 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1388 Update stored_args_watermark.
1389 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1390 (expand_call): Likewise. Maintain stack_usage_watermark and
1391 stored_args_watermark. Update calls to
1392 combine_pending_stack_adjustment_and_call.
1393 (emit_library_call_value_1): Handle polynomial argument sizes.
1394 Call stack_region_maybe_used_p and mark_stack_region_used.
1395 Maintain stack_usage_watermark.
1396 (store_one_arg): Likewise. Update call to
1397 mem_overlaps_already_clobbered_arg_p.
1398 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1400 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1401 * config/microblaze/microblaze.c (microblaze_function_prologue):
1403 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1404 TARGET_RETURN_POPS_ARGS interface.
1405 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1407 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1408 * config/i386/i386.c (ix86_return_pops_args): Update for new
1409 TARGET_RETURN_POPS_ARGS interface.
1410 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1411 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1412 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1413 TARGET_RETURN_POPS_ARGS interface.
1414 * config/vax/vax.c (vax_return_pops_args): Likewise.
1415 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1416 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1417 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1418 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1419 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1420 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1421 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1422 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1423 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1424 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1425 and x_stack_pointer_delta from int to poly_int64.
1426 * dojump.c (do_pending_stack_adjust): Update accordingly.
1427 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1428 stack_pointer_deltas.
1429 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1430 (pad_to_arg_alignment): Track polynomial offsets.
1431 (assign_parm_find_stack_rtl): Likewise.
1432 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1433 * toplev.c (output_stack_usage): Update reference to
1434 current_function_pushed_stack_size.
1436 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1437 Alan Hayward <alan.hayward@arm.com>
1438 David Sherwood <david.sherwood@arm.com>
1440 * function.c (in_arg_offset, var_offset, dynamic_offset)
1441 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1442 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1443 rather than a HOST_WIDE_INT.
1444 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1445 (instantiate_virtual_regs_in_insn): Likewise.
1447 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1448 Alan Hayward <alan.hayward@arm.com>
1449 David Sherwood <david.sherwood@arm.com>
1451 * rtl.h (get_args_size, add_args_size_note): New functions.
1452 (find_args_size_adjust): Return a poly_int64 rather than a
1454 (fixup_args_size_notes): Likewise. Make the same change to the
1455 end_args_size parameter.
1456 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1457 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1458 * calls.c (emit_call_1): Likewise.
1459 * explow.c (adjust_stack_1): Likewise.
1460 * cfgcleanup.c (old_insns_match_p): Update use of
1461 find_args_size_adjust.
1462 * combine.c (distribute_notes): Track polynomial arg sizes.
1463 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1464 end_true_args_size, beg_delay_args_size and end_delay_args_size
1465 from HOST_WIDE_INT to poly_int64.
1466 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1467 than a HOST_WIDE_INT.
1468 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1469 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1470 polynomial arg sizes.
1471 * emit-rtl.c (try_split): Use get_args_size.
1472 * recog.c (peep2_attempt): Likewise.
1473 * reload1.c (reload_as_needed): Likewise.
1474 * expr.c (find_args_size_adjust): Return the adjustment as a
1475 poly_int64 rather than a HOST_WIDE_INT.
1476 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1477 to a poly_int64 and change the return type in the same way.
1478 (emit_single_push_insn): Track polynomial arg sizes.
1480 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1481 Alan Hayward <alan.hayward@arm.com>
1482 David Sherwood <david.sherwood@arm.com>
1484 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1485 from HOST_WIDE_INT to poly_int64.
1486 * expr.c (push_block, emit_push_insn): Likewise.
1488 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1489 Alan Hayward <alan.hayward@arm.com>
1490 David Sherwood <david.sherwood@arm.com>
1492 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1494 (get_frame_size): Return the size as a poly_int64 rather than a
1496 (frame_offset_overflow): Take the offset as a poly_int64 rather
1497 than a HOST_WIDE_INT.
1498 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1499 (assign_stack_temp): Likewise for the size.
1500 * function.c (get_frame_size): Return a poly_int64 rather than
1502 (frame_offset_overflow): Take the offset as a poly_int64 rather
1503 than a HOST_WIDE_INT.
1504 (try_fit_stack_local): Take the start, length and size as poly_int64s
1505 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1506 rather than a HOST_WIDE_INT.
1507 (add_frame_space): Take the start and end as poly_int64s rather than
1509 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1510 (assign_stack_temp): Likewise for the size.
1511 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1513 (find_temp_slot_from_address): Handle polynomial offsets.
1514 (combine_temp_slots): Likewise.
1515 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1517 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1518 poly_int64 rather than a HOST_WIDE_INT.
1519 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1520 than a HOST_WIDE_INT.
1521 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1522 polynomial frame offsets.
1523 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1524 as a poly_int64 rather than an int.
1525 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1526 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1527 * config/v850/v850.c (compute_frame_size): Likewise.
1528 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1529 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1530 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1531 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1532 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1533 rather than a HOST_WIDE_INT.
1534 * explow.c (get_dynamic_stack_base): Likewise.
1535 * final.c (final_start_function): Use the constant lower bound
1536 of the frame size for -Wframe-larger-than.
1537 * ira.c (do_reload): Adjust for new get_frame_size return type.
1538 * lra.c (lra): Likewise.
1539 * reload1.c (reload): Likewise.
1540 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1541 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1542 * rtlanal.c (get_initial_register_offset): Return the offset as
1543 a poly_int64 rather than a HOST_WIDE_INT.
1545 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1546 Alan Hayward <alan.hayward@arm.com>
1547 David Sherwood <david.sherwood@arm.com>
1549 * reload1.c (elim_table): Change initial_offset, offset and
1550 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1551 (offsets_at): Change the target array's element type from
1552 HOST_WIDE_INT to poly_int64_pod.
1553 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1554 (elimination_costs_in_insn, update_eliminable_offsets)
1555 (verify_initial_elim_offsets, set_offsets_for_label)
1556 (init_eliminable_invariants): Update after above changes.
1558 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1559 Alan Hayward <alan.hayward@arm.com>
1560 David Sherwood <david.sherwood@arm.com>
1562 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1563 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1564 (decomposition): Change start and end from HOST_WIDE_INT
1566 (decompose, immune_p): Update accordingly.
1567 (find_inc_amount): Return a poly_int64 rather than an int.
1568 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
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 * tree.h (get_inner_reference): Return the bitsize and bitpos
1576 as poly_int64_pods rather than HOST_WIDE_INT.
1577 * fold-const.h (ptr_difference_const): Return the pointer difference
1578 as a poly_int64_pod rather than a HOST_WIDE_INT.
1579 * expr.c (get_inner_reference): Return the bitsize and bitpos
1580 as poly_int64_pods rather than HOST_WIDE_INT.
1581 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1583 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1584 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1585 (optimize_bit_field_compare): Update call to get_inner_reference.
1586 (decode_field_reference): Likewise.
1587 (fold_unary_loc): Track polynomial offsets and sizes.
1588 (split_address_to_core_and_offset): Return the bitpos as a
1589 poly_int64_pod rather than a HOST_WIDE_INT.
1590 (ptr_difference_const): Likewise for the pointer difference.
1591 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1592 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1593 * dbxout.c (dbxout_expand_expr): Likewise.
1594 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1595 (loc_list_from_tree_1, fortran_common): Likewise.
1596 * gimple-laddress.c (pass_laddress::execute): Likewise.
1597 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1598 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1599 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1600 * tree-affine.c (tree_to_aff_combination): Likewise.
1601 (get_inner_reference_aff): Likewise.
1602 * tree-data-ref.c (split_constant_offset_1): Likewise.
1603 (dr_analyze_innermost): Likewise.
1604 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1605 * tree-sra.c (ipa_sra_check_caller): Likewise.
1606 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1607 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1608 (instrument_bool_enum_load, instrument_object_size): Likewise.
1609 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1610 to get_inner_reference.
1611 * hsa-gen.c (gen_hsa_addr): Likewise.
1612 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1613 * tsan.c (instrument_expr): Likewise.
1614 * match.pd: Update call to ptr_difference_const.
1616 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1617 Alan Hayward <alan.hayward@arm.com>
1618 David Sherwood <david.sherwood@arm.com>
1620 * fold-const.c (fold_comparison): Track sizes and offsets as
1621 poly_int64s rather than HOST_WIDE_INTs when folding address
1624 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1625 Alan Hayward <alan.hayward@arm.com>
1626 David Sherwood <david.sherwood@arm.com>
1628 * expr.h (get_bit_range): Return the bitstart and bitend as
1629 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1630 as a poly_int64 rather than a HOST_WIDE_INT.
1631 * expr.c (get_bit_range): Likewise.
1632 (expand_assignment): Update call accordingly.
1633 * fold-const.c (optimize_bit_field_compare): Likewise.
1635 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1636 Alan Hayward <alan.hayward@arm.com>
1637 David Sherwood <david.sherwood@arm.com>
1639 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1640 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1641 as an optional final parameter.
1642 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1643 (aarch64_legitimate_address_p): Likewise.
1644 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1645 instead of an rtx code.
1646 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1647 (aarch64_legitimate_address_hook_p): Likewise.
1648 (aarch64_print_ldpstp_address): Likewise.
1649 (aarch64_print_operand_address): Likewise.
1650 (aarch64_address_cost): Likewise.
1651 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1652 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1653 (aarch64_mem_pair_lanes_operand): Likewise.
1655 2017-12-20 Richard Biener <rguenther@suse.de>
1657 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1658 update_stmt_if_modified.
1660 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1662 PR tree-optimization/83491
1663 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1664 before walking uses. Improve coding style and comments.
1666 2017-12-20 Tom de Vries <tom@codesourcery.com>
1668 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1670 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1673 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1675 (pass_ipa_free_fn_summary): Add small_p private data member,
1676 initialize to false in the ctor.
1677 (pass_ipa_free_fn_summary::clone,
1678 pass_ipa_free_fn_summary::set_pass_param,
1679 pass_ipa_free_fn_summary::gate): New methods.
1680 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1681 | TODO_dump_symtab if small_p.
1682 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1683 entry and add another instance of the pass with false parm after
1685 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1686 ipa_free_fn_summary here.
1688 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1690 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1692 2017-12-20 Martin Sebor <msebor@redhat.com>
1695 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1696 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1697 (expand_builtin_strncmp): Same.
1699 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1702 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1703 allowed debug stmts before labels.
1704 (expand_gimple_basic_block): Likewise.
1705 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1706 * gimple-iterator.h (gsi_after_labels): Likewise.
1707 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1708 rename reused variable, and simplify using gsi_move_before.
1709 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1710 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1711 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1712 (gimple_verify_flow_info, gimple_block_label): Likewise.
1713 (make_blocks): Move debug markers after adjacent labels.
1714 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1715 allowed debug insns outside blocks.
1716 * df-scan.c (df_insn_delete): Likewise.
1717 * lra-constraints.c (update_ebb_live_info): Likewise.
1718 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1719 (vt_initialize, delete_vta_debug_insns): Likewise.
1720 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1722 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1723 Alan Hayward <alan.hayward@arm.com>
1724 David Sherwood <david.sherwood@arm.com>
1726 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1727 (round_up_to_byte_boundary): Likewise.
1728 * expr.h (get_bit_range): Add temporary shim.
1729 * gimple-ssa-store-merging.c (store_operand_info): Change the
1730 bitsize, bitpos, bitregion_start and bitregion_end fields from
1731 unsigned HOST_WIDE_INT to poly_uint64.
1732 (merged_store_group): Likewise load_align_base.
1733 (compatible_load_p, compatible_load_p): Update accordingly.
1734 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1735 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1736 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1737 bitregion_start and bitregion_end as poly_uint64s rather than
1738 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1739 (handled_load): Take the bitsize, bitpos,
1740 bitregion_start and bitregion_end as poly_uint64s rather than
1741 unsigned HOST_WIDE_INTs.
1742 (pass_store_merging::process_store): Update call to
1743 mem_valid_for_store_merging.
1745 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1746 Alan Hayward <alan.hayward@arm.com>
1747 David Sherwood <david.sherwood@arm.com>
1749 * builtins.c (get_object_alignment_2): Track polynomial offsets
1750 and sizes. Update the alignment handling.
1752 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1753 Alan Hayward <alan.hayward@arm.com>
1754 David Sherwood <david.sherwood@arm.com>
1756 * tree.h (get_inner_reference): Add a version that returns the
1757 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1758 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1759 the case in which bitpos is not associated with the first byte.
1761 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1762 Alan Hayward <alan.hayward@arm.com>
1763 David Sherwood <david.sherwood@arm.com>
1765 * tree-affine.h (get_inner_reference_aff): Return the size as a
1767 * tree-affine.c (get_inner_reference_aff): Likewise.
1768 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1769 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1771 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1772 Alan Hayward <alan.hayward@arm.com>
1773 David Sherwood <david.sherwood@arm.com>
1775 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1776 HOST_WIDE_INT rather than a poly_int64.
1778 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1779 Alan Hayward <alan.hayward@arm.com>
1780 David Sherwood <david.sherwood@arm.com>
1782 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1783 HOST_WIDE_INT to poly_int64_pod.
1784 (perform_symbolic_merge): Update accordingly.
1785 (bswap_replace): Likewise.
1787 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1788 Alan Hayward <alan.hayward@arm.com>
1789 David Sherwood <david.sherwood@arm.com>
1791 * tree-affine.h (aff_tree::offset): Change from widest_int
1793 (wide_int_ext_for_comb): Delete.
1794 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1795 constants as poly_widest_int rather than widest_int.
1796 (aff_combination_constant_multiple_p): Return the multiplier
1797 as a poly_widest_int.
1798 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1800 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1801 version static and add an overload for poly_widest_int.
1802 (aff_combination_const, aff_combination_add_cst)
1803 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1804 the constants as poly_widest_int rather than widest_int.
1805 (tree_to_aff_combination): Generalize INTEGER_CST case to
1807 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1808 (aff_combination_add_product, aff_combination_mult): Handle
1810 (aff_combination_constant_multiple_p): Return the multiplier
1811 as a poly_widest_int.
1812 * tree-predcom.c (determine_offset): Return the offset as a
1814 (split_data_refs_to_components, suitable_component_p): Update
1816 (valid_initializer_p): Update call to
1817 aff_combination_constant_multiple_p.
1818 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1819 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1820 as a poly_int64 rather than a HOST_WIDE_INT.
1821 (get_address_cost): Handle polynomial offsets.
1822 (iv_elimination_compare_lt): Likewise.
1823 (rewrite_use_nonlinear_expr): Likewise.
1825 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1826 Alan Hayward <alan.hayward@arm.com>
1827 David Sherwood <david.sherwood@arm.com>
1829 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1830 as a poly_int64_pod rather than a HOST_WIDE_INT.
1831 (get_addr_base_and_unit_offset): Likewise.
1832 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1833 (get_addr_base_and_unit_offset): Likewise.
1834 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1835 to poly_int64 in example.
1836 * fold-const.c (fold_binary_loc): Update call to
1837 get_addr_base_and_unit_offset.
1838 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1839 (maybe_canonicalize_mem_ref_addr): Likewise.
1840 (gimple_fold_stmt_to_constant_1): Likewise.
1841 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1843 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1844 * match.pd: Likewise.
1845 * omp-low.c (lower_omp_target): Likewise.
1846 * tree-sra.c (build_ref_for_offset): Likewise.
1847 (build_debug_ref_for_model): Likewise.
1848 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1849 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1850 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1851 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1852 (constant_pointer_difference): Likewise.
1853 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1854 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1855 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1856 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1857 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1858 (set_ssa_val_to): Likewise.
1859 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1860 (maybe_diag_stxncpy_trunc): Likewise.
1861 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1862 * tree.c (build_simple_mem_ref_loc): Likewise.
1863 (array_at_struct_end_p): Likewise.
1865 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1866 Alan Hayward <alan.hayward@arm.com>
1867 David Sherwood <david.sherwood@arm.com>
1869 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1870 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1871 (get_ref_base_and_extent_hwi): Declare.
1872 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1873 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1874 (get_ref_base_and_extent_hwi): New function.
1875 * cfgexpand.c (expand_debug_expr): Update call to
1876 get_ref_base_and_extent.
1877 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1878 * gimple-fold.c (get_base_constructor): Return the offset as a
1879 poly_int64_pod rather than a HOST_WIDE_INT.
1880 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1881 * ipa-polymorphic-call.c
1882 (ipa_polymorphic_call_context::set_by_invariant)
1883 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1884 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1885 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1886 rather than get_ref_base_and_extent.
1887 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1888 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1889 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1891 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1892 call to get_ref_base_and_extent.
1893 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1894 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1895 (stmt_kills_ref_p): Likewise.
1896 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1897 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1899 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1900 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1901 when calling native_encode_expr.
1902 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1903 call to get_ref_base_and_extent.
1904 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1905 get_ref_base_and_extent.
1906 * var-tracking.c (track_expr_p): Likewise.
1908 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1909 Alan Hayward <alan.hayward@arm.com>
1910 David Sherwood <david.sherwood@arm.com>
1912 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1913 HOST_WIDE_INT to poly_int64_pod.
1914 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1915 polynomail parameter offsets.
1917 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1918 Alan Hayward <alan.hayward@arm.com>
1919 David Sherwood <david.sherwood@arm.com>
1921 * gengtype.c (main): Handle poly_int64_pod.
1922 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1923 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1924 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1925 from HOST_WIDE_INT to poly_int64_pod.
1926 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1927 (copy_cfa): New function.
1928 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1929 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1930 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1931 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1932 (initial_return_save): Treat offsets as poly_ints.
1933 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1934 offset is nonconstant.
1935 (reg_save): Take the offset as a poly_int64. Fall back to
1936 DW_CFA_expression for nonconstant offsets.
1937 (queue_reg_save): Take the offset as a poly_int64.
1938 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1940 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1941 Alan Hayward <alan.hayward@arm.com>
1942 David Sherwood <david.sherwood@arm.com>
1944 * rtl.h (operand_subword, operand_subword_force): Take the offset
1945 as a poly_uint64 an unsigned int.
1946 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1948 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1949 Alan Hayward <alan.hayward@arm.com>
1950 David Sherwood <david.sherwood@arm.com>
1952 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1953 'p' format code. Use INT_LIST rather than SUBREG as the example of
1954 a code with an XINT and an XEXP. Remove the implication that
1955 accessing an rtx field using XINT is expected to work.
1956 * rtl.def (SUBREG): Change format from "ei" to "ep".
1957 * rtl.h (rtunion::rt_subreg): New field.
1958 (XCSUBREG): New macro.
1959 (SUBREG_BYTE): Use it.
1960 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1961 Update constructor accordingly.
1962 (subreg_shape::operator ==): Update accordingly.
1963 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1964 than an unsigned int.
1965 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1966 a poly_uint64 rather than an unsigned int.
1967 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1968 than an unsigned int.
1969 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1970 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1971 an unsigned int. Take the sizes as poly_uint64s.
1972 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1973 an unsigned int. Take the shift as a poly_uint64 rather than
1975 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1976 as a poly_uint64 rather than an unsigned int.
1977 (simplify_subreg_regno): Likewise.
1978 (byte_lowpart_offset): Return the memory offset as a poly_int64
1980 (subreg_memory_offset): Likewise. Take the subreg offset as a
1981 poly_uint64 rather than an unsigned int.
1982 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1983 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1984 poly_uint64 rather than an unsigned int.
1985 * rtl.c (rtx_format): Describe 'p' in comment.
1986 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1987 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1988 offset as a poly_uint64 rather than an unsigned int.
1989 (byte_lowpart_offset): Return the memory offset as a poly_int64
1991 (subreg_memory_offset): Likewise. Take the subreg offset as a
1992 poly_uint64 rather than an unsigned int.
1993 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1994 mode sizes as poly_uint64s rather than unsigned ints. Return a
1995 poly_uint64 rather than an unsigned int.
1996 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1997 (copy_insn_1): Handle 'p'.
1998 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1999 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
2000 an unsigned int. Return the shift in the same way.
2001 (subreg_lsb): Return the shift as a poly_uint64 rather than an
2003 (subreg_size_offset_from_lsb): Take the sizes and shift as
2004 poly_uint64s rather than unsigned ints. Return the offset as
2006 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2007 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2009 * rtlhash.c (add_rtx): Handle 'p'.
2010 * genemit.c (gen_exp): Likewise.
2011 * gengenrtl.c (type_from_format, gendef): Likewise.
2012 * gensupport.c (subst_pattern_match, get_alternatives_number)
2013 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2014 (subst_dup): Likewise.
2015 * gengtype.c (adjust_field_rtx_def): Likewise.
2016 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2017 (match_pattern_2): Likewise.
2018 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2019 (rtx_test::subreg_field): New function.
2020 (operator ==, safe_to_hoist_p, transition_parameter_type)
2021 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2022 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2023 * genpeep.c (match_rtx): Likewise.
2024 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2025 (rtx_writer::print_rtx_operand): Handle 'p'.
2026 (print_value): Handle SUBREG.
2027 * read-rtl.c (apply_int_iterator): Likewise.
2028 (rtx_reader::read_rtx_operand): Handle 'p'.
2029 * alias.c (rtx_equal_for_memref_p): Likewise.
2030 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2031 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2033 * calls.c (expand_call): Likewise.
2034 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2035 (make_extraction, gen_lowpart_for_combine): Likewise.
2036 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2038 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2039 rather than an unsigned int. Treat subreg offsets as poly_ints.
2040 (exp_equiv_p): Handle 'p'.
2041 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2042 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2043 * dse.c (find_shift_sequence): Likewise.
2044 * dwarf2out.c (rtl_for_decl_location): Likewise.
2045 * expmed.c (extract_low_bits): Likewise.
2046 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2047 (expand_expr_real_2): Likewise.
2048 * final.c (alter_subreg): Likewise.
2049 (leaf_renumber_regs_insn): Handle 'p'.
2050 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2051 Treat subreg offsets as poly_ints.
2052 * fwprop.c (forward_propagate_and_simplify): Likewise.
2053 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2054 * ira.c (get_subreg_tracking_sizes): Likewise.
2055 * ira-conflicts.c (go_through_subreg): Likewise.
2056 * ira-lives.c (process_single_reg_class_operands): Likewise.
2057 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2058 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2059 as a poly_uint64 rather than an unsigned int.
2060 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2061 subreg offsets as poly_ints.
2062 * lra-constraints.c (operands_match_p): Handle 'p'.
2063 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2064 * lra-spills.c (assign_mem_slot): Likewise.
2065 * postreload.c (move2add_valid_value_p): Likewise.
2066 * recog.c (general_operand, indirect_operand): Likewise.
2067 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2068 (copyprop_hardreg_forward_1): Likewise.
2069 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2070 (record_subregs_of_mode): Likewise.
2071 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2072 * reload.c (operands_match_p): Handle 'p'.
2073 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2074 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2075 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2076 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2077 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2078 (simplify_const_poly_int_tests<N>::run): Likewise.
2079 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2080 a poly_uint64 rather than an unsigned int.
2081 * valtrack.c (debug_lowpart_subreg): Likewise.
2082 * var-tracking.c (var_lowpart): Likewise.
2083 (loc_cmp): Handle 'p'.
2085 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2086 Alan Hayward <alan.hayward@arm.com>
2087 David Sherwood <david.sherwood@arm.com>
2089 * ira.c (get_subreg_tracking_sizes): New function.
2090 (init_live_subregs): Take an integer size rather than a register.
2091 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2092 to init_live_subregs.
2094 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2095 Alan Hayward <alan.hayward@arm.com>
2096 David Sherwood <david.sherwood@arm.com>
2098 * expr.c (store_constructor_field): Change bitsize from a
2099 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2100 HOST_WIDE_INT to a poly_int64.
2101 (store_constructor): Change size from a HOST_WIDE_INT to
2103 (store_field): Likewise bitsize and bitpos.
2105 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2106 Alan Hayward <alan.hayward@arm.com>
2107 David Sherwood <david.sherwood@arm.com>
2109 * expmed.h (store_bit_field): Change bitregion_start and
2110 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2111 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2112 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2113 (store_fixed_bit_field, store_split_bit_field): Likewise.
2114 * expr.c (store_constructor_field, store_field): Likewise.
2115 (optimize_bitfield_assignment_op): Likewise. Make the same change
2116 to bitsize and bitpos.
2117 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2118 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2119 same change in the constructor arguments.
2120 (get_best_mode): Change bitregion_start and bitregion_end from
2121 unsigned HOST_WIDE_INT to poly_uint64.
2122 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2123 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2125 (bit_field_mode_iterator::next_mode): Update for new types
2126 of m_bitregion_start and m_bitregion_end.
2127 (get_best_mode): Change bitregion_start and bitregion_end from
2128 unsigned HOST_WIDE_INT to poly_uint64.
2130 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2131 Alan Hayward <alan.hayward@arm.com>
2132 David Sherwood <david.sherwood@arm.com>
2134 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2135 accepts poly_uint64 offsets.
2136 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2137 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2138 * expmed.c (lowpart_bit_field_p): Likewise.
2139 (extract_bit_field_as_subreg): New function, split out from...
2140 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2141 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2142 extractions, check that BITSIZE matches the size of the extracted
2143 value and that BITNUM is an exact multiple of that size.
2144 If all else fails, try forcing the value into memory if
2145 BITNUM is variable, and adjusting the address so that the
2146 offset is constant. Split the part that can only handle constant
2147 bitsize and bitnum out into...
2148 (extract_integral_bit_field): ...this new function.
2149 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2150 rather than unsigned HOST_WIDE_INTs.
2152 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2153 Alan Hayward <alan.hayward@arm.com>
2154 David Sherwood <david.sherwood@arm.com>
2156 * expmed.h (store_bit_field): Take bitsize and bitnum as
2157 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2158 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2159 that returns the byte size.
2160 (store_bit_field_1): Take bitsize and bitnum as
2161 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2162 to simple_mem_bitfield_p. Split the part that can only handle
2163 constant bitsize and bitnum out into...
2164 (store_integral_bit_field): ...this new function.
2165 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2166 than unsigned HOST_WIDE_INTs.
2167 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2169 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2170 Alan Hayward <alan.hayward@arm.com>
2171 David Sherwood <david.sherwood@arm.com>
2173 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2174 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2176 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2177 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2178 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2179 offset as a poly_int64 rather than an int.
2180 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2181 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2182 * lra-constraints.c (equiv_address_substitution): Track offsets
2184 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2185 (curr_insn_transform): Handle the new form of sp_offset.
2186 * lra-eliminations.c (lra_elim_table): Change previous_offset
2187 and offset from HOST_WIDE_INT to poly_int64.
2188 (print_elim_table, update_reg_eliminate): Update accordingly.
2189 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2190 (get_elimination): Update accordingly.
2191 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2192 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2193 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2194 poly_int64 offsets generally.
2195 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2196 (mark_not_eliminable, init_elimination): Update accordingly.
2197 (remove_reg_equal_offset_note): Return a bool and pass the new
2198 offset back by pointer as a poly_int64.
2199 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2200 rather than a HOST_WIDE_INT.
2201 (do_remat): Track offsets poly_int64s.
2202 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2204 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2205 Alan Hayward <alan.hayward@arm.com>
2206 David Sherwood <david.sherwood@arm.com>
2208 * rtl.h (mem_attrs): Add a default constructor. Change size and
2209 offset from HOST_WIDE_INT to poly_int64.
2210 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2211 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2212 (widen_memory_access): Take the sizes and offsets as poly_int64s
2213 rather than HOST_WIDE_INTs.
2214 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2215 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2217 (adjust_offset_for_component_ref): Change the offset from a
2218 HOST_WIDE_INT to a poly_int64.
2219 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2220 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2221 * dce.c (find_call_stack_args): Likewise.
2222 * dse.c (record_store): Likewise.
2223 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2224 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2225 * read-rtl-function.c (test_loading_mem): Likewise.
2226 * rtlanal.c (may_trap_p_1): Likewise.
2227 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2228 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2229 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2230 (mem_attrs::mem_attrs): New function.
2231 (set_mem_attributes_minus_bitpos): Change bitpos from a
2232 HOST_WIDE_INT to poly_int64.
2233 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2234 (clear_mem_offset, clear_mem_size, change_address)
2235 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2236 initialize mem_attrs.
2237 (set_mem_offset, set_mem_size, adjust_address_1)
2238 (adjust_automodify_address_1, offset_address, widen_memory_access):
2239 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2241 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2242 Alan Hayward <alan.hayward@arm.com>
2243 David Sherwood <david.sherwood@arm.com>
2245 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2246 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2247 rather than 0 to represent an unknown size. Assert that the size
2248 is known when the mode isn't BLKmode.
2249 (may_trap_p_1): Use -1 for unknown sizes.
2250 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2252 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2253 Alan Hayward <alan.hayward@arm.com>
2254 David Sherwood <david.sherwood@arm.com>
2256 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2257 to poly_int64. Update commentary for positions_needed.large.
2258 (read_info_type): Change offset and width from HOST_WIDE_INT
2260 (set_usage_bits): Likewise.
2261 (canon_address): Return the offset as a poly_int64 rather than
2262 a HOST_WIDE_INT. Use strip_offset_and_add.
2263 (set_all_positions_unneeded, any_positions_needed_p): Use
2264 positions_needed.large to track stores with non-constant widths.
2265 (all_positions_needed_p): Likewise. Take the offset and width
2266 as poly_int64s rather than ints. Assert that rhs is nonnull.
2267 (record_store): Cope with non-constant offsets and widths.
2268 Nullify the rhs of an earlier store if we can't tell which bytes
2270 (find_shift_sequence): Take the access_size and shift as poly_int64s
2272 (get_stored_val): Take the read_offset and read_width as poly_int64s
2273 rather than HOST_WIDE_INTs.
2274 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2275 non-constant offsets and widths.
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 * inchash.h (inchash::hash::add_poly_int): New function.
2282 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2283 Use poly_int64 rather than HOST_WIDE_INT.
2284 (ao_ref::max_size_known_p): New function.
2285 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2286 rather than HOST_WIDE_INT.
2287 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2288 to temporaries until its interface is adjusted to match.
2289 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2290 (aliasing_component_refs_p, decl_refs_may_alias_p)
2291 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2292 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2293 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2295 * alias.c (ao_ref_from_mem): Likewise.
2296 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2297 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2298 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2299 (maybe_trim_complex_store, maybe_trim_constructor_store)
2300 (live_bytes_read, dse_classify_store): Likewise.
2301 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2302 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2303 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2304 (vn_reference_lookup_3): Likewise.
2305 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2307 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2308 Alan Hayward <alan.hayward@arm.com>
2309 David Sherwood <david.sherwood@arm.com>
2311 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2312 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2313 instead of ranges_overlap_p.
2315 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2316 Alan Hayward <alan.hayward@arm.com>
2317 David Sherwood <david.sherwood@arm.com>
2319 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2320 sizes as poly_int64s rather than HOST_WIDE_INTs.
2322 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2323 Alan Hayward <alan.hayward@arm.com>
2324 David Sherwood <david.sherwood@arm.com>
2326 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2327 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2328 * gimple-fold.c (fold_ctor_reference): Likewise.
2330 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2331 Alan Hayward <alan.hayward@arm.com>
2332 David Sherwood <david.sherwood@arm.com>
2334 * target.def (dwarf_poly_indeterminate_value): New hook.
2335 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2336 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2337 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2338 * doc/tm.texi: Regenerate.
2339 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2340 offset as a poly_int64.
2341 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2342 offset as a poly_int64.
2343 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2344 Take the offset as a poly_int64.
2345 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2346 (frame_pointer_fb_offset): Change to a poly_int64.
2347 (int_loc_descriptor): Take the offset as a poly_int64. Use
2348 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2349 (based_loc_descr): Take the offset as a poly_int64.
2350 Use strip_offset_and_add to handle (plus X (const)).
2351 Use new_reg_loc_descr instead of an open-coded version of the
2352 previous implementation.
2353 (mem_loc_descriptor): Handle CONST_POLY_INT.
2354 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2355 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2357 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2358 Alan Hayward <alan.hayward@arm.com>
2359 David Sherwood <david.sherwood@arm.com>
2361 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2362 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2363 * inchash.h (inchash::hash::add_poly_hwi): New function.
2364 * gengtype.c (main): Register poly_int64.
2365 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2366 offset as a poly_int.
2367 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2368 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2369 offset as a poly_int64.
2370 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2371 * print-rtl.c (print_poly_int): New function.
2372 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2374 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2376 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2377 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2378 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2379 Enforce track_offset_p for parts of a PARALLEL.
2380 (vt_add_function_parameter): Use const_offset for the final
2381 offset to track. Use get_tracked_reg_offset for the parts
2384 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2385 Alan Hayward <alan.hayward@arm.com>
2386 David Sherwood <david.sherwood@arm.com>
2388 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2389 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2390 * doc/tm.texi: Regenerate.
2391 * hooks.h (hook_bool_uint_uint_true): Delete.
2392 (hook_bool_puint64_puint64_true): Declare.
2393 * hooks.c (hook_bool_uint_uint_true): Delete.
2394 (hook_bool_puint64_puint64_true): New function.
2395 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2396 instead of unsigned ints.
2397 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2398 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2400 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2401 Alan Hayward <alan.hayward@arm.com>
2402 David Sherwood <david.sherwood@arm.com>
2404 * optabs.h (expand_operand): Add an int_value field.
2405 (create_expand_operand): Add an int_value parameter and use it
2406 to initialize the new expand_operand field.
2407 (create_integer_operand): Replace with a declaration of a function
2408 that accepts poly_int64s. Move the implementation to...
2409 * optabs.c (create_integer_operand): ...here.
2410 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2411 the mode preserves the value of int_value, instead of calling
2412 const_int_operand on the rtx. Use gen_int_mode to generate
2415 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2416 Alan Hayward <alan.hayward@arm.com>
2417 David Sherwood <david.sherwood@arm.com>
2419 * dumpfile.h (dump_dec): Declare.
2420 * dumpfile.c (dump_dec): New function.
2421 * pretty-print.h (pp_wide_integer): Turn into a function and
2422 declare a poly_int version.
2423 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2425 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2426 Alan Hayward <alan.hayward@arm.com>
2427 David Sherwood <david.sherwood@arm.com>
2429 * doc/generic.texi (POLY_INT_CST): Document.
2430 * tree.def (POLY_INT_CST): New tree code.
2431 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2432 * tree-core.h (tree_poly_int_cst): New struct.
2433 (tree_node): Add a poly_int_cst field.
2434 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2435 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2436 instead of a wide_int_ref.
2437 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2439 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2440 instead of an unsigned HOST_WIDE_INT.
2441 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2442 (ptrdiff_tree_p): Declare.
2443 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2444 extern inline implementations if the target doesn't use POLY_INT_CST.
2445 (poly_int_tree_p): New function.
2446 (wi::unextended_tree): New class.
2447 (wi::int_traits <unextended_tree>): New override.
2448 (wi::extended_tree): Add a default constructor.
2449 (wi::extended_tree::get_tree): New function.
2450 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2451 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2452 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2453 (wi::tree_to_poly_wide_ref): New typedefs.
2454 (wi::ints_for): Provide overloads for extended_tree and
2456 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2457 (wi::to_wide): New functions.
2458 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2459 * tree.c (poly_int_cst_hasher): New struct.
2460 (poly_int_cst_hash_table): New variable.
2461 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2462 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2464 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2465 (init_ttree): Initialize poly_int_cst_hash_table.
2466 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2467 a poly_int64 instead of a HOST_WIDE_INT.
2468 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2469 instead of an unsigned HOST_WIDE_INT.
2470 (wide_int_to_tree): Rename to...
2471 (wide_int_to_tree_1): ...this.
2472 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2473 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2474 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2475 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2476 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2477 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2479 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2480 (streamer_read_tree_body): Likewise.
2481 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2482 (streamer_write_tree_body): Likewise.
2483 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2484 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2485 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2486 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2487 * gimple-expr.h (is_gimple_constant): Likewise.
2488 * gimplify.c (maybe_with_size_expr): Likewise.
2489 * print-tree.c (print_node): Likewise.
2490 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2491 * tree-pretty-print.c (dump_generic_node): Likewise.
2492 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2493 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2494 * tree-vrp.c (compare_values_warnv): Likewise.
2495 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2496 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2497 (force_expr_to_var_cost): Likewise.
2498 * tree-ssa-loop.c (for_each_index): Likewise.
2499 * fold-const.h (build_invariant_address, size_int_kind): Take a
2500 poly_int64 instead of a HOST_WIDE_INT.
2501 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2502 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2504 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2505 (int_const_binop_2): New function, split out from...
2506 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2507 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2508 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2509 * gimple-ssa-strength-reduction.c (slsr_process_add)
2510 (slsr_process_mul): Check for INTEGER_CSTs before using them
2512 * stor-layout.c (bits_from_bytes): New function.
2513 (bit_from_pos): Use it.
2514 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2515 by BITS_PER_UNIT to get the TYPE_SIZE.
2516 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2517 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2519 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2520 Alan Hayward <alan.hayward@arm.com>
2521 David Sherwood <david.sherwood@arm.com>
2523 * doc/rtl.texi (const_poly_int): Document. Also document the
2524 rtl sharing behavior.
2525 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2526 * rtl.h (const_poly_int_def): New struct.
2527 (rtx_def::u): Add a cpi field.
2528 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2529 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2530 (wi::rtx_to_poly_wide_ref): New typedef
2531 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2532 (poly_int_rtx_p): New functions.
2533 (trunc_int_for_mode): Declare a poly_int64 version.
2534 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2535 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2537 (strip_offset): Declare.
2538 (strip_offset_and_add): New function.
2539 * rtl.def (CONST_POLY_INT): New rtx code.
2540 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2541 (shared_const_p): Use poly_int_rtx_p.
2542 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2544 (gen_int_shift_amount): Likewise.
2545 * emit-rtl.c (const_poly_int_hasher): New class.
2546 (const_poly_int_htab): New variable.
2547 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2548 (const_poly_int_hasher::hash): New function.
2549 (const_poly_int_hasher::equal): Likewise.
2550 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2551 (immed_wide_int_const): Rename to...
2552 (immed_wide_int_const_1): ...this and make static.
2553 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2554 instead of a wide_int_ref.
2555 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2556 (gen_lowpart_common): Handle CONST_POLY_INT.
2557 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2558 * cselib.c (cselib_hash_rtx): Likewise.
2559 * dwarf2out.c (const_ok_for_output_1): Likewise.
2560 * expr.c (convert_modes): Likewise.
2561 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2562 * rtlhash.c (add_rtx): Likewise.
2563 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2564 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2565 Handle existing CONST_POLY_INT rtxes.
2566 * expmed.h (expand_shift): Take a poly_int64 instead of a
2568 * expmed.c (expand_shift): Likewise.
2569 * rtlanal.c (strip_offset): New function.
2570 (commutative_operand_precedence): Give CONST_POLY_INT the same
2571 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2573 * rtl-tests.c (const_poly_int_tests): New struct.
2574 (rtl_tests_c_tests): Use it.
2575 * simplify-rtx.c (simplify_const_unary_operation): Handle
2577 (simplify_const_binary_operation): Likewise.
2578 (simplify_binary_operation_1): Fold additions of symbolic constants
2579 and CONST_POLY_INTs.
2580 (simplify_subreg): Handle extensions and truncations of
2582 (simplify_const_poly_int_tests): New struct.
2583 (simplify_rtx_c_tests): Use it.
2584 * wide-int.h (storage_ref): Add default constructor.
2585 (wide_int_ref_storage): Likewise.
2586 (trailing_wide_ints): Use GTY((user)).
2587 (trailing_wide_ints::operator[]): Add a const version.
2588 (trailing_wide_ints::get_precision): New function.
2589 (trailing_wide_ints::extra_size): Likewise.
2591 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2592 Alan Hayward <alan.hayward@arm.com>
2593 David Sherwood <david.sherwood@arm.com>
2595 * emit-rtl.h (gen_int_shift_amount): Declare.
2596 * emit-rtl.c (gen_int_shift_amount): New function.
2597 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2599 * calls.c (shift_return_value): Likewise.
2600 * cse.c (fold_rtx): Likewise.
2601 * dse.c (find_shift_sequence): Likewise.
2602 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2603 (expand_shift, expand_smod_pow2): Likewise.
2604 * lower-subreg.c (shift_cost): Likewise.
2605 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2606 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2607 (expand_vec_perm_var): Likewise.
2608 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2609 (simplify_binary_operation_1): Likewise.
2610 * combine.c (try_combine, find_split_point, force_int_to_mode)
2611 (simplify_shift_const_1, simplify_shift_const): Likewise.
2612 (change_zero_ext): Likewise. Use simplify_gen_binary.
2614 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2616 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2618 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2620 * doc/invoke.texi (ARM Options): Document accepted extension options
2621 for -march=armv8.3-a.
2623 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2626 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2627 or --with-float=softfp, set the default CPU to arm10e.
2629 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2631 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2632 * config/visium/predicates.md (const_shift_operand): Likewise.
2633 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2634 (visium_legitimize_reload_address): Likewise.
2636 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2639 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2641 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2644 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2646 2017-12-20 Julia Koval <julia.koval@intel.com>
2648 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2649 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2650 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2651 * config.gcc: Include vpclmulqdqintrin.h.
2652 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2653 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2655 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2656 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2657 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2658 * config/i386/i386.c (isa2_opts): Add -mcx16.
2659 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2660 (ix86_option_override_internal): Move mcx16 to flags2.
2661 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2662 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2663 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2664 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2665 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2666 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2667 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2668 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2669 * doc/invoke.texi: Add -mvpclmulqdq.
2671 2017-12-20 Tom de Vries <tom@codesourcery.com>
2674 * config/i386/i386.c (ix86_static_chain): Move
2675 DECL_STATIC_CHAIN test ...
2676 * calls.c (rtx_for_static_chain): ... here. New function.
2677 * calls.h (rtx_for_static_chain): Declare.
2678 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2679 instead of targetm.calls.static_chain.
2680 * df-scan.c (df_get_entry_block_def_set): Same.
2682 2017-12-19 Tom de Vries <tom@codesourcery.com>
2684 PR tree-optimization/83493
2685 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2688 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2690 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2691 inform with hardcoded english plural handling.
2693 2017-12-18 Jeff Law <law@redhat.com>
2695 PR tree-optimization/83477
2696 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2697 a non-virtual PHI, always push a new range.
2699 2017-12-19 Martin Sebor <msebor@redhat.com>
2702 * builtins.c (compute_objsize): Handle non-constant offsets.
2704 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2706 PR tree-optimization/83444
2707 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2708 character load case, if get_stridx on MEM_REF's operand doesn't
2709 look usable, retry with get_addr_stridx.
2711 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2714 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2715 (variable_tracking_main_1): Keep markers even when VTA fails.
2718 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2719 even if there are markers before it.
2720 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2722 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2724 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2725 typical order conditions.
2726 * sel-sched.c (extract_new_fences_from): Likewise.
2727 * config/visium/constraints.md (J, K, L): Likewise.
2728 * config/visium/predicates.md (const_shift_operand): Likewise.
2729 * config/visium/visium.c (visium_legitimize_address,
2730 visium_legitimize_reload_address): Likewise.
2731 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2732 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2733 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2734 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2735 SET_DUMP_DETAIL): Likewise.
2736 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2737 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2738 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2739 avr_set_core_architecture, avr_set_current_function,
2740 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2741 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2742 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2743 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2744 avr_map_decompose, avr_fold_builtin): Likewise.
2745 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2746 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2747 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2748 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2749 * config/m32c/m32c.c (m32c_conditional_register_usage,
2750 m32c_address_cost): Likewise.
2751 * config/m32c/predicates.md (shiftcount_operand,
2752 longshiftcount_operand): Likewise.
2753 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2754 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2755 can_use_cdx_ldstw): Likewise.
2756 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2757 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2759 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2760 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2761 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2762 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2763 * config/vax/vax.c (adjacent_operands_p): Likewise.
2764 * config/ft32/constraints.md (L, b, KA): Likewise.
2765 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2767 * cfgexpand.c (expand_stack_alignment): Likewise.
2768 * gcse.c (insert_expr_in_table): Likewise.
2769 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2770 * cgraphunit.c (cgraph_node::expand): Likewise.
2771 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2772 * emit-rtl.c (add_insn): Likewise.
2773 * input.c (dump_location_info): Likewise.
2774 * passes.c (NEXT_PASS): Likewise.
2775 * read-rtl-function.c (parse_note_insn_name,
2776 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2778 * sched-rgn.c (sched_rgn_init): Likewise.
2779 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2780 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2781 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2783 * explow.c (eliminate_constant_term): Likewise.
2784 * final.c (leaf_renumber_regs_insn): Likewise.
2785 * cfgrtl.c (print_rtl_with_bb): Likewise.
2786 * genhooks.c (emit_init_macros): Likewise.
2787 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2788 * tree-data-ref.c (conflict_fn): Likewise.
2789 * selftest.c (assert_streq): Likewise.
2790 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2791 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2792 fold_binary_loc, multiple_of_p): Likewise.
2793 * reload.c (push_reload, find_equiv_reg): Likewise.
2794 * et-forest.c (et_nca, et_below): Likewise.
2795 * dbxout.c (dbxout_symbol_location): Likewise.
2796 * reorg.c (relax_delay_slots): Likewise.
2797 * dojump.c (do_compare_rtx_and_jump): Likewise.
2798 * gengtype-parse.c (type): Likewise.
2799 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2800 simplify_const_relational_operation): Likewise.
2801 * reload1.c (do_output_reload): Likewise.
2802 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2803 * gengtype.c (type_for_name): Likewise.
2804 * gimple-ssa-sprintf.c (format_directive): Likewise.
2806 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2809 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2810 accessing it. Adjust comment.
2812 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2815 * predict.c (zero_one_minusone): New function.
2816 (apply_return_prediction): Avoid return prediction for functions
2817 returning only -1, 0 and 1 values, unless they only return -1 and 0
2820 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2822 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2823 fall-through warning.
2824 (arc_compute_frame_size): Remove unused variables.
2825 (arc_print_operand): Fix fprintif format.
2826 (arc_can_follow_jump): Clean fall-through warning.
2828 2017-12-19 Marek Polacek <polacek@redhat.com>
2831 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2834 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2836 * config/arc/arc.c (overriderregs): New variable.
2837 (arc_override_options): Track fixed/call saved/call options.
2838 (arc_conditional_register_usage): Check against overrideregs
2839 variable whenever we change register properties.
2841 2017-12-19 Nathan Sidwell <nathan@acm.org>
2843 * opts.c (finish_options): Don't prefix dump_base_name if it
2844 already contains directories.
2846 2017-12-19 Martin Liska <mliska@suse.cz>
2848 PR rtl-optimization/82675
2849 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2850 more element in sbitmap.
2852 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2854 * gimplify.c (gimplify_expr): Use error_operand_p.
2856 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2859 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2862 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2864 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2865 the jump is to a label.
2867 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2869 PR tree-optimization/83444
2870 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2875 * ipa-inline.c (flatten_remove_node_hook): New function.
2876 (ipa_inline): Keep only nodes with flatten attribute at the end of
2877 the array in the order from ipa_reverse_postorder, only walk that
2878 portion of array for flattening, if there is more than one such
2879 node, temporarily register a removal hook and ignore removed nodes.
2881 PR tree-optimization/80631
2882 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2883 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2884 IFN_REDUC_MAX or IFN_REDUC_MIN.
2886 2017-12-18 Jeff Law <law@redhat.com>
2888 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2889 record symbolic equivalences from backedges in the CFG.
2892 2017-11-19 Jeff Law <law@redhat.com>
2894 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2895 of degenerates resulting from ignoring an edge.
2897 2017-12-18 Martin Sebor <msebor@redhat.com>
2900 PR tree-optimization/78450
2901 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2902 (handle_builtin_strlen): Call it.
2904 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2906 PR rtl-optimization/83424
2907 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2909 2017-12-18 Marek Polacek <polacek@redhat.com>
2912 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2913 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2916 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2918 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2919 (umaddsidi4, umaddsidi_split): Likewise.
2921 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2923 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2926 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2929 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2931 2017-12-18 Richard Biener <rguenther@suse.de>
2933 PR tree-optimization/81877
2934 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2935 (outermost_indep_loop): Adjust.
2936 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2937 (can_sm_ref_p): Adjust.
2939 2017-12-18 Richard Biener <rguenther@suse.de>
2942 * tree.c (array_at_struct_end_p): Return true if the underlying
2943 object has space for at least one element in excess of what
2944 the array domain specifies.
2946 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2948 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2949 example to avoid overfull hbox.
2950 * doc/invoke.texi (Option Summary): Add missing @gol.
2951 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2954 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2955 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2957 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2958 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2959 (AArch64 Options): Document -mverbose-cost-dump.
2960 (ARM Options): Likewise, plus -mflip-thumb.
2962 2017-12-17 Martin Sebor <msebor@redhat.com>
2965 * gimple-ssa-warn-restrict.c
2966 (builtin_memref::offset_out_of_bounds): Correct the handling of
2969 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2971 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2973 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2975 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2977 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2979 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2982 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2984 * doc/invoke.texi: Fix some typos.
2986 2017-12-16 Martin Sebor <msebor@redhat.com>
2988 PR tree-optimization/78918
2989 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2990 * builtins.c (check_sizes): Rename...
2991 (check_access): ...to this. Rename function arguments for clarity.
2992 (check_memop_sizes): Adjust names.
2993 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2994 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2995 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2996 (check_strncat_sizes, expand_builtin_strncat): Same.
2997 (expand_builtin_strncpy, expand_builtin_memset): Same.
2998 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2999 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
3000 (maybe_emit_sprintf_chk_warning): Same.
3001 (expand_builtin_strcpy): Adjust.
3002 (expand_builtin_stpcpy): Same.
3003 (expand_builtin_with_bounds): Detect out-of-bounds accesses
3004 in pointer-checking forms of memcpy, memmove, and mempcpy.
3005 (gcall_to_tree_minimal, max_object_size): Define new functions.
3006 * builtins.h (max_object_size): Declare.
3007 * calls.c (alloc_max_size): Call max_object_size instead of
3008 hardcoding ssizetype limit.
3009 (get_size_range): Handle new argument.
3010 * calls.h (get_size_range): Add a new argument.
3011 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3012 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3013 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3015 (gimple_fold_builtin_memory_chk): Same.
3016 (gimple_fold_builtin_stxcpy_chk): New function.
3017 * gimple-ssa-warn-restrict.c: New source.
3018 * gimple-ssa-warn-restrict.h: New header.
3019 * gimple.c (gimple_build_call_from_tree): Propagate location.
3020 * passes.def (pass_warn_restrict): Add new pass.
3021 * tree-pass.h (make_pass_warn_restrict): Declare.
3022 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3024 (handle_builtin_strcat): Same.
3025 (strlen_optimize_stmt): Rename...
3026 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3027 stpncpy, strncpy, and their checking forms.
3029 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3031 PR rtl-optimization/82849
3032 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3033 and get_max_loop_iterations_int.
3035 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3036 Alan Hayward <alan.hayward@arm.com>
3037 David Sherwood <david.sherwood@arm.com>
3039 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3040 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3041 size as a poly_uint64.
3042 (mode_for_vector, mode_for_int_vector): Take the number of vector
3043 elements as a poly_uint64.
3044 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3045 size as a poly_uint64.
3046 (mode_for_vector, mode_for_int_vector): Take the number of vector
3047 elements as a poly_uint64.
3049 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3050 Alan Hayward <alan.hayward@arm.com>
3051 David Sherwood <david.sherwood@arm.com>
3053 * machmode.h (MACRO_MODE): New macro.
3054 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3055 * targhooks.c (default_libcall_value, default_secondary_reload)
3056 (default_memory_move_cost, default_register_move_cost)
3057 (default_class_max_nregs): Likewise.
3059 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3060 Alan Hayward <alan.hayward@arm.com>
3061 David Sherwood <david.sherwood@arm.com>
3063 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3064 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3066 * genautomata.c (main): Likewise.
3067 * genconditions.c (write_header): Likewise.
3068 * genemit.c (main): Likewise.
3069 * genextract.c (print_header): Likewise.
3070 * genopinit.c (main): Likewise.
3071 * genoutput.c (output_prologue): Likewise.
3072 * genpeep.c (main): Likewise.
3073 * genpreds.c (write_insn_preds_c): Likewise.
3074 * genrecog.c (writer_header): Likewise.
3075 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3076 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3077 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3078 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3079 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3080 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3081 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3082 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3083 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3084 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3085 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3086 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3087 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3088 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3089 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3090 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3091 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3092 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3093 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3094 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3095 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3096 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3097 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3098 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3099 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3100 * config/darwin.c (IN_TARGET_CODE): Likewise.
3101 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3102 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3103 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3104 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3105 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3106 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3107 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3108 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3109 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3110 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3111 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3112 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3113 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3114 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3115 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3116 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3117 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3118 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3119 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3120 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3121 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3122 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3123 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3124 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3125 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3126 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3127 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3128 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3129 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3130 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3131 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3132 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3133 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3134 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3135 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3136 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3137 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3138 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3139 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3140 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3141 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3142 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3143 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3144 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3145 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3146 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3147 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3148 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3149 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3150 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3151 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3152 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3153 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3154 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3155 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3156 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3157 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3158 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3159 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3160 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3161 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3162 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3163 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3164 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3165 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3166 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3167 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3168 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3169 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3170 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3171 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3172 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3173 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3174 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3175 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3176 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3177 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3178 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3179 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3180 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3181 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3182 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3183 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3184 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3185 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3186 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3187 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3188 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3189 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3190 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3191 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3192 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3193 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3194 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3195 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3196 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3197 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3198 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3199 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3200 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3201 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3202 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3203 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3204 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3205 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3206 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3208 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3209 Alan Hayward <alan.hayward@arm.com>
3210 David Sherwood <david.sherwood@arm.com>
3212 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3215 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3216 Alan Hayward <alan.hayward@arm.com>
3217 David Sherwood <david.sherwood@arm.com>
3219 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3220 * doc/md.texi (vec_series@var{m}): Document.
3221 * tree.def (VEC_SERIES_EXPR): New tree code.
3222 * tree.h (build_vec_series): Declare.
3223 * tree.c (build_vec_series): New function.
3224 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3225 * tree-pretty-print.c (dump_generic_node): Likewise.
3226 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3227 * tree-inline.c (estimate_operator_cost): Likewise.
3228 * expr.c (expand_expr_real_2): Likewise.
3229 * optabs-tree.c (optab_for_tree_code): Likewise.
3230 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3231 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3232 * expmed.c (make_tree): Handle VEC_SERIES.
3233 * optabs.def (vec_series_optab): New optab.
3234 * optabs.h (expand_vec_series_expr): Declare.
3235 * optabs.c (expand_vec_series_expr): New function.
3236 * tree-vect-generic.c (expand_vector_operations_1): Check that
3237 the operands also have vector type.
3239 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3240 Alan Hayward <alan.hawyard@arm.com>
3241 David Sherwood <david.sherwood@arm.com>
3243 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3244 (VEC_COND_EXPR): Add missing @tindex.
3245 * doc/md.texi (vec_duplicate@var{m}): Document.
3246 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3247 * tree.c (build_vector_from_val): Add stubbed-out handling of
3248 variable-length vectors, using VEC_DUPLICATE_EXPR.
3249 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3250 * cfgexpand.c (expand_debug_expr): Likewise.
3251 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3252 * tree-inline.c (estimate_operator_cost): Likewise.
3253 * tree-pretty-print.c (dump_generic_node): Likewise.
3254 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3255 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3256 (test_vec_duplicate_folding): New function.
3257 (fold_const_c_tests): Call it.
3258 * optabs.def (vec_duplicate_optab): New optab.
3259 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3260 * optabs.h (expand_vector_broadcast): Declare.
3261 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3262 vec_duplicate_optab.
3263 * expr.c (store_constructor): Try using vec_duplicate_optab for
3265 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3267 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3270 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3271 div/mod latencies a bit.
3273 2017-12-15 Jeff Law <law@redhat.com>
3275 PR tree-optimization/36550
3276 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3277 (mark_threaded_blocks): Rewrite code to avoid block copying when
3278 optimizing for size. Don't pessimize blocks which will be
3279 copied, but all the statements will be dead.
3281 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3283 PR tree-optimization/81165
3284 * tree-ssa-threadupdate.c (uses_in_bb): New.
3285 (estimate_threading_killed_stmts): New.
3286 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3287 * tree-ssa-threadedge.c
3288 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3291 2017-12-15 Jeff Law <law@redhat.com>
3293 PR tree-optimization/83410
3294 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3295 threads when parallelizing loops.
3297 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3299 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3301 * config/alpha/alpha.c (vms_attribute_table): Swap
3302 affects_type_identity and handler fields, adjust comments.
3303 * config/mips/mips.c (mips_attribute_table): Likewise.
3304 * config/visium/visium.c (visium_attribute_table): Likewise.
3305 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3306 * config/microblaze/microblaze.c (microblaze_attribute_table):
3308 * config/spu/spu.c (spu_attribute_table): Likewise.
3309 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3310 * config/arc/arc.c (arc_attribute_table): Likewise.
3311 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3312 * config/v850/v850.c (v850_handle_interrupt_attribute,
3313 v850_handle_data_area_attribute): Formatting fixes.
3314 (v850_attribute_table): Swap affects_type_identity and handler
3315 fields, adjust comments.
3316 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3317 * config/arm/arm.c (arm_attribute_table): Likewise.
3318 * config/avr/avr.c (avr_attribute_table): Likewise.
3319 * config/s390/s390.c (s390_attribute_table): Likewise.
3320 * config/sh/sh.c (sh_attribute_table): Likewise.
3321 * config/i386/i386.c (ix86_handle_cconv_attribute,
3322 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3323 (ix86_attribute_table): Swap affects_type_identity and handler
3324 fields, adjust comments.
3325 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3326 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3327 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3328 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3329 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3330 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3331 * config/rx/rx.c (rx_attribute_table): Likewise.
3332 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3333 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3334 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3335 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3336 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3337 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3338 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3339 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3340 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3341 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3342 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3344 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3346 (register_scoped_attributes, decl_attributes): Formatting fixes.
3348 PR tree-optimization/83269
3349 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3350 subtraction in arg0's type if type is signed and arg0 is unsigned.
3354 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3355 convert? on @0 instead of convert. Check type of @1, not @0.
3356 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3357 convert? on @0 instead of convert. Check type of @1, not @0.
3358 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3359 only optimize if either both @1 and @2 types are narrower
3360 precision, or both are wider or equal precision, and in the former
3361 case only if both have undefined overflow.
3363 2017-12-15 Richard Biener <rguenther@suse.de>
3366 * internal-fn.def (IFN_NOP): Add.
3367 * internal-fn.c (expand_NOP): Do nothing.
3368 * lto-streamer-in.c (input_function): Instead of removing
3369 sanitizer calls replace them with IFN_NOP calls.
3371 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3372 Alan Hayward <alan.hayward@arm.com>
3373 David Sherwood <david.sherwood@arm.com>
3375 * dse.c (store_info, read_info_type): Replace begin and end with
3377 (print_range): New function.
3378 (set_all_positions_unneeded, any_positions_needed_p)
3379 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3381 (record_store): Likewise. Optimize the case in which all positions
3383 (get_stored_val): Replace read_begin and read_end with read_offset
3385 (replace_read): Update call accordingly.
3387 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3389 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3390 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3391 (loop_cand::loop_cand): Initialize above members.
3392 (loop_cand::supported_operations): Delete.
3393 (loop_cand::can_interchange_p): Inline above function.
3394 (loop_cand::classify_simple_reduction): Record number of constant
3395 initialized simple reductions.
3396 (should_interchange_loops): New parameters. Check stmt cost of loops
3398 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3399 Update call to should_interchange_loops.
3400 (should_interchange_loop_nest): Update call to
3401 should_interchange_loops.
3403 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3406 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3407 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3408 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3410 2017-12-15 Julia Koval <julia.koval@intel.com>
3412 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3413 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3415 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3416 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3417 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3419 2017-12-15 Julia Koval <julia.koval@intel.com>
3421 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3422 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3423 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3424 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3425 _mm_aesenc_epi128): New intrinsics.
3427 2017-12-15 Julia Koval <julia.koval@intel.com>
3429 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3430 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3432 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3433 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3434 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3436 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3438 * gimple-ssa-strength-reduction.c (analyze_increments):
3439 Distinguish replacement costs for constant strides from those for
3442 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3444 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3446 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3448 * doc/invoke.texi: Document -Wcast-function-type.
3449 * recog.h (stored_funcptr): Change signature.
3450 * tree-dump.c (dump_node): Avoid warning.
3451 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3453 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3456 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3459 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3461 PR tree-optimization/83312
3462 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3463 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3464 "val" param, and to cope with arbitrary basic blocks.
3465 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3466 handle NULL_TREE for "val", dropping "bb" param.
3467 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3468 drop "bb" param. Handle NULL_TREE for "val".
3469 (find_case_label_for_value): Make "switch_stmt" param const.
3470 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3472 (vrp_prop::check_all_array_refs): Reimplement as...
3473 (check_array_bounds_dom_walker::before_dom_children): ...this new
3474 vfunc. Replace linear search through BB block list, excluding
3475 those with non-executable in-edges via dominator walk.
3477 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3479 * config/arm/arm.opt (mverbose-cost-dump): New option.
3480 * config/arm/arm.c (arm_rtx_costs): Use it.
3482 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3485 * reload1.c (emit_input_reload_insns): Skip debug markers.
3487 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3489 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3493 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3494 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3495 * config/sh/sh.c: Skip debug insns besides notes.
3496 * config/sh/sh.md: Likewise.
3497 * config/sh/sh_treg_combine.cc: Likewise.
3498 * config/sh/sync.md: Likewise.
3500 2017-12-14 Tom de Vries <tom@codesourcery.com>
3502 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3503 Add item for weak_undefined.
3505 2017-12-14 Richard Biener <rguenther@suse.de>
3507 PR tree-optimization/67842
3508 * sese.h (bb_in_region): Remove #if 0'ed code.
3510 2017-12-14 Richard Biener <rguenther@suse.de>
3512 PR tree-optimization/83326
3513 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3514 may_be_zero parameter and handle it by not marking the first
3515 peeled copy as not exiting the loop.
3516 (try_peel_loop): Likewise.
3517 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3518 to handle the case of constant or zero iterations and perform
3519 loop header copying on-the-fly.
3521 2017-12-14 Richard Biener <rguenther@suse.de>
3523 PR tree-optimization/83418
3524 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3525 Instead of asserting we don't get unfolded comparisons deal with them.
3527 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3530 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3531 basic blocks. Assert debug bind insns don't appear outside of bbs,
3532 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3535 PR tree-optimization/83198
3536 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3537 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3538 value if arg is a REAL_CST with incompatible type.
3540 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3541 Bin Cheng <bin.cheng@arm.com>
3544 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3546 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3549 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3551 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3552 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3554 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3555 Alan Hayward <alan.hayward@arm.com>
3556 David Sherwood <david.sherwood@arm.com>
3558 * poly-int.h: New file.
3559 * poly-int-types.h: Likewise.
3560 * coretypes.h: Include them.
3561 (POLY_INT_CONVERSION): Define.
3562 * target.def (estimated_poly_value): New hook.
3563 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3564 * doc/tm.texi: Regenerate.
3565 * doc/poly-int.texi: New file.
3566 * doc/gccint.texi: Include it.
3567 * doc/rtl.texi: Describe restrictions on subreg modes.
3568 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3569 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3570 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3571 * targhooks.h (default_estimated_poly_value): Declare.
3572 * targhooks.c (default_estimated_poly_value): New function.
3573 * target.h (estimated_poly_value): Likewise.
3574 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3575 (wi::unary_traits): Delete.
3576 (wi::binary_traits::signed_shift_result_type): Define for
3577 offset_int << HOST_WIDE_INT, etc.
3578 (generic_wide_int::operator <<=): Define for all types and use
3579 wi::lshift instead of <<.
3580 (wi::hwi_with_prec): Add a default constructor.
3581 (wi::ints_for): New class.
3582 (operator <<): Define for all wide-int types.
3583 (operator /): New function.
3584 (operator %): Likewise.
3585 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3586 (ASSERT_MAYBE_NE_AT): New macros.
3588 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3589 Dominik Vogt <vogt@linux.vnet.ibm.com>
3592 * emit-rtl.c (init_emit): Remove ??? comment.
3593 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3594 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3595 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3596 alignment of 3 virtual registers to BITS_PER_WORD.
3598 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3600 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3602 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3603 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3605 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3608 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3610 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3612 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3613 with the exception of debug begin stmt markers appear before
3617 * final.c (rest_of_handle_final): Call variable_tracking_main only
3618 if !flag_var_tracking.
3620 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3621 Jakub Jelinek <jakub@redhat.com>
3625 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3626 labels debug stmts that can only appear after labels.
3628 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3630 PR rtl-optimization/82398
3631 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3632 EXPR_USEFULNESS in priority comparison.
3634 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3636 PR rtl-optimization/83393
3637 * combine.c (move_deaths): If reg_stat points to a too new insn in
3638 last_death, do not use it: find the proper insn instead.
3640 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3643 * config/rs6000/rs6000-string.c (expand_block_compare,
3644 expand_strn_compare): Fix set_mem_size() calls.
3646 2017-12-12 Jeff Law <law@redhat.com>
3648 PR tree-optimization/83298
3649 PR tree-optimization/83362
3650 PR tree-optimization/83383
3651 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3652 push_value_range a public interface. Add new argument to
3653 record_ranges_from_stmt.
3654 * gimple-ssa-evrp-analyze.c
3655 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3656 Update comments. Handle recording temporary equivalences.
3657 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3658 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3659 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3660 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3661 gimple-ssa-evrp-analyze.h.
3662 (record_temporary_equivalences_from_phis): Add new argument. When
3663 the PHI arg is an SSA_NAME, set the result's range to the range
3665 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3666 from statements too.
3667 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3668 Pass it down to children as needed.
3669 (thread_outgoing_edges): Likewise.
3670 (thread_across_edge): Likewise. Push/pop range state as needed.
3671 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3673 2017-12-12 Julia Koval <julia.koval@intel.com>
3675 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3676 (PTA_CANNONLAKE): Remove PTA_CLWB.
3678 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3681 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3682 (vcondv2div2df): Likewise.
3683 (vconduv2dfv2di): Likewise.
3685 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3687 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3688 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3689 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3690 ATTR_MATHFN_FPROUNDING_ERRNO.
3692 2017-12-12 Richard Biener <rguenther@suse.de>
3694 PR tree-optimization/83385
3695 * graphite-scop-detection.c (get_order, order): Remove.
3696 (bb_to_rpo): New global.
3698 (build_scops): Sort pbbs in RPO order.
3700 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3702 * combine.c (simplify_set): Do not transform subregs to zero_extends
3703 if the destination mode is a vector mode.
3705 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3707 PR tree-optimization/83359
3708 * tree-cfg.h (fold_loop_internal_call): Declare.
3709 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3710 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3711 (find_loop_dist_alias): New function.
3712 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3713 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3716 PR tree-optimization/80631
3717 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3718 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3719 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3720 hardcoding zero as the value if COND_EXPR is never true. For
3721 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3722 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3723 hardcoding MAX_EXPR as the reduction operation.
3724 (is_nonwrapping_integer_induction): Allow negative step.
3725 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3726 vect_create_epilog_for_reduction, if no value is suitable, don't
3727 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3729 2017-12-12 Richard Biener <rguenther@suse.de>
3731 PR tree-optimization/81889
3732 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3733 range info from the non-wrapping IV instead of just the range
3736 2017-12-12 Julia Koval <julia.koval@intel.com>
3738 * config.gcc: Add vaesintrin.h.
3739 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3741 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3742 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3744 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3745 * config/i386/immintrin.h: Include vaesintrin.h.
3746 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3747 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3748 _mm_aesdec_epi128): New intrinsics.
3750 2017-12-12 Julia Koval <julia.koval@intel.com>
3752 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3753 OPTION_MASK_ISA_VAES_UNSET): New.
3754 (ix86_handle_option): Handle -mvaes.
3755 * config/i386/cpuid.h: Define bit_VAES.
3756 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3757 * config/i386/i386-c.c (__VAES__): New.
3758 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3759 (ix86_valid_target_attribute_inner_p): Ditto.
3760 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3761 * config/i386/i386.opt: Add -mvaes.
3762 * doc/invoke.texi: Ditto.
3764 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3766 * debug.h (gcc_debug_hooks): Add inline_entry.
3767 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3768 * debug.c (do_nothing_debug_hooks): Likewise.
3769 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3770 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3771 (dwarf2_lineno_debug_hooks): Likewise.
3773 * common.opt (gstatement-frontiers): New, setting
3774 debug_nonbind_markers_p.
3775 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3776 * toplev.c (process_options): Autodetect value for debug statement
3778 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3779 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3781 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3782 markers. Integrate source bind into debug stmt expand loop.
3783 (pass_expand::execute): Check debug marker limit. Avoid deep
3784 TER and expand debug locations for debug bind insns only.
3785 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3787 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3788 * final.c (reemit_insn_block_notes): Take current block from
3789 nonbind markers. Declare note where it's first set.
3790 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3791 begin stmt markers if enabled.
3792 (notice_source_line): Handle nonbind markers. Fail if their
3793 location is unknown or that of builtins.
3794 (rest_of_handle_final): Convert begin stmt markers to notes if
3795 var-tracking didn't run.
3796 (rest_of_clean_state): Skip begin stmt markers.
3797 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3799 * function.c (allocate_struct_function): Set begin_stmt_markers.
3800 * function.h (struct function): Add debug_marker_count counter
3801 and debug_nonbind_markers flag.
3802 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3803 * gimple-low.c (lower_function_body): Adjust
3804 debug_nonbind_markers.
3805 (lower_stmt): Drop or skip gimple debug stmts.
3806 (lower_try_catch): Skip debug stmts.
3807 * gimple.c (gimple_build_debug_begin_stmt): New.
3808 (gimple_copy): Increment debug_marker_count if copying one.
3809 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3810 * gimplify.c (rexpr_location): New.
3811 (rexpr_has_location): New.
3812 (warn_switch_unreachable_r): Handle gimple debug stmts.
3813 (shortcut_cond_r): Call expr_location.
3815 (find_goto_label): New.
3816 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3818 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3820 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3821 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3822 (LANG_HOOKS_INITIALIZER): ... this.
3823 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3824 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3826 (update_ebb_live_info): Skip debug insn markers.
3827 * lra.c (debug_insn_static_data): Rename to...
3828 (debug_bind_static_data): ... this.
3829 (debug_marker_static_data): New.
3830 (lra_set_insn_recog_data): Select one of the above depending
3832 (lra_update_isn_regno_info): Don't assume debug insns have
3834 (push_insns): Skip debug insns.
3835 * lto-streamer-in.c (input_function): Drop debug stmts
3836 depending on active options. Adjust debug_nonbind_markers.
3837 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3838 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3839 begin stmt marker notes.
3840 (print_insn): Likewise.
3841 * recog.c (extract_insn): Recognize rtl for debug markers.
3842 * rtl.def (DEBUG_MARKER): New.
3843 * tree-inline.c: Include params.h.
3844 (remap_gimple_stmt): Handle nonbind markers.
3845 (maybe_move_debug_stmts_to_successors): Likewise.
3846 (copy_debug_stmt): Likewise.
3847 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3848 markers regardless of no side effects.
3849 (tsi_link_before): Don't update container's side effects when adding
3850 a begin stmt marker.
3851 (tsi_link_after): Likewise.
3852 (expr_first): Skip begin stmt markers.
3853 (expr_last): Likewise.
3854 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3855 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3856 Disregard nonbind markers.
3857 * tree.c (make_node_stat): Don't set side effects for begin stmt
3859 (build1_stat): Likewise.
3860 * tree.def (DEBUG_BEGIN_STMT): New.
3861 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3862 * var-tracking.c (delete_debug_insns): Renamed to...
3863 (delete_vta_debug_insns): ... this.
3864 (reemit_marker_as_note): New.
3865 (vt_initialize): Reemit markers.
3866 (delete_vta_debug_insns): Likewise.
3867 (vt_debug_insns_local): Reemit or delete markers.
3868 (variable_tracking_main_1): Likewise.
3869 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3870 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3871 (gimple_debug_nonbind_marker_p): New.
3872 (gimple_build_debug_bind): Adjust.
3873 (gimple_build_debug_begin_stmt): New.
3874 * doc/invoke.texi (max-debug-marker-count): New param.
3875 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3876 (debug_parameter_ref, debug_marker): New.
3877 (NOTE_INSN_BEGIN_STMT): New.
3878 (DEBUG_INSN): Describe begin stmt markers.
3880 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3881 without debug insns, we wouldn't, but clean up debug insns
3882 after a control flow insn nevertheless.
3884 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3885 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3886 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3887 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3888 (rtl_tidy_fallthru_edge): Likewise.
3889 (rtl_verify_fallthru): Likewise.
3890 (rtl_verify_bb_layout): Likewise.
3891 (skip_insns_after_block): Likewise.
3892 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3893 * dwarf2out.c: Include print-rtl.h.
3894 (dwarf2out_next_real_insn): New.
3895 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3896 Dump debug binds in asm comments.
3897 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3898 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3899 callers to use gsi_start_nondebug_bb instead.
3900 (gsi_after_labels): Skip gimple debug stmts.
3901 (gsi_start_nondebug): New.
3902 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3903 (proper_loop_form_for_interchange): Adjust.
3904 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3905 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3906 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3907 (collect_fallthrough_labels): Likewise.
3908 (should_warn_for_implicit_fallthrough): Likewise.
3909 (warn_implicit_fallthrough_r): Likewise.
3910 (expand_FALLTHROUGH_r): Likewise.
3911 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3912 (graphite_copy_stmts_from_block): Skip nonbind markers.
3913 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3914 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3915 * jump.c (clean_barriers): Skip debug insns.
3916 * omp-expand.c (expand_parallel_call): Skip debug insns.
3917 (expand_task_call): Likewise.
3918 (remove_exit_barrier): Likewise.
3919 (expand_omp_taskreg): Likewise.
3920 (expand_omp_for_init_counts): Likewise.
3921 (expand_omp_for_generic): Likewise.
3922 (expand_omp_for_static_nochunk): Likewise.
3923 (expand_omp_for_static_chunk): Likewise.
3924 (expand_omp_simd): Likewise.
3925 (expand_omp_taskloop_for_outer): Likewise.
3926 (expand_omp_taskloop_for_inner): Likewise.
3927 (expand_oacc_for): Likewise.
3928 (expand_omp_sections): Likewise.
3929 (expand_omp_single): Likewise.
3930 (expand_omp_synch): Likewise.
3931 (expand_omp_atomic_load): Likewise.
3932 (expand_omp_atomic_store): Likewise.
3933 (expand_omp_atomic_fetch_op): Likewise.
3934 (expand_omp_atomic_pipeline): Likewise.
3935 (expand_omp_atomic_mutex): Likewise.
3936 (expand_omp_target): Likewise.
3937 (grid_expand_omp_for_loop): Likewise.
3938 (grid_expand_target_grid_body): Likewise.
3939 (build_omp_regions_1): Likewise.
3940 * omp-low.c (check_combined_parallel): Skip debug stmts.
3941 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3942 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3943 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3945 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3946 (make_edges): Likewise.
3947 (cleanup_dead_labels): Likewise.
3948 (gimple_can_merge_blocks_p): Likewise.
3949 (stmt_starts_bb_p): Likewise.
3950 (gimple_block_label): Likewise.
3951 (gimple_redirect_edge_and_branch): Likewise.
3952 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3954 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3956 * tree-pretty-print (print_declaration): Omit initializer in slim
3958 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3960 (eliminate_unnecessary_stmts): Stabilize block removal order.
3961 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3962 * var-tracking.c (get_first_insn): New.
3963 (vt_emit_notes): Call it.
3964 (vt_initialize): Walk any insns before the first BB.
3965 (delete_debug_insns): Likewise.
3967 * gimple.h (enum gimple_debug_subcode): Add
3968 GIMPLE_DEBUG_BEGIN_STMT.
3969 (gimple_debug_begin_stmt_p): New.
3970 (gimple_debug_nonbind_marker_p): New.
3971 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3972 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3973 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3974 * insn-notes.def (BEGIN_STMT): New.
3975 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3976 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3977 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3978 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3979 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3980 (INSN_DEBUG_MARKER_KIND): New.
3981 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3982 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3983 (INSN_VAR_LOCATION_PTR): New.
3984 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3986 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3987 deps only if debug bind insns are enabled.
3988 (pass_expand::execute): Avoid deep TER and expand
3989 debug locations for debug bind insns only.
3990 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3991 debug stmts special handling down to debug bind stmts.
3992 * combine.c (try_combine): Narrow debug insns special handling
3993 down to debug bind insns.
3994 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3995 Narrow debug insns preexisting special handling down to debug
3997 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3998 handling down to debug bind insns.
3999 * function.c (instantiate_virtual_regs): Skip debug markers,
4000 adjust handling of debug binds.
4001 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
4002 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
4003 * haifa-sched.c (schedule_insn): Narrow special handling of debug
4004 insns to debug bind insns.
4005 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
4006 special handling of debug stmts to debug bind stmts.
4007 * ipa-split.c (split_function): Likewise.
4008 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4009 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4011 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4012 * regrename.c (build_def_use): Likewise.
4013 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4014 (pass_cprop_hardreg): Narrow special casing of debug insns to
4016 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4017 * reload1.c (reload): Likewise.
4018 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4019 casing of debug stmts to debug bind stmts.
4020 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4021 * ssa-iterators.h (num_imm_uses): Likewise.
4022 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4023 debug stmts to debug bind stmts.
4024 * tree-inline.c (tree_function_versioning): Narrow special casing
4025 of debug stmts to debug bind stmts.
4026 * tree-loop-distribution.c (generate_loops_for_partition):
4027 Narrow special casing of debug stmts to debug bind stmts.
4028 * tree-sra.c (analyze_access_subtree): Narrow special casing
4029 of debug stmts to debug bind stmts.
4030 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4031 stmts to debug bind stmts.
4032 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4033 casing of debug stmts to debug bind stmts.
4034 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4035 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4036 casing of debug stmts to debug bind stmts.
4037 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4039 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4040 debug stmts to debug bind stmts.
4041 (gimple_replace_ssa_lhs): Likewise.
4042 (insert_debug_temp_for_var_def): Likewise.
4043 (insert_debug_temps_for_defs): Likewise.
4044 (reset_debug_uses): Likewise.
4045 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4046 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4047 (adjust_debug_stmts): Likewise.
4048 (adjust_phi_and_debug_stmts): Likewise.
4049 (vect_do_peeling): Likewise.
4050 * tree-vect-loop.c (vect_transform_loop): Likewise.
4051 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4052 * var-tracking.c (adjust_mems): Narrow special casing of debug
4053 insns to debug bind insns.
4054 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4055 (compute_bb_dataflow, vt_find_locations): Likewise.
4056 (vt_expand_loc, emit_notes_for_changes): Likewise.
4057 (vt_init_cfa_base): Likewise.
4058 (vt_emit_notes): Likewise.
4059 (vt_initialize): Likewise.
4060 (vt_finalize): Likewise.
4062 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4063 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4064 (next_nonnote_nondebug_insn_bb): New.
4065 (prev_nonnote_nondebug_insn_bb): New.
4066 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4067 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4068 (prev_nonnote_nondebug_insn_bb): Declare.
4069 (next_nonnote_nondebug_insn_bb): Declare.
4070 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4071 * cfgrtl.c (get_last_bb_insn): Likewise.
4072 * lra.c (push_insns): Likewise.
4074 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4077 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4079 * selftest.c (selftest::test_runner::test_runner): New ctor.
4080 (selftest::test_runner::~test_runner): New dtor.
4081 * selftest.h (class selftest::test_runner): New class.
4083 2017-12-11 Carl Love <cel@us.ibm.com>
4085 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4086 vec_extract_fp32_from_shortl]): Add #defines.
4087 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4088 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4089 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4090 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4091 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4092 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4093 * doc/extend.texi: Add documentation for the added builtins.
4095 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4097 PR rtl-optimization/80693
4098 PR rtl-optimization/81019
4099 PR rtl-optimization/81020
4100 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4101 are not mentioned in i3. Place the REG_UNUSED note on i2,
4102 possibly modified to REG_DEAD, if it did not originate in i3.
4104 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4106 * recog.c (store_data_bypass_p_1): New function.
4107 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4108 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4110 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4112 PR rtl-optimization/83361
4113 * ifcvt.c (if_convert): Call fixup_partitions.
4115 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4117 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4118 early folding of splat_u{8,16,32}.
4120 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4122 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4123 output_operand_lossage first argument with capital letter.
4124 (aarch64_override_options): Don't start error and sorry first argument
4125 with capital letter.
4127 2017-12-11 Andi Kleen <ak@linux.intel.com>
4129 PR gcov-profile/83355
4130 * auto-profile.c (string_table::get_index_by_decl): Don't
4131 recurse when abstract origin points to itself.
4133 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4135 PR tree-optimization/83320
4136 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4137 (prune_datarefs_not_in_loop): Ditto.
4139 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4141 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4143 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4145 PR tree-optimization/83337
4146 * gimple-loop-interchange.cc (compute_access_stride): Handle
4147 bitfield DRs properly.
4149 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4151 PR tree-optimization/83338
4152 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4153 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4154 vector of unsigned integers to vector of signed integers.
4156 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4158 PR rtl-optimization/83317
4159 * lra-constraints.c (process_address_1): Add insn code check.
4161 2017-12-08 Michael Matz <matz@suse.de>
4163 Fix PR tree-optimization/83323
4164 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4165 head-controlled loops and loop BBs.
4166 * common.opt (funroll-and-jam): Remove, instead ...
4167 (floop-unroll-and-jam): ... reuse this option.
4168 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4169 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4170 (-floop-unroll-and-jam): ... this option.
4172 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4174 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4175 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4176 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4178 2017-12-08 Richard Biener <rguenther@suse.de>
4181 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4182 handle accesses outside of zero-sized vars.
4184 2017-12-08 Martin Jambor <mjambor@suse.cz>
4186 PR tree-optimization/83141
4187 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4188 test for MEM_REFs implicitely changing types with padding. Remove
4190 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4192 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4194 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4196 * config/avr/avr.c (avr_attribute_table): Likewise.
4197 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4198 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4199 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4200 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4201 of struct attribute_spec.
4202 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4204 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4207 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4209 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4212 * profile-count.c (profile_count::from_gcov_type): Move from
4213 profile-count.h; handle overflow.
4214 * profile-count.h (profile_count::from_gcov_type): Move offline.
4216 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4218 PR rtl-optimization/83304
4219 * combine.c (move_deaths): If we do not know where a register died,
4222 2017-12-08 Richard Biener <rguenther@suse.de>
4224 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4225 Provide -fopt-info-loop feedback when we interchange in a nest.
4227 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4229 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4230 for armv6 ARM CPU IDs.
4232 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4234 * common/config/arm/arm-common.c: Include <algorithm>.
4235 (INCLUDE_VECTOR): Define.
4236 (compare_opt_names): New function.
4237 (arm_rewrite_selected_arch): Only strip out extensions that can be
4238 expressed through -mfpu. Sort the remaining extensions
4241 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4243 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4244 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4245 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4246 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4248 2017-06-08 Tristan Gingold <gindold@adacore.com>
4251 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4252 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4254 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4256 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4257 of struct attribute_spec.
4259 2017-12-08 Julia Koval <julia.koval@intel.com>
4261 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4262 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4263 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4264 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4265 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4266 _mm_maskz_dpwssds_epi32): Ditto.
4268 2017-12-08 Richard Biener <rguenther@suse.de>
4270 PR tree-optimization/81303
4271 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4272 conditions try to create a comparison vector type matching
4273 the data vector type.
4274 (vectorizable_condition): Adjust.
4275 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4276 Leave invariant conditions alone in case we can vectorize those.
4278 2017-12-08 Julia Koval <julia.koval@intel.com>
4280 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4281 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4282 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4283 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4284 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4286 2017-12-08 Julia Koval <julia.koval@intel.com>
4288 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4289 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4290 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4291 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4292 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4293 _mm_maskz_dpbusds_epi32): New intrinsics.
4295 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4297 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4299 (ld<bh>io_signed): Likewise.
4300 (st<bhw>io): Likewise.
4301 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4302 SMALL_INT12 constant integer operand.
4304 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4306 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4308 (nios2_large_constant_allowed): ...this. Adjust uses.
4309 (nios2_plus_symbolic_constant_p): Rename to...
4310 (nios2_plus_large_constant_p): ...this. Adjust uses.
4311 (nios2_legitimate_address_p): Correct CONST_INT handling.
4312 (nios2_symbolic_memory_operand_p): Rename to...
4313 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4314 (nios2_large_constant_p): Check for large constant integers too.
4315 (nios2_split_large_constant): Handle constant integers.
4316 (nios2_split_symbolic_memory_operand): Rename to...
4317 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4318 (nios2_legitimize_constant_address): Handle constant integers.
4319 (r0rel_constant_p): Handle small constant integers.
4320 (nios2_print_operand_address): Handle r0-relative integer addresses.
4321 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4322 * config/nios2/nios2.md: Adjust for renamed functions.
4324 2017-12-07 Andrew Waterman <andrew@sifive.com>
4326 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4327 (TARGET_HAVE_SRODATA_SECTION): New define.
4328 (riscv_select_section): New function.
4330 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4331 Alexander Monakov <amonakov@ispras.ru>
4332 Jakub Jelinek <jakub@redhat.com>
4335 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4337 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4339 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4340 a bool success value. Don't call output_operand_lossage here.
4341 (aarch64_print_ldpstp_address): Return a bool success value.
4342 (aarch64_print_operand_address): Call output_addr_const if
4343 aarch64_print_address_internal fails.
4344 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4345 'y'; call output_operand_lossage instead. Call output_operand_lossage
4346 if aarch64_print_ldpstp_address fails.
4348 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4350 * tree-vector-builder.h
4351 (tree_vector_builder::binary_encoded_nelts): Declare.
4352 * tree-vector-builder.c
4353 (tree_vector_builder::binary_encoded_nelts): New function.
4354 * fold-const.c (negate_expr_p): Likewise.
4355 (operand_equal_p, fold_checksum_tree): Likewise.
4356 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4357 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4358 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4359 (uniform_vector_p): Likewise.
4360 * varasm.c (const_hash_1, compare_constant): Likewise.
4361 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4362 (valid_lattice_transition): Operate directly on the VECTOR_CST
4364 * ipa-icf.c: Include tree-vector-builder.h.
4365 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4366 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4368 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4370 * tree.c (build_vector): Delete.
4371 * tree.h (build_vector): Make static and move into the self-testing
4374 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4376 * vector-builder.h (vector_builder::derived): New const overload.
4377 (vector_builder::elt): New function.
4378 * tree-vector-builder.h (tree_vector_builder::type): New function.
4379 (tree_vector_builder::apply_step): Declare.
4380 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4382 * gimple-fold.h (tree_vector_builder): Declare.
4383 (gimple_build_vector): Take a tree_vector_builder instead of a
4384 type and vector of elements.
4385 * gimple-fold.c (gimple_build_vector): Likewise.
4386 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4388 (get_initial_defs_for_reduction): Likewise.
4389 (vectorizable_induction): Likewise.
4391 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4393 * tree-vector-builder.h
4394 (tree_vector_builder::new_binary_operation): Declare.
4395 * tree-vector-builder.c
4396 (tree_vector_builder::new_binary_operation): New function.
4397 * fold-const.c (fold_relational_const): Use it.
4398 (const_binop): Likewise. Check that both input vectors have
4399 the same number of elements, thus excluding things like WIDEN_SUM.
4400 Check whether it is possible to operate directly on the encodings
4403 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4405 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4406 new_unary_operation, operating only on the encoded elements.
4407 (const_unop): Likewise.
4408 (exact_inverse): Likewise.
4409 (distributes_over_addition_p): New function.
4410 (const_binop): Use tree_vector_builder and new_unary_operation
4411 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4412 on the encoded elements unless the encoding is strided and the
4413 operation does not distribute over addition.
4414 (fold_convert_const): Use tree_vector_builder and
4415 new_unary_operation. Operate only on the encoded elements
4416 for truncating integer conversions, or for non-stepped encodings.
4418 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4420 * config/sparc/sparc.c: Include tree-vector-builder.h.
4421 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4422 * expmed.c: Include tree-vector-builder.h.
4423 (make_tree): Use tree_vector_builder instead of build_vector.
4424 * fold-const.c: Include tree-vector-builder.h.
4425 (const_binop): Use tree_vector_builder instead of build_vector.
4426 (const_unop): Likewise.
4427 (native_interpret_vector): Likewise.
4428 (fold_vec_perm): Likewise.
4429 (fold_ternary_loc): Likewise.
4430 * gimple-fold.c: Include tree-vector-builder.h.
4431 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4433 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4434 (simplify_vector_constructor): Use tree_vector_builder instead
4436 * tree-vect-generic.c: Include tree-vector-builder.h.
4437 (add_rshift): Use tree_vector_builder instead of build_vector.
4438 (expand_vector_divmod): Likewise.
4439 (optimize_vector_constructor): Likewise.
4440 * tree-vect-loop.c: Include tree-vector-builder.h.
4441 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4442 of build_vector. Explicitly use a stepped encoding for
4444 * tree-vect-slp.c: Include tree-vector-builder.h.
4445 (vect_get_constant_vectors): Use tree_vector_builder instead
4447 (vect_transform_slp_perm_load): Likewise.
4448 (vect_schedule_slp_instance): Likewise.
4449 * tree-vect-stmts.c: Include tree-vector-builder.h.
4450 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4451 (vect_gen_perm_mask_any): Likewise.
4452 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4453 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4455 (build_vector_from_val): Likewise. Explicitly use a duplicate
4458 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4460 * doc/generic.texi (VECTOR_CST): Describe new representation of
4462 * vector-builder.h: New file.
4463 * tree-vector-builder.h: Likewise.
4464 * tree-vector-builder.c: Likewise.
4465 * Makefile.in (OBJS): Add tree-vector-builder.o.
4466 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4467 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4468 (tree_vector): Change the number of elements to
4469 vector_cst_encoded_nelts.
4470 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4471 (VECTOR_CST_ELTS): Delete.
4472 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4473 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4474 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4475 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4476 (VECTOR_CST_ENCODED_ELT): Likewise.
4477 (vector_cst_encoded_nelts): New function.
4478 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4479 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4480 (vector_cst_int_elt, vector_cst_elt): Declare.
4481 * tree.c: Include tree-vector-builder.h.
4482 (tree_code_size): Abort if passed VECTOR_CST.
4483 (tree_size): Update for new VECTOR_CST layout.
4484 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4485 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4486 (build_vector): Use tree_vector_builder.
4487 (vector_cst_int_elt, vector_cst_elt): New functions.
4488 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4489 encoded elements and then create the vector in the canonical form.
4490 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4491 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4492 (tree_c_tests): Call test_vector_cst_patterns.
4493 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4495 (hash_tree): Likewise.
4496 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4497 (streamer_write_tree_header): Likewise.
4498 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4499 (streamer_alloc_tree): Likewise. Update call to make_vector.
4500 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4502 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4504 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4505 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4506 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4508 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4509 Richard Biener <rguenther@suse.de>
4511 PR tree-optimization/81303
4512 * Makefile.in (gimple-loop-interchange.o): New object file.
4513 * common.opt (floop-interchange): Reuse the option from graphite.
4514 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4515 -floop-interchange and mention it for -O3.
4516 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4517 * gimple-loop-interchange.cc: New file.
4518 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4519 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4520 * passes.def (pass_linterchange): New pass.
4521 * timevar.def (TV_LINTERCHANGE): New time var.
4522 * tree-pass.h (make_pass_linterchange): New declaration.
4523 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4524 interchange. Record IV before/after increment in new parameters.
4525 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4526 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4528 (check_reduction_path): ...New function here.
4529 * tree-vectorizer.h (check_reduction_path): New declaration.
4531 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4534 PR rtl-optimization/80818
4535 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4536 always early clobbered.
4537 * lra-lives.c (process_bb_lives): Check input hard regs for early
4538 clobbered non-operand hard reg.
4540 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4543 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4544 types_compatible_p, just that TYPE_MODE is the same.
4546 2017-12-07 Martin Sebor <msebor@redhat.com>
4549 * attribs.c (empty_attribute_table): Initialize new member of
4550 struct attribute_spec.
4551 (decl_attributes): Add argument. Handle mutually exclusive
4552 combinations of attributes.
4553 (selftests::test_attribute_exclusions): New function.
4554 (selftests::attribute_c_tests): Ditto.
4555 * attribs.h (decl_attributes): Add default argument.
4556 * selftest.h (attribute_c_tests): Declare.
4557 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4558 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4560 * doc/extend.texi (Common Function Attributes): Update const and pure.
4561 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4562 of struct attribute_spec.
4563 * config/arc/arc.c (arc_attribute_table): Same.
4564 * config/arm/arm.c (arm_attribute_table): Same.
4565 * config/avr/avr.c ( avr_attribute_table): Same.
4566 * config/bfin/bfin.c (bfin_attribute_table): Same.
4567 * config/cr16/cr16.c (cr16_attribute_table): Same.
4568 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4569 * config/h8300/h8300.c (h8300_attribute_table): Same.
4570 * config/i386/i386.c (ix86_attribute_table): Same.
4571 * config/ia64/ia64.c (ia64_attribute_table): Same.
4572 * config/m32c/m32c.c (m32c_attribute_table): Same.
4573 * config/m32r/m32r.c (m32r_attribute_table): Same.
4574 * config/m68k/m68k.c (m68k_attribute_table): Same.
4575 * config/mcore/mcore.c (mcore_attribute_table): Same.
4576 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4577 * config/mips/mips.c (mips_attribute_table): Same.
4578 * config/msp430/msp430.c (msp430_attribute_table): Same.
4579 * config/nds32/nds32.c (nds32_attribute_table): Same.
4580 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4581 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4582 * config/rl78/rl78.c (rl78__attribute_table): Same.
4583 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4584 * onfig/rx/rx.c (rx_attribute_table): Same.
4585 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4586 * config/sh/sh.c (sh_attribute_table): Same.
4587 * config/sparc/sparc.c (sparc_attribute_table): Same.
4588 * config/spu/spu.c (spu_attribute_table): Same.
4589 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4590 * config/v850/v850.c (v850_attribute_table): Same.
4591 * config/visium/visium.c (visium_attribute_table): Same.
4593 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4596 * config/arm/arm.c (INCLUDE_STRING): Define.
4597 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4598 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4601 2017-12-07 Michael Matz <matz@suse.de>
4603 Add unroll and jam pass
4605 * gimple-loop-jam.c: New file.
4606 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4607 * common.opt (funroll-and-jam): New option.
4608 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4609 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4610 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4611 * passes.def: Add pass_loop_jam.
4612 * timevar.def (TV_LOOP_JAM): Add.
4613 * tree-pass.h (make_pass_loop_jam): Declare.
4614 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4615 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4616 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4617 to flow_loop_tree_node_add.
4618 (duplicate_subloops, copy_loops_to): Append to sibling list.
4619 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4620 * doc/invoke.texi (-funroll-and-jam): Document new option.
4621 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4623 2017-12-07 Richard Biener <rguenther@suse.de>
4625 PR tree-optimization/83296
4626 PR tree-optimization/67769
4627 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4628 flow sensitive info in an unrelated BB.
4629 (value_replacement): Use reset_flow_sensitive_info.
4630 (minmax_replacement): Reset flow sensitive info on the def
4631 we move. Do not reset flow sensitive info in the whole BB
4632 we move the stmt to.
4633 (abs_replacement): Likewise.
4635 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4638 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4639 rs6000_cpu to the given -mcpu=, or to the default processor.
4641 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4643 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4644 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4645 instead of rs6000_cpu_attr.
4646 (rs6000_adjust_cost): Ditto.
4647 (is_microcoded_insn): Ditto.
4648 (rs6000_adjust_priority): Ditto.
4649 (rs6000_issue_rate): Ditto.
4650 (rs6000_use_sched_lookahead): Ditto.
4651 (rs6000_use_sched_lookahead_guard): Ditto.
4652 (rs6000_sched_reorder): Ditto.
4653 (force_new_group): Ditto.
4654 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4655 (group_ending_nop): Ditto.
4657 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4659 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4660 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4661 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4662 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4664 (rs6000_reassociation_width): Ditto.
4665 (rs6000_emit_epilogue): Ditto.
4666 (rs6000_adjust_cost): Ditto.
4667 (is_microcoded_insn): Ditto.
4668 (is_cracked_insn): Ditto.
4669 (rs6000_adjust_priority): Ditto.
4670 (rs6000_sched_reorder): Ditto.
4671 (rs6000_sched_reorder2): Ditto.
4672 (insn_must_be_first_in_group): Ditto.
4673 (insn_must_be_last_in_group): Ditto.
4674 (rs6000_register_move_cost): Ditto.
4675 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4678 2017-12-07 Julia Koval <julia.koval@intel.com>
4680 * config.gcc: Add vaesintrin.h.
4681 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4683 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4684 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4686 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4687 * config/i386/immintrin.h: Include vaesintrin.h.
4688 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4689 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4690 _mm_aesdec_epi128): New intrinsics.
4692 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4694 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4695 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4696 c-family/c-spellcheck.cc.
4697 (best_macro_match::best_macro_match): Likewise.
4699 (struct edit_distance_traits<cpp_hashnode *>): Move to
4700 c-family/c-spellcheck.h.
4701 (class best_macro_match): Likewise.
4703 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4705 PR tree-optimization/83293
4706 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4707 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4708 might insert into empty bb.
4711 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4712 simplify for plus with :c added, and pointer_plus without that.
4713 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4714 with undefined overflow and the conversion is not widening,
4715 perform negation in utype and only convert to type afterwards.
4716 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4717 simplify for plus with :c added, and pointer_plus without that.
4718 If type is integral with undefined overflow and the conversion is
4719 not widening, perform minus in utype and only convert to type
4720 afterwards. Move the last pointer_diff_expr simplify into the
4723 2017-12-06 Martin Sebor <msebor@redhat.com>
4725 PR tree-optimization/82646
4726 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4727 strncpy, not maxlen.
4729 2017-12-06 Martin Sebor <msebor@redhat.com>
4731 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4734 PR tree-optimization/83075
4735 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4736 strncat/strncpy don't change length of source string.
4738 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4741 2017-11-29 Martin Aberg <maberg@gaisler.com>
4743 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4744 to prevent b2bst errata sequence.
4745 (sqrtdf2_fix): Likewise.
4747 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4749 PR tree-optimization/81945
4750 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4751 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4752 to dest_cfun has orig_loop_num set, either remap it to the new
4753 loop number if the loop got moved too, or clear it.
4755 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4757 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4759 (thunderx2t99_nothing): New insn reservation.
4760 (thunderx2t99_mrs): New insn reservation.
4761 (thunderx2t99_multiple): New insn reservation.
4762 (thunderx2t99_alu_basi): Add bfx to reservation.
4763 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4765 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4768 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4770 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4772 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4773 tree-ssa-pre.c::remove_dead_inserted_code.
4774 * tree-ssa-dce.h: New file.
4775 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4776 (remove_dead_inserted_code): Move and rename to function
4777 tree-ssa-dce.c::simple_dce_from_worklist.
4778 (pass_pre::execute): Update use.
4780 2017-12-05 Richard Biener <rguenther@suse.de>
4782 PR tree-optimization/83277
4783 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4784 to code-gen liveout vars.
4786 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4788 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4789 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4790 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4793 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4795 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4797 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4798 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4801 2017-12-05 Richard Biener <rguenther@suse.de>
4803 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4804 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4805 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4806 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4808 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4810 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4812 2017-12-05 Martin Liska <mliska@suse.cz>
4813 Jakub Jelinek <jakub@redhat.com>
4815 * doc/invoke.texi: Document the options.
4816 * flag-types.h (enum sanitize_code): Add
4817 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4818 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4819 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4820 * opts.c: Define new sanitizer options.
4821 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4822 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4824 2017-12-05 Julia Koval <julia.koval@intel.com>
4826 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4827 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4828 (ix86_handle_option): Handle -mavx512vnni.
4829 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4830 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4831 * config/i386/i386-c (__AVX512VNNI__): New.
4832 * config/i386/i386.c (ix86_target_string): Handle new option.
4833 (ix86_valid_target_attribute_inner_p): Handle new option.
4834 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4835 * config/i386/i386.opt (mavx512vnni): New option.
4837 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4840 * athlon.md: Disable for generic.
4841 * haswell.md: Enable for generic.
4842 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4843 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4845 (ix86_adjust_cost): Move generic to haswell path.
4847 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4849 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4850 instead of MEM_P in a couple more places. Fix formatting issues.
4852 2017-12-04 Jim Wilson <jimw@sifive.com>
4854 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4855 instead of GP_REG_LAST-1.
4856 (riscv_adjust_libcall_cfi_prologue): Likewise.
4857 (riscv_adjust_libcall_cri_epilogue): Likewise.
4858 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4861 2017-12-04 Luis Machado <luis.machado@linaro.org>
4863 * ipa-pure-const.c (check_decl): Add missing newline.
4864 (state_from_flags): Likewise.
4866 2017-12-04 Jeff Law <law@redhat.com>
4868 PR tree-optimizatin/78496
4869 * gimple-ssa-evrp-analyze.h
4870 (evrp_range_analyzer::get_vr_values): Simplify.
4871 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4872 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4873 and gimple-ssa-evrp-analyze.h.
4874 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4875 (simplify_stmt_for_jump_threading): Copy a blob of code from
4876 tree-vrp.c to use ranges to simplify statements.
4877 (dom_opt_dom_walker::before_dom_children): Call
4878 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4879 (dom_opt_dom_walker::after_dom_children): Similarly for
4880 evrp_range_analyzer::leave.
4881 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4884 * gimple-ssa-evrp-analyze.c
4885 (evrp_range_analyzer::extract_range_from_stmt): Always use
4886 vr_values::update_value_range so preexisting range info is
4887 medged with new range info, even if the new range is VR_VARYING.
4889 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4891 * combine.c: Adjust comment.
4892 (use_crosses_set_p): Delete.
4893 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4894 (try_combine): Ditto.
4896 2017-12-04 Richard Biener <rguenther@suse.de>
4898 PR tree-optimization/83255
4899 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4900 Re-add zero-iteration check.
4902 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4904 PR rtl-optimization/83245
4905 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4906 hard registers as earlyclobber, also if not in an asm.
4908 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4912 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4915 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4916 rs6000_cpu based on cpu_index, not tune_index.
4918 2017-12-04 Richard Biener <rguenther@suse.de>
4920 PR tree-optimization/83238
4921 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4922 code match comment, rejecting invalid SESE regions.
4924 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4926 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4927 require base operand is a REG_POINTER prior to reload on targets
4928 with non-equivalent space registers.
4930 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4932 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4933 (update_specialized_profile): Fix updating of counts.
4934 (perhaps_add_new_callers): Likewise.
4936 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4939 * x86-tune.def: Remove obsolette FIXMEs.
4940 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4941 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4942 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4944 (X86_TUNE_PAD_RETURNS): Disable for generic.
4946 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4948 PR tree-optimization/83170
4949 PR tree-optimization/83241
4950 * gimple-ssa-store-merging.c
4951 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4952 gimple_vuse (ins_stmt) in case it has changed.
4953 (imm_store_chain_info::output_merged_store): Likewise.
4955 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4959 * tree-cfg.c (pass_warn_function_return::execute): Handle
4960 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4961 with BUILTINS_LOCATION.
4965 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4966 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4967 instead of DECL_MODE.
4969 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4970 last argument to rtx pointer.
4971 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4972 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4973 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4974 depending on the chosen ISAs.
4975 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4976 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4977 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4979 * config/i386/sse.md (mov<mode>_internal): Likewise.
4980 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4982 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4984 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4986 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4988 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4989 parameters from prototype.
4990 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4991 parameters. Don't print a comment.
4992 (emit_fusion_gpr_load): Adjust.
4993 (emit_fusion_load_store): Adjust.
4994 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4995 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4996 comment on the second line.
4998 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5001 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5002 rs6000_cpu based on cpu_index, not tune_index.
5004 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5006 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5007 which_alternative instead of which_alternative + 1.
5008 (output_asm_insn): Print an extra tab if the template is short.
5010 2017-12-01 Jim Wilson <jimw@sifive.com>
5012 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5014 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5015 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5017 * doc/tm.texi: Regenerate.
5019 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5022 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5023 whether we can allocate pseudos before trying to fix an address.
5024 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5025 memory address is indexed or indirect.
5026 (floatuns_<mode>si2_hw2): Likewise.
5028 2017-12-01 Jason Merrill <jason@redhat.com>
5030 * Makefile.in (TAGS): Add c-family/*.cc.
5032 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5034 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5035 (call_value_insn): Likewise.
5036 (sibcall_insn): Likewise.
5037 (sibcall_value_insn): Likewise.
5038 (movsi_aarch64): Likewise.
5039 (movdi_aarch64): Likewise.
5040 (add_losym_): Likewise.
5041 (ldr_got_small_): Likewise.
5042 (ldr_got_small_sidi): Likewise.
5043 (ldr_got_small_28k_): Likewise.
5044 (ldr_got_small_28k_sidi): Likewise.
5045 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5046 Move output_addr_const to symbolic case. Add error check.
5048 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5050 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5051 * config/s390/s390.md: Disable MVC merging peephole if it would
5052 disable operand forwarding.
5053 (new peephole2): Split MVCs if it would turn them into up to 2
5056 2017-12-01 Richard Biener <rguenther@suse.de>
5058 PR tree-optimization/83232
5059 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5060 detection of same access. Instead of breaking the group here
5061 do not consider the duplicate. Add comment explaining real fix.
5063 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5065 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5067 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5069 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5071 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5073 * function.h (struct function): Remove cilk_frame_decl,
5074 is_cilk_function and calls_cilk_spawn fields.
5075 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5077 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5078 cilk_elemental field.
5079 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5080 * target.def: Adjust comment.
5081 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5082 Don't test cilk_elemental.
5084 PR tree-optimization/83233
5085 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5086 bswap_stat name for the struct.
5089 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5090 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5091 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5092 SWITCH_BREAK_LABEL_P set on the label.
5093 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5094 added for default case if it was missing and not all cases covered.
5095 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5096 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5098 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5099 to avoid -Wimplicit-fallthrough warning.
5100 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5101 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5103 PR tree-optimization/83221
5104 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5106 (init_reassoc): Formatting fix.
5109 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5110 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5114 * tree-cfg.c: Include asan.h.
5115 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5117 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5119 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5121 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5123 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5124 vpcmpeqd instruction.
5126 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5128 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5129 registers type for 128bit mode.
5131 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5133 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5135 * gengtype-state.c (read_state_pair): Likewise.
5136 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5137 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5138 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5139 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5140 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5141 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5142 * builtins.c (fold_builtin_3): Likewise.
5143 * graphite-scop-detection.c
5144 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5145 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5148 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5151 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5152 parameter. Return true if ignore_decl_rtl_set_p is true and other
5153 conditions are satisfied.
5154 * asan.h (asan_protect_global): Add new parameter.
5155 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5156 to asan_protect_global calls.
5158 2017-11-30 Jim Wilson <jimw@sifive.com>
5160 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5161 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5162 -mstrict-align, add info on default value. Delete redundant lines for
5163 -mabi. Add missing -mexplicit-relocs docs.
5165 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5167 * config/arc/arc.md (trap): New pattern.
5169 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5171 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5172 instruction to end into a delay slot.
5173 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5174 can be placed into a delay slot against reg_note.
5176 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5178 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5179 labels number of usages.
5181 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5183 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5185 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5187 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5190 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5191 multiplication by power of 2 constant into two shifts + comparison.
5193 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5196 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5198 2017-11-30 Richard Biener <rguenther@suse.de>
5200 PR tree-optimization/83202
5201 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5202 allow_peel argument and guard peeling.
5203 (canonicalize_loop_induction_variables): Likewise.
5204 (canonicalize_induction_variables): Pass false.
5205 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5206 peeling from cunrolli.
5208 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5210 * combine.c (try_combine): Print a message to dump file whenever
5211 I0, I1, or I2 cannot be combined into I3.
5213 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5215 PR rtl-optimization/83156
5216 PR rtl-optimization/82621
5217 * combine.c (try_combine): Don't split an I2 if one of the dests is
5218 set again before I3. Allow unused dests.
5220 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5222 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5224 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5226 PR rtl-optimization/80818
5227 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5228 recursively. Use insn code for clobber.
5229 (lra_set_insn_recog_data): Pass the new arg to
5230 collect_non_operand_hard_regs.
5231 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5233 (lra_update_insn_regno_info): Pass insn to
5234 add_regs_to_insn_regno_info.
5236 2017-11-29 Jim Wilson <jimw@sifive.com>
5237 Andrew Waterman <andrew@sifive.com>
5239 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5240 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5241 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5242 use SINGLE_SHIFT_COST.
5243 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5244 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5246 2017-11-29 Julia Koval <julia.koval@intel.com>
5248 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5249 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5250 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5251 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5252 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5253 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5254 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5255 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5256 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5257 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5258 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5259 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5260 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5261 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5262 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5263 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5264 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5265 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5266 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5267 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5268 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5269 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5270 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5271 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5272 __builtin_ia32_vpshldv_v2di_mask,
5273 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5274 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5275 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5277 2017-11-29 Julia Koval <julia.koval@intel.com>
5279 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5280 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5281 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5282 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5283 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5284 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5285 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5286 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5287 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5288 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5289 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5290 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5291 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5292 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5293 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5294 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5295 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5296 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5297 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5298 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5299 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5301 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5303 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5304 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5307 2017-11-29 Martin Aberg <maberg@gaisler.com>
5309 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5310 to prevent b2bst errata sequence.
5311 (sqrtdf2_fix): Likewise.
5313 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5315 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5316 (div_sqrt_insn_p): New function.
5317 (sparc_do_work_around_errata): Insert NOP instructions to
5318 prevent sequences that could trigger the TN-0013 errata for
5319 certain LEON3 processors.
5320 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5321 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5322 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5323 (in_branch_delay): Prevent div and sqrt in delay slot if
5325 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5327 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5329 * config/sparc/sparc.c (atomic_insn_p): New function.
5330 (sparc_do_work_around_errata): Insert NOP instructions to
5331 prevent sequences that could trigger the TN-0010 errata for
5333 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5334 instruction referable in atomic_insns_p.
5336 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5338 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5339 (atomic_compare_and_swap_leon3_1): Likewise.
5342 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5344 * config/sparc/sparc.c (fpop_insn_p): New function.
5345 (sparc_do_work_around_errata): Insert NOP instructions to
5346 prevent sequences that could trigger the TN-0012 errata for
5348 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5349 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5350 (in_branch_annul_delay): Prevent floating-point instructions
5351 in delay slot of annulled integer branch.
5353 2017-11-29 Richard Biener <rguenther@suse.de>
5355 PR tree-optimization/83202
5356 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5358 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5359 nodes vectorized to the same stmts multiple times.
5360 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5361 (vect_analyze_slp_instance): Adjust.
5362 (scalar_stmts_to_slp_tree_map_t): New typedef.
5363 (vect_schedule_slp_instance): Add a map recording the SLP node
5364 representing the vectorized stmts for a set of scalar stmts.
5365 Avoid code-generating redundancies.
5366 (vect_schedule_slp): Allocate map and pass it down.
5368 2017-11-29 Nathan Sidwell <nathan@acm.org>
5371 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5374 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5376 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5377 cases for printing LDP/STP memory addresses.
5378 (aarch64_print_address_internal): Renamed from
5379 aarch64_print_operand_address, added parameter, add Pmode check.
5380 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5381 (aarch64_print_operand_address): Indirect to
5382 aarch64_print_address_internal.
5383 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5384 'y' operand output specifier.
5386 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5389 * tree.c (build_simple_mem_ref_loc): Handle
5390 get_addr_base_and_unit_offset returning a MEM_REF.
5393 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5396 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5399 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5401 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5403 (layout::move_to_column): Likewise.
5405 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5408 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5409 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5411 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5412 Martin Jambor <mjambor@suse.cz>
5415 * tree.h (expr_type_first_operand_type_p): Declare
5416 * tree.c (expr_type_first_operand_type_p): New function.
5417 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5418 (ipa_value_from_jfunc): Adjust declaration.
5419 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5420 Use it as result type for arithmetics, unless it is NULL in which case
5421 be more conservative.
5422 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5423 ipa_get_jf_pass_through_result.
5424 (propagate_vals_across_pass_through): Likewise.
5425 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5426 is to propagate_vals_across_pass_through.
5427 (propagate_constants_across_call): Pass PARM_TYPE to
5428 propagate_scalar_across_jump_function.
5429 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5430 ipa_value_from_jfunc.
5431 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5432 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5433 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5434 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5435 target_type, pass it to ipa_value_from_jfunc.
5436 (update_indirect_edges_after_inlining): Pass parameter type to
5437 try_make_edge_direct_simple_call.
5439 2017-11-28 Jeff Law <law@redhat.com>
5441 * gimple-ssa-evrp-analyze.c
5442 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5443 refine ranges if scev_initialized_p returns true.
5444 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5446 2017-11-28 Julia Koval <julia.koval@intel.com>
5448 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5449 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5450 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5451 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5452 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5453 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5454 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5455 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5456 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5457 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5458 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5459 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5460 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5461 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5462 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5463 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5464 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5465 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5466 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5467 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5468 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5469 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5471 2017-11-28 Julia Koval <julia.koval@intel.com>
5473 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5474 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5475 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5476 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5477 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5478 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5479 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5480 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5481 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5482 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5483 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5484 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5485 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5486 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5487 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5488 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5489 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5490 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5491 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5492 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5493 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5494 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5495 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5496 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5497 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5498 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5499 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5500 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5501 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5502 __builtin_ia32_vpshld_v2di_mask): New builtins.
5503 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5504 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5506 2017-11-28 Richard Biener <rguenther@suse.de>
5508 PR tree-optimization/80776
5509 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5511 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5513 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5514 If the incoming edge is an effective fallthru because the other
5515 edge only reaches a __builtin_unreachable () then record ranges
5516 derived from the controlling condition in SSA info.
5517 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5518 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5520 2017-11-28 Olivier Hainque <hainque@adacore.com>
5522 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5523 on mingw build hosts.
5525 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5528 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5530 2017-11-28 Richard Biener <rguenther@suse.de>
5533 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5534 copies generated from memcpy use a character array as reference
5537 2017-11-28 Julia Koval <julia.koval@intel.com>
5538 Sebastian Peryt <sebastian.peryt@intel.com>
5540 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5541 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5542 cilk-common.o, cilk.h, cilk-common.c): Remove.
5544 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5545 * builtins.c (is_builtin_name): Remove cilkplus condition.
5546 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5547 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5548 cilk-builtins.def, cilkplus.def): Remove.
5549 * cif-code.def (CILK_SPAWN): Remove.
5550 * cilk-builtins.def: Delete.
5551 * cilk-common.c: Ditto.
5553 * cilkplus.def: Ditto.
5554 * config/darwin.h (fcilkplus): Delete.
5555 * cppbuiltin.c: Ditto.
5556 * doc/extend.texi: Remove cilkplus doc.
5557 * doc/generic.texi: Ditto.
5558 * doc/invoke.texi: Ditto.
5559 * doc/passes.texi: Ditto.
5560 * gcc.c (fcilkplus): Remove.
5561 * gengtype.c (cilk.h): Remove.
5562 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5564 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5566 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5568 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5569 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5570 cilkplus conditions.
5571 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5572 inline_read_section): Ditto.
5573 * ipa-inline-analysis.c (cilk.h): Remove.
5574 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5575 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5576 append_linker_options): Remove condition for fcilkplus.
5577 * lto/lto-lang.c (cilk.h): Remove.
5578 (lto_init): Remove condition for fcilkplus.
5579 * omp-expand.c (expand_cilk_for_call): Delete.
5580 (expand_omp_taskreg, expand_omp_for_static_chunk,
5581 expand_omp_for): Remove cilkplus
5583 (expand_cilk_for): Delete.
5584 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5585 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5586 execute_lower_omp, diagnose_sb_0): Ditto.
5587 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5588 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5589 * tree-nested.c: Ditto.
5590 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5591 (dump_generic_node): Ditto.
5592 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5593 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5595 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5597 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5599 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5600 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5601 for vector mode and !TARGET_SIMD.
5603 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5605 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5607 * tree.h (SWITCH_LABELS): Remove.
5608 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5609 assert SWITCH_BODY is non-NULL.
5610 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5612 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5614 PR tree-optimization/80788
5615 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5616 has TREE_OVERFLOW set, call drop_tree_overflow.
5618 2017-11-28 Richard Biener <rguenther@suse.de>
5620 PR tree-optimization/83158
5621 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5623 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5626 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5627 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5629 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5631 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5632 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5634 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5637 * varasm.c (compare_constant): Take the mode of the constants into
5638 account when comparing floating point constants.
5640 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5642 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5643 from explicit instantiation of debug_helper.
5644 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5646 2017-11-27 Richard Biener <rguenther@suse.de>
5648 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5651 2017-11-27 Richard Biener <rguenther@suse.de>
5653 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5654 (type_hash_canon): Also clear int_cst_hash_table entry for
5656 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5658 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5660 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5661 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5662 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5664 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5666 * hash-map.h (gt_cleare_cache): Avoid UB.
5668 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5670 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5671 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5672 (decide_unroll_runtime_iterations): Likewise.
5673 (decide_unroll_stupid): Likewise.
5675 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5678 * config/i386/i386.md: Add a loop with incssp.
5680 2017-11-27 Martin Jambor <mjambor@suse.cz>
5682 PR tree-optimization/81248
5683 * tree-sra.c (splice_param_accesses): Remove size check.
5684 (decide_one_param_reduction): Fix size check.
5685 * gimple-pretty-print.c (dump_profile): Silence warning.
5686 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5688 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5691 * dbxout.c (lastlineno): New variable.
5692 (dbx_debug_hooks): Use dbxout_switch_text_section as
5693 switch_text_section debug hook.
5694 (dbxout_function_end): Switch to current_function_section
5695 rather than function_section. If crtl->has_bb_partition,
5696 output just one N_FUN, depending on in_cold_section_p.
5697 (dbxout_source_line): Remember last lineno in lastlineno.
5698 (dbxout_switch_text_section): New function.
5699 (dbxout_function_decl): Adjust dbxout_block caller.
5700 (dbx_block_with_cold_children): New function.
5701 (dbxout_block): Return true if any LBRAC/RBRAC have been
5702 emitted. Use dbx_block_with_cold_children at depth == 0
5703 in second partition. Add PARENT_BLOCKNUM argument, pass
5704 it optionally adjusted to children. Output LBRAC/RBRAC
5705 around recursive call only if the block is in the current
5706 partition, if not and anything was output, emit empty
5708 * final.c (final_scan_insn): Compute cold_function_name
5709 before calling switch_text_section debug hook. Call
5710 that hook even if dwarf2out_do_frame if not emitting
5714 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5715 TREE_READONLY decls.
5717 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5719 PR rtl-optimization/82488
5720 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5722 2017-11-26 Julia Koval <julia.koval@intel.com>
5724 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5726 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5729 2017-11-26 Julia Koval <julia.koval@intel.com>
5731 * config/i386/driver-i386.c (host_detect_local_cpu):
5732 Detect skylake-avx512.
5734 2017-11-26 Julia Koval <julia.koval@intel.com>
5736 * config.gcc: Add -march=cannonlake.
5737 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5738 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5739 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5740 (PTA_CANNONLAKE): New.
5741 (processor_target_table): Add cannonlake.
5742 (ix86_option_override_internal): Ditto.
5743 (fold_builtin_cpu): Ditto.
5744 (get_builtin_code_for_version): Handle cannonlake.
5745 (M_INTEL_COREI7_CANNONLAKE): New.
5746 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5747 * doc/invoke.texi: Add -march=cannonlake.
5749 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5751 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5752 (try_init_one_plugin): Alternative implementation for MinGW.
5753 * Makefile.in (plugin_implib): New.
5754 (gengtype-lex.c): Fix broken AIX workaround.
5755 * configure: Regenerate.
5756 * doc/plugins.texi: Document support for MinGW.
5758 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5760 PR rtl-optimization/81553
5761 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5762 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5763 is a shift where C1 has different mode than the whole shift, use C1's
5764 mode for MULT rather than the shift's mode.
5767 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5768 builtins not enabled in the currently selected ISA.
5770 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5772 PR tree-optimization/71026
5773 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5774 (insert_reciprocals): Change to insert reciprocals before a division
5775 by a square and to insert the square of a reciprocal.
5776 (execute_cse_reciprocals_1): Change to consider division by a square.
5777 (register_division_in): Add importance parameter.
5779 2017-11-24 Richard Biener <rguenther@suse.de>
5781 PR tree-optimization/82402
5782 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5783 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5785 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5787 * match.pd (0-ptr): New transformation.
5789 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5792 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5793 when skipping checking.
5795 2017-11-24 Richard Biener <rguenther@suse.de>
5797 PR tree-optimization/83128
5798 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5799 (vn_reference_lookup_3): Likewise.
5801 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5804 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5805 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5808 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5809 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5810 check for the latter with gimple_call_builtin_p. Do not handle
5811 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5813 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5815 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5818 2017-11-23 Julia Koval <julia.koval@intel.com>
5820 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5821 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5822 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5823 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5824 _mm512_maskz_expandloadu_epi16): New intrinsics.
5825 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5826 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5827 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5828 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5829 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5830 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5831 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5832 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5833 _mm256_maskz_expandloadu_epi8): New intrinsics.
5834 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5835 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5836 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5837 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5838 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5839 * config/i386/sse.md (VI248_VLBW): New iterator.
5840 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5842 2017-11-23 Julia Koval <julia.koval@intel.com>
5844 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5845 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5846 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5847 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5848 _mm512_mask_compressstoreu_epi16): New.
5849 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5850 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5851 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5852 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5853 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5854 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5855 _mm256_mask_compressstoreu_epi8): New.
5856 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5857 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5858 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5859 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5860 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5861 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5862 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5863 __builtin_ia32_compresshi128_mask,
5864 __builtin_ia32_compressstoreuqi512_mask,
5865 __builtin_ia32_compressstoreuhi512_mask,
5866 __builtin_ia32_compressstoreuqi256_mask,
5867 __builtin_ia32_compressstoreuqi128_mask,
5868 __builtin_ia32_compressstoreuhi256_mask,
5869 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5870 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5872 (ix86_expand_special_args_builtin): Handle new types.
5873 (s4fma_expand): Handle new builtin array.
5874 * config/i386/immintrin.h: Include new headers.
5875 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5876 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5878 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5880 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5881 fixes. Declare temp and g variables at the top in order to avoid
5882 {} in most of the cases.
5884 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5886 * match.pd (ptr-0): New transformation.
5888 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5890 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5891 (struct addr_mode_cost_table): New.
5892 (struct tune_params): Add field addr_mode_costs.
5893 * config/arm/arm.c (generic_addr_mode_costs): New.
5894 (arm_slowmul_tune): Initialise addr_mode_costs field.
5895 (arm_fastmul_tune): Likewise.
5896 (arm_strongarm_tune): Likewise.
5897 (arm_xscale_tune): Likewise.
5898 (arm_9e_tune): Likewise.
5899 (arm_marvell_pj4_tune): Likewise.
5900 (arm_v6t2_tune): Likewise.
5901 (arm_cortex_tune): Likewise.
5902 (arm_cortex_a8_tune): Likewise.
5903 (arm_cortex_a7_tune): Likewise.
5904 (arm_cortex_a15_tune): Likewise.
5905 (arm_cortex_a35_tune): Likewise.
5906 (arm_cortex_a53_tune): Likewise.
5907 (arm_cortex_a57_tune): Likewise.
5908 (arm_exynosm1_tune): Likewise.
5909 (arm_xgene1_tune): Likewise.
5910 (arm_cortex_a5_tune): Likewise.
5911 (arm_cortex_a9_tune): Likewise.
5912 (arm_cortex_a12_tune): Likewise.
5913 (arm_cortex_a73_tune): Likewise.
5914 (arm_v7m_tune): Likewise.
5915 (arm_cortex_m7_tune): Likewise.
5916 (arm_v6m_tune): Likewise.
5917 (arm_fa726te_tune): Likewise.
5918 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5920 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5922 * config/arm/arm.c (arm_mem_costs): New function.
5923 (arm_rtx_costs_internal): Use arm_mem_costs.
5925 2017-11-23 Mark Wielaard <mark@klomp.org>
5927 * dwarf2out.c (init_sections_and_labels): Use generation to create
5928 unique ranges_section_label and ranges_base_label. Return generation.
5929 (output_rnglists): Add generation argument. Use generation to create
5930 unique ranges labels.
5931 (dwarf2out_finish): Get generation from init_sections_and_labels
5932 and pass generation to output_rnglists.
5934 2017-11-23 Mike Stump <mikestump@comcast.net>
5935 Eric Botcazou <ebotcazou@adacore.com>
5937 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5938 * cfgloop.h (struct loop): Add unroll field.
5939 * function.h (struct function): Add has_unroll bitfield.
5940 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5941 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5942 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5943 (pass_rtl_unroll_loops::gate): Likewise.
5944 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5945 for which loop->unroll==1.
5946 (decide_unroll_constant_iterations): Use note for consistency and
5947 take loop->unroll into account. Return early if loop->unroll is set.
5948 Fix thinko in existing test.
5949 (decide_unroll_runtime_iterations): Use note for consistency and
5950 take loop->unroll into account.
5951 (decide_unroll_stupid): Likewise.
5952 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5953 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5954 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5956 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5957 (print_loop): Print loop->unroll if set.
5958 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5959 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5960 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5962 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5963 loop->unroll is set and smaller than the trip count. Otherwise bypass
5964 entirely the heuristics if loop->unroll is set. Remove dead note.
5965 Fix off-by-one bug in other note.
5966 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5967 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5969 (tree_unroll_loops_completely): Make static.
5970 (pass_complete_unroll::execute): Use correct type for variable.
5971 (pass_complete_unrolli::execute): Fix formatting.
5972 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5974 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5976 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5977 enable when TARGET_PREFER_AVX128 is set.
5979 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5981 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5983 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5984 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5986 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5988 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5989 frequencies. Fix estimation of aggregate parameters.
5991 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5993 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5996 2017-11-23 Tom de Vries <tom@codesourcery.com>
5998 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
6000 2017-11-23 Tom de Vries <tom@codesourcery.com>
6002 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
6005 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6008 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6009 sibcall_value_pcrel_fdpic): Use local variable instead of
6011 (calli_tbr_rel): Add missing operand 2.
6012 (call_valuei_tbr_rel): Add missing operand 3.
6014 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6017 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6018 bitpos/bitsize covering the whole destination, use store_expr only if
6019 the complex mode is the same. Otherwise, use expand_normal and if
6020 it returns CONCAT, subreg each part separately instead of trying to
6021 subreg the whole result.
6023 2017-11-23 Richard Biener <rguenther@suse.de>
6025 PR tree-optimization/23094
6026 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6027 come here from walking over backedges in the first iteration.
6028 (vn_reference_lookup_3): Skip clobbers that store the same value.
6030 2017-11-23 Richard Biener <rguenther@suse.de>
6032 PR tree-optimization/81403
6033 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6034 a block we need a leader relative to.
6035 (phi_translate_1): For nary processing require a leader from
6036 get_representative_for given we run expression simplification
6037 using match-and-simplify. Remove previous fix.
6039 2017-11-22 Jeff Law <law@redhat.com>
6041 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6042 Use new method allocate_value_range rather than accessing the
6043 vrp_value_range_pool data member directly.
6044 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6045 to use extract_range_from_stmt method to avoid need for
6046 extract_range_from_assignment method.
6047 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6048 method rather than setting values_propgated data member directly.
6049 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6050 and values propagated data members and extract_range_from_assignment
6051 method. Reorder private data members to conform to standards.
6052 Add new methods set_lattice_propagation_complete and
6053 allocate_value_range.
6055 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6057 PR rtl-optimization/83030
6058 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6059 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6060 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6061 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6062 CROSSING_JUMP_P flag is already set before setting it.
6063 * cfgrtl.c (fixup_partition_crossing): Likewise.
6064 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6067 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6069 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6070 Handle the case where both arguments are using gen_const_vec_series.
6072 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6075 * pretty-print.c (pp_format): Move quoting implementation to
6076 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6077 to pass address of "quote" local.
6078 (pp_begin_quote): New function.
6079 (pp_end_quote): New function.
6080 * pretty-print.h (printer_fn): Convert penultimate param from bool
6082 (pp_begin_quote): New decl.
6083 (pp_end_quote): New decl.
6084 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6085 param from bool to bool *.
6086 * tree-diagnostic.h (default_tree_printer): Likewise.
6088 2017-11-22 Jeff Law <law@redhat.com>
6090 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6091 Initialize vr_values.
6092 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6093 vr_values via vr_values class instance rather than delegators.
6094 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6095 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6096 (evrp_range_analyzer::push_value_range): Likewise.
6097 (evrp_range_analyzer::pop_value_range): Likewise.
6098 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6099 most delegators. Those remaining are exposed as public interfaces.
6100 Make vr_values a pointer and private.
6101 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6103 (evrp_range_analyzer::get_vr_value): New method.
6104 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6105 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6106 (class evrp_dom_walker): Attach evrp_folder class, initialize
6107 it in the ctor. Remove temporary delegators.
6108 (evrp_dom_walker::before_dom_children): Call methods in attached
6109 evrp_range_analyzer class via class instance pointer. Use
6110 free value_range_constant_singleton to remove need for
6111 op_with_constant_singleton_value delegator method. Do not
6112 create a vrp_prop class instance for every call! Narrow
6113 scope of a couple variables.
6114 (evrp_dom_walker::cleanup): Call methods in attached
6115 evrp_range_analyzer class via class instance pointer.
6116 * vr-values.h (class vr_values): Privatize many methods and
6119 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6121 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6122 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6123 * expr.c (expand_expr_real_2): Likewise.
6124 * fold-const.c (const_unop): Likewise.
6125 * optabs-tree.c (optab_for_tree_code): Likewise.
6126 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6127 * tree-inline.c (estimate_operator_cost): Likewise.
6128 * tree-pretty-print.c (dump_generic_node): Likewise.
6129 (op_code_prio): Likewise.
6130 (op_symbol_code): Likewise.
6131 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6132 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6133 * internal-fn.c (direct_internal_fn_optab): New function.
6134 (direct_internal_fn_array, direct_internal_fn_supported_p
6135 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6136 * fold-const-call.c (fold_const_reduction): New function.
6137 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6139 * tree-vect-loop.c: Include internal-fn.h.
6140 (reduction_code_for_scalar_code): Rename to...
6141 (reduction_fn_for_scalar_code): ...this and return an internal
6143 (vect_model_reduction_cost): Take an internal_fn rather than
6145 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6147 (vectorizable_reduction): Use internal functions rather than tree
6148 codes for the reduction operation. Update calls to the functions
6150 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6151 Use calls to internal functions rather than REDUC tree codes.
6152 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6154 2017-11-22 Olivier Hainque <hainque@adacore.com>
6156 * config/vxworks.c (vxworks_override_options): Pick default
6157 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6158 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6159 DWARF_GNAT_ENCODINGS_DEFAULT.
6160 * config/vxworksae.h: Likewise.
6162 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6164 PR tree-optimization/83104
6165 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6168 2017-11-22 Marek Polacek <polacek@redhat.com>
6169 H.J. Lu <hongjiu.lu@intel.com>
6170 Jason Merrill <jason@redhat.com>
6175 * calls.c (initialize_argument_information): Call
6176 warn_parameter_passing_abi target hook.
6177 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6178 argument onto stack.
6179 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6180 * common.opt: Update -fabi-version description.
6181 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6182 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6184 (ix86_is_empty_record): New function.
6185 (ix86_warn_parameter_passing_abi): New function.
6186 (TARGET_EMPTY_RECORD_P): Redefine.
6187 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6188 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6189 * doc/tm.texi: Regenerated.
6190 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6191 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6192 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6193 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6194 instead of int_size_in_bytes.
6195 * expr.c (copy_blkmode_to_reg): Likewise.
6196 * function.c (aggregate_value_p): Return 0 for empty types.
6197 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6199 (locate_and_pad_parm): Call arg size_in_bytes instead
6201 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6202 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6203 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6205 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6206 (std_gimplify_va_arg_expr): Skip empty records. Call
6207 arg_size_in_bytes instead size_in_bytes.
6208 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6209 * tree-core.h (tree_type_common): Add empty_flag.
6210 (tree_decl_common): Update comments.
6211 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6213 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6214 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6216 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6217 * tree.c (default_is_empty_type): New function.
6218 (default_is_empty_record): New function.
6219 (arg_int_size_in_bytes): New function.
6220 (arg_size_in_bytes): New function.
6221 (get_ultimate_context): New function.
6222 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6223 TRANSLATION_UNIT_WARN_EMPTY_P.
6224 (default_is_empty_record, arg_int_size_in_bytes,
6225 arg_size_in_bytes, get_ultimate_context): Declare.
6227 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6229 * config/arm/arm.c (cmse_clear_registers): New function.
6230 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6231 code by call to cmse_clear_registers.
6232 (cmse_nonsecure_entry_clear_before_return): Likewise.
6234 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6236 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6237 (vdot_s32, vdotq_s32): New.
6238 (vdot_lane_u32, vdotq_lane_u32): New.
6239 (vdot_lane_s32, vdotq_lane_s32): New.
6242 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6245 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6246 for unsigned values with fewer HWIs than the precision.
6247 (test_overflow): New function.
6248 (wide_int_cc_tests): Call it.
6250 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6252 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6253 has the same mode class as Pmode.
6255 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6257 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6258 padding_bits_to_clear_ptr.
6259 (cmse_nonsecure_entry_clear_before_return): Likewise.
6261 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6263 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6264 auto_sbitap instead of integer bitfield to control register needing
6267 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6269 PR tree-optimization/83044
6270 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6271 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6272 that need the upper bound. Subtract offset from
6273 get_addr_base_and_unit_offset only if positive and subtract it
6274 before division by eltsize rather than after it.
6277 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6278 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6279 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6283 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6284 expand_binop with *mul_widen_optab, make sure at least one of the
6285 operands doesn't have VOIDmode.
6288 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6290 PR rtl-optimization/82044
6291 PR tree-optimization/82042
6292 * dse.c (record_store): Check for overflow.
6293 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6294 add_wild_read instead of clear_rhs_from_active_local_stores on
6295 overflow and log it into dump_file.
6297 2017-11-22 Richard Biener <rguenther@suse.de>
6299 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6300 fake edges to exit when looking for a place to insert.
6301 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6303 (insert_into_preds_of_block): Commit edge insertion immediately,
6304 assert that doesn't require new BBs.
6305 (fini_pre): Release expressions.
6306 (pass_pre::execute): Shuffle things around a bit, if the fn
6307 is too large do not compute AVAIL either as this is really the
6310 2017-11-22 Richard Biener <rguenther@suse.de>
6312 PR tree-optimization/83089
6313 * tree-if-conv.c (pass_if_conversion::execute): If anything
6314 changed reset SCEV and free the number of iteration estimates.
6316 2017-11-21 Martin Sebor <msebor@redhat.com>
6318 PR tree-optimization/82945
6319 * calls.h (warn_nonstring_bound): Remove unused function.
6321 2017-11-21 Martin Sebor <msebor@redhat.com>
6323 PR tree-optimization/82945
6324 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6325 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6326 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6328 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6329 * calls.h (get_attr_nonstring_decl): Declare new function.
6330 * doc/extend.texi (attribute nonstring): Update.
6331 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6332 get_attr_nonstring_decl and handle it.
6333 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6334 detection of nul-termination.
6335 (strlen_to_stridx): Change to a pointer.
6336 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6337 (pass_strlen::execute): Same.
6339 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6341 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6342 for the new option -mprefer-vector-width=[none|128|256|512].
6343 * config/i386/i386.c (ix86_target_string): remove old style options
6344 -mprefer-avx256 and make -mprefer-avx128 as alias.
6345 (ix86_option_override_internal): Apply defaults for the
6346 -mprefer-vector-width=[128|256] option.
6347 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6348 Implement macros to work with -mprefer-vector-width=.
6349 * config/i386/i386.opt: Implemented option
6350 -mprefer-vector-width=[none|128|256|512].
6351 * doc/invoke.texi: Documentation for
6352 -mprefer-vector-width=[none|128|256|512].
6354 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6356 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6358 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6360 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6362 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6364 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6366 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6367 * expr.c (expand_expr_real_2): Likewise.
6368 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6369 fold_binary_loc): Likewise.
6370 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6371 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6372 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6373 MINUS_EXPR transformations.
6374 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6375 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6376 * tree-inline.c (estimate_operator_cost): Likewise.
6377 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6378 op_symbol_code): Likewise.
6379 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6380 * vr-values.c (extract_range_from_binary_expr): Likewise.
6381 * varasm.c (initializer_constant_valid_p_1): Likewise.
6382 * tree.def: New tree code POINTER_DIFF_EXPR.
6384 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6386 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6387 integer suffix to movbe mnemonic.
6388 (*bswaphi2_movbe): Ditto.
6389 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6391 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6394 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6395 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6396 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6398 2017-11-21 Martin Liska <mliska@suse.cz>
6400 * tree-inline.c (expand_call_inline): Remove not needed
6403 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6404 Jakub Jelinek <jakub@redhat.com>
6407 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6408 Remove static keyword from f variable.
6410 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6412 PR tree-optimization/83086
6413 * gimple-ssa-store-merging.c
6414 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6415 rather than n.base_addr.
6417 2017-11-21 Martin Liska <mliska@suse.cz>
6419 PR rtl-optimization/82044
6420 PR tree-optimization/82042
6421 * dse.c (check_mem_read_rtx): Check for overflow.
6423 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6425 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6428 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6430 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6431 and its corresponding call arg location note.
6433 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6434 Andrew Burgess <andrew.burgess@embecosm.com>
6436 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6438 (arc_return_slot_offset): Likewise.
6439 (arc_eh_return_address_location): New declaration.
6440 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6441 (MUST_SAVE_REGISTER): Add exception handler case.
6442 (MUST_SAVE_RETURN_ADDR): Likewise.
6443 (arc_frame_pointer_required): Likewise.
6444 (arc_frame_pointer_needed): New function.
6445 (arc_compute_frame_size): Changed.
6446 (arc_expand_prologue): Likewise.
6447 (arc_expand_epilogue): Likewise.
6448 (arc_initial_elimination_offset): Likewise.
6449 (arc_return_slot_offset): Delete.
6450 (arc_eh_return_address_location): New function.
6451 (arc_builtin_setjmp_frame_value): Likewise.
6452 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6453 (EH_RETURN_STACKADJ_RTX): Define.
6454 (EH_RETURN_HANDLER_RTX): Likewise.
6455 * config/arc/arc.md (eh_return): Delete.
6457 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6459 * print-tree.h (debug_vec_tree): Remove prototype.
6460 * gdbinit.in (pvt): Remove macro.
6462 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6464 PR tree-optimization/83047
6465 * gimple-ssa-store-merging.c
6466 (imm_store_chain_info::output_merged_store): If the loads with the
6467 same vuse are in different basic blocks, for load_gsi pick a load
6468 location that is dominated by the other loads.
6471 * config/i386/i386.c (ix86_memmodel_check): Start
6472 -Winvalid-memory-model diagnostics with lowercase letter.
6475 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6476 set in_cold_section_p to the partition containing loc_list->first.
6477 When seeing loc_list->last_before_switch node, update secname and
6478 perform range_across_switch second partition handling only after that.
6481 * run-rtl-passes.c: Include debug.h.
6482 (run_rtl_passes): Call debug_hooks->assembly_start.
6483 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6487 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6488 OPTAB_DIRECT in calls to expand_simple_binop.
6490 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6493 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6495 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6498 * spellcheck.h (best_match::blithely_get_best_candidate): New
6501 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6503 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6504 with lower case letter.
6506 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6508 * config/i386/i386.md (bswaphi2): New expander.
6509 (*bswaphi2_movbe): New insn pattern.
6510 (bswaphi -> rorhi pepehole2): New peephole pattern.
6512 2017-11-20 Jeff Law <law@redhat.com>
6514 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6515 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6516 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6517 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6518 gimple-ssa-evrp-analyze.h.
6520 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6521 set BB_VISITED here.
6522 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6524 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6526 * tree-predcom.c: Add general comment on Store-Store chains.
6527 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6529 (get_chain_last_ref_at): Rename into...
6530 (get_chain_last_write_at): ...this.
6531 (get_chain_last_write_before_load): New function.
6532 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6533 CT_STORE_STORE when write reference is added.
6534 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6535 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6536 (initialize_root_vars_store_elim_1): Ditto.
6537 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6538 definition is created.
6539 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6540 chain by replacing it with dominant stored value.
6542 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6544 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6546 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6548 * vr-values.c (extract_range_from_binary_expr): Use a full range
6551 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6553 * config/arm/arm.md (R4_REGNUM): Define constant.
6554 (nonsecure_call_internal): Remove r4 clobber.
6555 (nonsecure_call_value_internal): Likewise.
6556 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6557 clobber and resequence match_operands.
6558 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6559 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6560 (nonsecure_call_value_reg_thumb2): Likewise.
6562 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6564 PR tree-optimization/78821
6565 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6566 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6567 to the address of the base rather than the base itself.
6568 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6569 (find_bswap_or_nop_finalize): New function.
6570 (find_bswap_or_nop): Use it.
6571 (bswap_replace): Return a tree rather than bool, change first
6572 argument from gimple * to gimple_stmt_iterator, allow inserting
6573 into an empty sequence, allow ins_stmt to be NULL - then emit
6574 all stmts into gsi. Fix up MEM_REF address gimplification.
6575 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6576 (struct store_immediate_info): Add N and INS_STMT non-static
6578 (store_immediate_info::store_immediate_info): Initialize them
6579 from newly added ctor args.
6580 (merged_store_group::apply_stores): Formatting fixes. Sort by
6582 (stmts_may_clobber_ref_p): For stores call also
6583 refs_anti_dependent_p.
6584 (gather_bswap_load_refs): New function.
6585 (imm_store_chain_info::try_coalesce_bswap): New method.
6586 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6587 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6588 (imm_store_chain_info::output_merged_store): Fail if number of
6589 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6590 and NOP_EXPR rhs_code.
6591 (pass_store_merging::process_store): Compute n and ins_stmt, if
6592 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6593 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6595 (pass_store_merging::execute): Calculate dominators.
6597 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6598 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6599 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6600 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6601 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6602 class pass_optimize_bswap, bswap_replace,
6603 pass_optimize_bswap::execute): Moved to ...
6604 * gimple-ssa-store-merging.c: ... this file.
6605 Include optabs-tree.h.
6606 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6607 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6608 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6609 anonymous namespace, remove static keywords.
6610 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6611 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6613 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6616 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6618 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6620 * vec.h (debug_helper): New function.
6621 (DEFINE_DEBUG_VEC): New macro.
6622 * hash-set.h (debug_helper): New function.
6623 (DEFINE_DEBUG_HASH_SET): New macro.
6624 * cfg.c (debug_slim (edge)): New function.
6625 Call DEFINE_DEBUG_VEC for edges.
6626 Call DEFINE_DEBUG_HASH_SET for edges.
6627 * cfghooks.c (debug_slim (basic_block)): New function.
6628 Call DEFINE_DEBUG_VEC for basic blocks.
6629 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6630 * print-tree.c (debug_slim): New function to handle trees.
6631 Call DEFINE_DEBUG_VEC for trees.
6632 Call DEFINE_DEBUG_HASH_SET for trees.
6633 (debug (vec<tree, va_gc>) &): Remove.
6634 (debug (<vec<tree, va_gc>) *): Remove.
6635 * print-rtl.c (debug_slim): New function to handle const_rtx.
6636 Call DEFINE_DEBUG_VEC for rtx_def.
6637 Call DEFINE_DEBUG_VEC for rtx_insn.
6638 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6639 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6640 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6641 (debug (vec<rtx_insn *> *ptr): Remove.
6642 (debug_insn_vector): Remove.
6643 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6645 2017-11-20 Tom de Vries <tom@codesourcery.com>
6647 PR rtl-optimization/82020
6648 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6649 IF_THEN_ELSE condition.
6651 2017-11-19 Jeff Law <law@redhat.com>
6653 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6654 of degenerates resulting from ignoring an edge.
6656 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6659 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6661 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6664 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6665 for uninitialied counts.
6667 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6670 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6672 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6675 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6677 2017-11-19 Tom de Vries <tom@codesourcery.com>
6679 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6682 2017-11-19 Tom de Vries <tom@codesourcery.com>
6684 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6687 2017-11-19 Tom de Vries <tom@codesourcery.com>
6689 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6692 2017-11-19 Tom de Vries <tom@codesourcery.com>
6694 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6696 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6698 2017-11-19 Tom de Vries <tom@codesourcery.com>
6700 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6702 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6703 semicolon after MCORE_EXPORT_NAME call.
6705 2017-11-19 Tom de Vries <tom@codesourcery.com>
6708 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6709 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6711 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6713 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6714 (cgraph_edge::clone): Cleanup updating of profile.
6715 * ipa-cp.c (update_profiling_info): Likewise.
6716 * ipa-inline-transform.c (inline_transform): Likewise.
6717 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6718 * ipa-split.c (execute_split_functions): Do not split when function
6720 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6721 * profile-count.c (profile_count::dump): Add global0.
6722 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6724 (profile_count::to_sreal_scale): Likewise.
6725 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6726 (profile_count::combine_with_ipa_count): New function.
6727 * profile-count.h (profile_guessed_global0adjusted): New.
6728 (profile_count::adjusted_zero): New.
6729 (profile_count::global0adjusted): New.
6730 (profile_count::combine_with_ipa_count): New.
6731 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6732 correct profile of return block of split functions.
6733 (copy_cfg_body): Remove unused profile_count.
6734 (copy_body): Likewise.
6735 (expand_call_inline): Update.
6736 (tree_function_versioning): Update.
6738 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6740 * hash-set.h (hash_set::empty): New.
6741 * tree-ssa-threadbackward.h: Delete.
6742 * tree-ssa-threadbackward.c (class thread_jumps): New.
6743 Move max_threaded_paths into class.
6744 (fsm_find_thread_path): Remove arguments that are now in class.
6745 (profitable_jump_thread_path): Rename to...
6746 (thread_jumps::profitable_jump_thread_path): ...this.
6747 (convert_and_register_jump_thread_path): Rename to...
6748 (thread_jumps::convert_and_register_current_path): ...this.
6749 (check_subpath_and_update_thread_path): Rename to...
6750 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6751 (register_jump_thread_path_if_profitable): Rename to...
6752 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6753 (handle_phi): Rename to...
6754 (thread_jumps::handle_phi): ...this.
6755 (handle_assignment): Rename to...
6756 (thread_jumps::handle_assignment): ...this.
6757 (fsm_find_control_statement_thread_paths): Rename to...
6758 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6759 (find_jump_threads_backwards): Rename to...
6760 (thread_jumps::find_jump_threads_backwards): ...this.
6761 Initialize path local data.
6762 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6763 from within thread_jumps class.
6764 (pass_early_thread_jumps::execute): Same.
6766 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6768 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6770 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6772 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6773 to canonical location.
6775 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6778 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6780 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6782 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6784 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6785 rcrt1.o%s/grcrt1.o%s for -static-pie.
6787 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6789 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6790 ix86_shift_rotate_cost): Break out from ...
6791 (ix86_rtx_costs): ... here.
6792 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6795 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6797 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6798 when entry block was promoted unlikely.
6799 (estimate_bb_frequencies): Increase frequency scale.
6800 * profile-count.h (profile_count): Export precision info.
6802 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6804 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6805 disturbin profile of entry block.
6807 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6809 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6812 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6814 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6815 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6817 2017-11-17 Jeff Law <law@redhat.com>
6819 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6820 from evrp_dom_walker class. Various methods moved into new class.
6821 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6822 (evrp_range_analyzer::enter): New method.
6823 (evrp_range_analyzer::leave): New method.
6824 (evrp_dom_walker): Remove delegators no longer needed by this class.
6825 Replace vr_values data member with evrp_range_analyzer
6827 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6828 method extracted from evrp_dom_walker::before_dom_children.
6829 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6830 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6832 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6833 Add private copy constructor and move assignment operators.
6834 Privatize methods and class data where trivially possible.
6835 (evrp_dom_walker::cleanup): New function, extracted from
6836 execute_early_vrp. Simplify access to class data.
6838 * vr-values.h (get_output_for_vrp): Prototype.
6839 * vr-values.c (get_output_for_vrp): New function extracted from
6840 vrp_visit_assignment_or_call and extract_range_from_stmt.
6841 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6843 2017-11-17 Luis Machado <luis.machado@linaro.org>
6845 * config/aarch64/aarch64.c
6846 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6847 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6848 tune_params::AUTOPREFETCHER_WEAK.
6850 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6853 * config/arm/arm.c (arm_valid_target_attribute_rec):
6854 Parse "arch=" and "+<ext>".
6855 (arm_valid_target_attribute_tree): Re-init global options.
6856 (arm_option_override): Make non-static.
6857 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6858 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6859 (__ARM_FEATURE_CRC32): Support undef.
6860 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6861 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6863 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6865 * gdbinit.in (break-on-diagnostic): New command.
6867 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6869 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6870 return to indirect jump for EH return if control-flow
6871 protection is enabled. Change explicit 'false' argument in
6872 pro_epilogue_adjust_stack with a value of
6874 * config/i386/i386.md (simple_return_indirect_internal):
6875 Remove SImode restriction to support 64-bit.
6877 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6879 * combine.c (added_notes_insn): New.
6880 (try_combine): Handle added_notes_insn like added_links_insn.
6881 Rewrite return value code.
6882 (distribute_notes): Set added_notes_insn to the earliest insn we added
6885 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6887 PR rtl-optimization/82621
6888 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6889 dest of one of those SETs is unused.
6891 2017-11-17 Richard Biener <rguenther@suse.de>
6894 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6895 * tree-pretty-print.c (dump_generic_node): Handle
6896 annot_expr_parallel_kind.
6897 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6898 * gimplify.c (gimple_boolify): Likewise.
6900 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6902 * config.gcc (extra_headers): Add cet.h for x86 targets.
6903 * config/i386/cet.h: New file.
6904 * doc/install.texi: Add --enable-cet/--disable-cet.
6906 2017-11-17 Richard Biener <rguenther@suse.de>
6908 PR tree-optimization/83017
6909 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6910 (gen_parallel_loop): Properly count iterations.
6911 (parallelize_loops): Handle loop->can_be_parallel independent
6912 of flag_loop_parallelize_all. Make static profitability test match
6914 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6915 * invoke.texi (parloops-min-per-thread): Document.
6917 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6919 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6920 upstreaming review comments.
6922 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6924 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6925 with fast unaligned access.
6926 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6928 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6930 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6931 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6933 2017-11-17 Richard Biener <rguenther@suse.de>
6935 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6936 folding of references.
6938 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6941 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6942 of replacing call to strncmp with corresponding call to strcmp when
6945 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6947 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6948 option prefer-avx256 for skylake-avx512 configuration.
6949 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6950 (get_builtin_code_for_version): Ditto.
6952 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6953 Monk Chiang <sh.chiang04@gmail.com>
6955 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6956 (FIXED_REGISTERS): Reserve more register numbers.
6957 (CALL_USED_REGISTERS): Likewise.
6958 (REG_ALLOC_ORDER): Likewise.
6959 (REG_CLASS_CONTENTS): Likewise.
6960 (REGISTER_NAMES): Likewise.
6962 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6963 Kito Cheng <kito.cheng@gmail.com>
6965 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6967 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6969 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6971 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6972 parameter list for vec_splats.
6974 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6976 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6978 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6980 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6983 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6984 with the required version of automake 1.11.6.
6986 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6988 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6990 (DU_C2_3_power9): Correct reservation combinations.
6991 (FP_DIV_power9, VEC_DIV_power9): New.
6992 (power9-alu): Split out rotate/shift...
6993 (power9-rot): ...to here, correct dispatch resource.
6994 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6996 (power9-fp): Correct latency.
6997 (power9-sdiv): Add div/sqrt resource.
6998 (power9-ddiv): Correct latency, add div/sqrt resource.
6999 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
7000 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
7002 (power9-qpdiv, power9-qpmul): Adjust resource usage.
7004 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
7006 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7007 switch statement mapping KF built-ins to TF built-ins if we don't
7008 have the proper ISA 3.0 assembler support.
7010 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7012 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7013 (lower_emutls_function_body): Do not compute it.
7015 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7017 * ipa-split.c (split_bb_info): Turn time to sreal.
7018 (split_point): Likewise.
7019 (dump_split_point): Likewise.
7020 (fine_split_points): Likewise.
7021 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7023 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7025 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7028 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7030 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7032 (expensive_function_p): Remove useless assert.
7033 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7035 2017-11-16 Martin Sebor <msebor@redhat.com>
7037 PR tree-optimization/82588
7038 PR tree-optimization/82583
7039 * tree-vrp.c (check_array_ref): Handle flexible array members,
7040 string literals, and inner indices.
7041 (search_for_addr_array): Add detail to diagnostics.
7043 2017-11-16 Nathan Sidwell <nathan@acm.org>
7047 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7048 (SET_DECL_ASSEMBLER_NAME): Forward to
7049 overwrite_decl_assembler_name.
7050 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7051 (overwrite_decl_assembler_name): Declare.
7052 * tree.c (overwrite_decl_assembler_name): New.
7053 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7054 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7055 (LANG_HOOKS_INITIALIZER): Add it.
7056 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7057 * langhooks.c (lhd_set_decl_assembler_name): Use
7058 SET_DECL_ASSEMBLER_NAME.
7059 (lhd_overwrite_decl_assembler_name): Default implementation.
7061 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7062 Jackson Woodruff <jackson.woodruff@arm.com>
7064 PR tree-optimization/71026
7065 * match.pd: Canonicalize constant multiplies in division.
7067 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7069 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7070 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7071 Remove OPT_fomit_frame_pointer entry.
7072 * common/config/arc/arc-common.c: Likewise.
7073 * common/config/arm/arm-common.c: Likewise.
7074 * common/config/avr/avr-common.c: Likewise.
7075 * common/config/c6x/c6x-common.c: Likewise.
7076 * common/config/cr16/cr16-common.c: Likewise.
7077 * common/config/cris/cris-common.c: Likewise.
7078 * common/config/epiphany/epiphany-common.c: Likewise.
7079 * common/config/fr30/fr30-common.c: Likewise.
7080 * common/config/frv/frv-common.c: Likewise.
7081 * common/config/ia64/ia64-common.c: Likewise.
7082 * common/config/iq2000/iq2000-common.c: Likewise.
7083 * common/config/lm32/lm32-common.c: Likewise.
7084 * common/config/m32r/m32r-common.c: Likewise.
7085 * common/config/mcore/mcore-common.c: Likewise.
7086 * common/config/microblaze/microblaze-common.c: Likewise.
7087 * common/config/mips/mips-common.c: Likewise.
7088 * common/config/mmix/mmix-common.c: Likewise.
7089 * common/config/mn10300/mn10300-common.c: Likewise.
7090 * common/config/nios2/nios2-common.c: Likewise.
7091 * common/config/pa/pa-common.c: Likewise.
7092 * common/config/pdp11/pdp11-common.c: Likewise.
7093 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7094 * common/config/riscv/riscv-common.c: Likewise.
7095 * common/config/rs6000/rs6000-common.c: Likewise.
7096 * common/config/rx/rx-common.c: Likewise.
7097 * common/config/s390/s390-common.c: Likewise.
7098 * common/config/sh/sh-common.c: Likewise.
7099 * common/config/sparc/sparc-common.c: Likewise.
7100 * common/config/tilegx/tilegx-common.c: Likewise.
7101 * common/config/tilepro/tilepro-common.c: Likewise.
7102 * common/config/v850/v850-common.c: Likewise.
7103 * common/config/visium/visium-common.c: Likewise.
7104 * common/config/xstormy16/xstormy16-common.c: Likewise.
7105 * common/config/xtensa/xtensa-common.c: Likewise.
7106 * invoke.texi (-fomit-frame-pointer): Update documentation.
7108 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7110 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7112 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7114 * cfg.c (scale_bbs_frequencies_int,
7115 cale_bbs_frequencies_gcov_type): Remove.
7116 * cfg.h (scale_bbs_frequencies_int,
7117 cale_bbs_frequencies_gcov_type): Remove.
7119 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7121 * tree-ssa-loop-manip.c
7122 (scale_dominated_blocks_in_loop): Update to profile counts.
7123 (tree_transform_and_unroll_loop): Likewise.
7125 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7127 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7128 scale_bbs_frequencies_int.
7130 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7132 * final.c (compute_alignments): Use counts rather than frequencies.
7134 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7136 * cfgloopanal.c: Include sreal.h
7137 (average_num_loop_insns): Use counts and sreal for accounting.
7139 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7141 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7144 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7145 Kito Cheng <kito.cheng@gmail.com>
7147 * config/nds32/constraints.md: Provide more constraints.
7148 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7149 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7150 support constraints usage.
7152 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7154 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7156 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7157 Kito Cheng <kito.cheng@gmail.com>
7159 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7160 * config/nds32/nds32.opt: Refine the layout.
7161 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7162 TARGET_EXT_STRING): Support new options.
7163 * config/nds32/nds32.h: Likewise.
7164 * config/nds32/nds32.md: Likewise.
7165 * config/nds32/nds32-predicates.c: Likewise.
7166 * config/nds32/constraints.md: Likewise.
7167 * common/config/nds32/nds32-common.c: Likewise.
7169 2017-11-16 Julia Koval <julia.koval@intel.com>
7172 * config/i386/gfniintrin.h: Add sse check.
7173 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7175 2017-11-16 Julia Koval <julia.koval@intel.com>
7177 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7178 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7179 (ix86_handle_option): Handle -mavx512vbmi2.
7180 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7181 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7182 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7183 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7184 (ix86_valid_target_attribute_inner_p): Ditto.
7185 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7186 * config/i386/i386.opt (mavx512vbmi2): New option.
7187 * doc/invoke.texi: Add new option.
7189 2017-11-16 Julia Koval <julia.koval@intel.com>
7191 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7192 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7193 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7194 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7195 _mm512_gf2p8mul_epi8): New intrinsics.
7196 * config/i386/i386-builtin-types.def
7197 (V64QI_FTYPE_V64QI_V64QI): New type.
7198 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7199 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7200 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7201 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7202 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7203 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7205 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7207 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7208 explicitly as a stream of bytes.
7210 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7212 * config/rs6000/altivec.h (vec_xst_be): New #define.
7213 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7214 and externalize from *altivec_vperm_<mode>_internal.
7215 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7217 (XL_BE_V8HI): Likewise.
7218 (XL_BE_V4SI): Likewise.
7219 (XL_BE_V4SI): Likewise.
7220 (XL_BE_V2DI): Likewise.
7221 (XL_BE_V4SF): Likewise.
7222 (XL_BE_V2DF): Likewise.
7223 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7224 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7225 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7226 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7227 VSX_BUILTIN_VEC_XST_BE.
7228 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7229 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7231 (altivec_init_builtins): Replace conditional calls to def_builtin
7232 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7233 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7234 with unconditional calls. Remove calls to def_builtin for
7235 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7236 __builtin_vec_xst_be.
7237 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7238 to define_expand, and add alternate RTL generation for P8.
7239 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7240 vsx_ld_elemrev_v8hi.
7241 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7242 add alternate RTL generation for P8.
7243 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7244 vsx_ld_elemrev_v16qi.
7245 (vsx_st_elemrev_v8hi): Convert define_insn
7246 to define_expand, and add alternate RTL generation for P8.
7247 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7248 vsx_st_elemrev_v8hi.
7249 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7250 add alternate RTL generation for P8.
7251 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7252 vsx_st_elemrev_v16qi.
7254 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7257 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7258 TARGET_AVX512ER check.
7259 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7260 neither -mzeroupper nor -mno-zeroupper is used and
7261 TARGET_EMIT_VZEROUPPER is set.
7262 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7263 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7265 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7267 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7268 folding of vector compares.
7269 (fold_build_vec_cmp): New helper function.
7270 (fold_compare_helper): New helper function.
7271 (builtin_function_type): Add compare builtins to the list of functions
7272 having unsigned arguments. Cosmetic updates to comment indentation.
7273 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7274 the not+eq combination.
7276 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7278 PR tree-optimization/82726
7279 PR tree-optimization/70754
7280 * tree-predcom.c (order_drefs_by_pos): New function.
7281 (combine_chains): Move code setting has_max_use_after to...
7282 (try_combine_chains): ...here. New parameter. Sort combined chains
7283 according to position information.
7284 (tree_predictive_commoning_loop): Update call to above function.
7285 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7287 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7289 PR tree-optimization/82726
7291 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7293 PR tree-optimization/70754
7294 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7295 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7296 combined stmt before it if not NULL.
7297 (combine_chains): Process refs reversely and compute dominance point
7301 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7303 PR tree-optimization/79663
7304 * tree-predcom.c (combine_chains): Process refs in reverse order
7305 only for ZERO length chains, and add explaining comment.
7307 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7309 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7310 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7311 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7313 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7315 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7317 2017-11-15 Martin Liska <mliska@suse.cz>
7319 * tree-cfg.c (pass_warn_function_return::execute):
7320 Compare warn_return_type for greater than zero.
7322 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7326 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7327 to return true on Xeon and not on Xeon Phi.
7328 (ix86_check_avx256_register): Changed to ...
7329 (ix86_check_avx_upper_register): ... this. Add extra check for
7330 VALID_AVX512F_REG_OR_XI_MODE.
7331 (ix86_avx_u128_mode_needed): Changed
7332 ix86_check_avx256_register to ix86_check_avx_upper_register.
7333 (ix86_check_avx256_stores): Changed to ...
7334 (ix86_check_avx_upper_stores): ... this. Changed
7335 ix86_check_avx256_register to ix86_check_avx_upper_register.
7336 (ix86_avx_u128_mode_after): Changed
7337 avx_reg256_found to avx_upper_reg_found. Changed
7338 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7339 (ix86_avx_u128_mode_entry): Changed
7340 ix86_check_avx256_register to ix86_check_avx_upper_register.
7341 (ix86_avx_u128_mode_exit): Ditto.
7342 * config/i386/i386.h: (host_detect_local_cpu): New define.
7344 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7346 * config/arm/xgene1.md (xgene1): Split into automatons
7347 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7348 (xgene1_f_load): Adjust reservations and/or types.
7349 (xgene1_f_store): Likewise.
7350 (xgene1_load_pair): Likewise.
7351 (xgene1_store_pair): Likewise.
7352 (xgene1_fp_load1): Likewise.
7353 (xgene1_load1): Likewise.
7354 (xgene1_store1): Likewise.
7355 (xgene1_move): Likewise.
7356 (xgene1_alu): Likewise.
7357 (xgene1_simd): Likewise.
7358 (xgene1_bfm): Likewise.
7359 (xgene1_neon_load1): Likewise.
7360 (xgene1_neon_store1): Likewise.
7361 (xgene1_neon_logic): Likewise.
7362 (xgene1_neon_st1): Likewise.
7363 (xgene1_neon_ld1r): Likewise.
7364 (xgene1_alu_cond): Added.
7365 (xgene1_shift_reg): Likwise.
7366 (xgene1_bfx): Likewise.
7367 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7369 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7372 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7374 (can_widen_mult_without_libcall): New function.
7375 (expand_mul_overflow): If only checking unsigned mul overflow,
7376 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7377 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7378 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7379 (expand_DIVMOD): Formatting fix.
7380 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7381 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7382 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7384 PR tree-optimization/82977
7385 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7386 constructed temporary to strlen_to_stridx.put.
7388 2017-11-15 Martin Liska <mliska@suse.cz>
7390 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7391 * configure: Regenerate.
7393 2017-11-15 Martin Liska <mliska@suse.cz>
7396 * config/sh/sh-mem.cc: Use proper probability for
7399 2017-11-14 Jeff Law <law@redhat.com>
7401 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7402 the red zone for stack_clash_protection_final_dynamic_probe targets
7403 when the total dynamic stack size is zero bytes.
7405 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7406 blocks is post order.
7408 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7410 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7411 * final.c (rest_of_clean_state): Set it for the
7412 -fcompare-debug dump.
7413 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7414 class when TDF_COMPARE_DEBUG is set.
7416 * dwarf2out.c (gen_producer_string): Discard
7419 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7422 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7423 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7424 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7425 (__TGMATH_REAL_2_3): Remove macros.
7426 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7427 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7429 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7431 (remquo): Define using __TGMATH_REAL_3.
7433 2017-11-14 Jeff Law <law@redhat.com>
7435 * vr-values.c: New file with contents extracted from tree-vrp.c.
7436 * Makefile.in (OBJS): Add vr-values.o
7437 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7438 (set_value_range, set_and_canonicalize_value_range): Likewise.
7439 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7440 (value_range_constant_singleton, symbolic_range_p): Likewise.
7441 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7442 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7443 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7444 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7445 (find_case_label_range, find_case_label_index): Likewise.
7446 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7447 (range_int_cst_singleton_p, value_inside_range): Likewise.
7448 (get_single_symbol): Likewise.
7449 (switch_update): Move structure definition here.
7450 (to_remove_edges, to_update_switch_stmts): Provide externs.
7451 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7452 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7453 (vrp_val_is_min, set_value_range): Likewise.
7454 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7455 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7456 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7457 (range_is_nonnull, range_int_cst_p): Likewwise.
7458 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7459 (get_single_symbol, operand_less_p): Likewise
7460 (compare_values_warnv, compare_values): Likewise.
7461 (value_inside_range, value_range_constant_singleton): Likewise.
7462 (zero_nonzero_bitgs_from_vr): Likewise.
7463 (extract_range_from_binary_expr_1): Likewise.
7464 (overflow_comparison_p): Likewise.
7465 (to_remove_edges, to_update_switch_stmts): Likewise.
7466 (find_case_label-index, find_case_label_range): Likewise.
7467 (switch_update, set_value_range_to_nonnegative): Remove.
7468 (set_value_range_to_truthvalue): Likewise.
7469 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7470 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7471 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7472 (find_case_label_ranges, test_for_singularity): Likewise.
7473 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7474 (x_vr_values): Move to its remaining use site.
7476 2017-11-10 Jeff Law <law@redhat.com>
7478 * vr-values.h (VR_INITIALIZER): Move #define here.
7479 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7480 * Makefile.in (OBJS): Add tree-evrp.o
7481 * tree-vrp.h (assert_info): Move structure definition here.
7482 (set_value_range_to_varying): Prototype.
7483 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7484 (infer_value_range, register_edge_assert_for): Likewise.
7485 (stmt_interesting_for_vrp): Likewise.
7486 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7487 (set_value_range_to_varying): No longer static.
7488 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7489 (infer_value_range, register_edge_assert_for): Likewise.
7491 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7493 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7494 generate the XXBRD instruction.
7496 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7497 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7498 as being compatible if -mabi=ieeelongdouble.
7499 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7500 to setup float128 built-ins with hardware support.
7501 (BU_FLOAT128_HW_2): Likewise.
7502 (BU_FLOAT128_HW_3): Likewise.
7503 (BU_FLOAT128_HW_VSX_1): Likewise.
7504 (BU_FLOAT128_HW_VSX_2): Likewise.
7505 (scalar_extract_expq): Change float128 built-in functions to
7506 accommodate having both KFmode and TFmode functions. Use the
7507 KFmode variant as the default.
7508 (scalar_extract_sigq): Likewise.
7509 (scalar_test_neg_qp): Likewise.
7510 (scalar_insert_exp_q): Likewise.
7511 (scalar_insert_exp_qp): Likewise.
7512 (scalar_test_data_class_qp): Likewise.
7513 (sqrtf128_round_to_odd): Delete processing the round to odd
7514 built-in functions as special built-in functions, and define them
7515 as float128 built-ins. Use the KFmode variant as the default.
7516 (truncf128_round_to_odd): Likewise.
7517 (addf128_round_to_odd): Likewise.
7518 (subf128_round_to_odd): Likewise.
7519 (mulf128_round_to_odd): Likewise.
7520 (divf128_round_to_odd): Likewise.
7521 (fmaf128_round_to_odd): Likewise.
7522 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7523 support for KFmode and TFmode xststdcqp calls.
7524 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7525 point, switch the built-in handlers for the get/set float128
7526 exponent, get float128 mantissa, float128 test built-ins, and the
7527 float128 round to odd built-in functions. Eliminate creating the
7528 float128 round to odd built-in functions as special built-ins.
7529 (rs6000_init_builtins): Eliminate special creation of the float128
7530 round to odd built-in functions.
7531 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7532 function insns to support both TFmode and KFmode variants.
7533 (xsxsigqp_<mode>): Likewise.
7534 (xsiexpqpf_<mode>): Likewise.
7535 (xsiexpqp_<mode>): Likewise.
7536 (xststdcqp_<mode>): Likewise.
7537 (xststdcnegqp_<mode>): Likewise.
7538 (xststdcqp_<mode>): Likewise.
7540 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7542 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7543 unused path_in_freq_ptr parameter.
7544 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7546 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7548 * ipa-inline.c (edge_badness): Dump sreal frequency.
7549 (compute_inlined_call_time): Match natural implementaiton ...
7550 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7551 forgotten division by CGRAPH_FREQ_BASE.
7553 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7555 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7556 Solaris 11. Update comment.
7557 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7559 * config/sol2.h (STARTFILE_SPEC): Likewise.
7560 * configure: Regenerate.
7562 2017-11-14 Carl Love <cel@us.ibm.com>
7564 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7565 le_ and be_ prefixes to swap* variables. Remove
7566 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7569 2017-11-14 Jason Merrill <jason@redhat.com>
7571 Support GTY((cache)) on hash_map.
7572 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7573 (ggc_cache_remove): Override it instead of ggc_mx.
7574 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7575 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7576 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7577 (simple_cache_map_traits): Override maybe_mx.
7578 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7579 (hash_map): Friend gt_cleare_cache.
7580 (gt_cleare_cache): New.
7581 * tree.h (tree_cache_traits): New hash_map traits class.
7582 (tree_cache_map): New typedef.
7584 2017-11-14 Richard Biener <rguenther@suse.de>
7586 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7587 paramter and handling.
7588 (cleanup_control_flow_bb): Likewise.
7589 (cleanup_control_flow_pre): New helper performing a DFS walk
7590 to call cleanup_control_flow_bb in PRE order.
7591 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7592 via cleanup_control_flow_pre.
7594 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7596 * config/aarch64/aarch64-simd.md
7597 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7598 (*aarch64_simd_bsl<mode>_alt): Likewise.
7599 (aarch64_simd_bsldi_internal): New.
7600 (aarch64_simd_bsldi_alt): Likewise.
7602 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7604 * tracer.c (better_p): Do not compare frequencies.
7605 * reg-stack.c (better_edge): Likewise.
7606 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7609 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7611 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7612 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7613 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7614 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7615 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7616 * predict.c (maybe_hot_frequency_p): Inline to ...
7617 (maybe_hot_count_p): ... here; rewrite to counts.
7618 (counts_to_freqs): Rename to ...
7619 (update_max_bb_count): ... this one.
7620 (expensive_function_p): Use counts.
7621 (estimate_bb_frequencies): Update.
7622 (rebuild_frequencies): Update.
7623 * predict.h (counts_to_freqs): Rename to ...
7624 (update_max_bb_count): ... this one.
7625 * profile.c (compute_branch_probabilities): Add debug info
7626 * tree-inline.c (expand_call_inline): Update debug info.
7627 (optimize_inline_calls): Use update_max_bb_count..
7628 (tree_function_versioning): Use update_max_bb_count..
7629 * value-prof.c (gimple_value_profile_transformations):
7630 Do not use update_max_bb_count.
7632 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7634 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7635 always use frequencies.
7637 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7639 * bb-reorder.c: Remove frequencies from comments.
7640 (better_edge_p): Use profile counts.
7641 (find_traces): Dump profile counts.
7642 (rotate_loop): Use profile counts.
7643 (find_traces_1_round): Likewise.
7644 (connect_better_edge_p): Use counts instead of probabilities for
7646 (copy_bb_p): Drop early check for non-0 frequency.
7647 (sanitize_hot_paths): Update comments.
7649 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7651 * ipa-split.c (struct split_point): Add count.
7652 (consider_split): Do not compute incoming frequency; compute incoming
7653 count and store it to split_point.
7654 (split_function): Set count of the call to split part correctly.
7656 2017-11-13 Carl Love <cel@us.ibm.com>
7658 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7660 2017-11-13 Tom Tromey <tom@tromey.com>
7662 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7664 2017-11-13 Carl Love <cel@us.ibm.com>
7666 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7667 Add support for builtins:
7668 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7669 vector {un,}signed {char,int,short},
7670 vector {un,}signed {char,int,short}) arguments.
7671 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7672 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7673 Add BU_P9V_AV_2 expansions for the builtins.
7674 * config/rs6000/altivec.h (vec_first_match_index,
7675 vec_first_mismatch_index, vec_first_match_or_eos_index,
7676 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7677 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7678 new extern declaration.
7679 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7680 * config/rs6000/vsx.md (first_match_index_<mode>,
7681 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7682 first_mismatch_or_eos_index_<mode>): Add define expand.
7683 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7684 * doc/extend.texi: Update the built-in documenation file for the new
7687 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7689 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7690 into the min/max operations for _Float<N> and _Float<N>X types.
7692 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7695 * dwarf2out.c (do_eh_frame): New static variable.
7696 (dwarf2out_begin_prologue): Set it.
7697 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7699 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7701 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7703 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7705 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7708 * debug.h (dwarf2out_do_eh_frame): Declare.
7709 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7710 (dwarf2out_do_frame): Use it.
7711 (dwarf2out_do_cfi_asm): Likewise.
7712 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7713 (dwarf2out_assembly_start): Likewise.
7714 (dwarf2out_begin_prologue): Fix comment.
7715 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7716 if the target needs either debug or unwind DWARF2 info.
7717 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7718 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7719 -ftrapping-math, -ftrapv and -fwrapv.
7721 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7723 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7724 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7725 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7726 (estimate_edge_size_and_time): Likewise.
7727 (ipa_merge_fn_summary_after_inlining): Likewise.
7728 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7729 (compute_uninlined_call_time): Use sreal_frequency.
7730 (compute_inlined_call_time): Likewise.
7731 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7732 * profile-count.c: Include sreal.h.
7733 (profile_count::to_sreal_scale): New.
7734 * profile-count.h: Forward declare sreal.
7735 (profile_count::to_sreal_scale): Declare.
7737 2017-11-13 Nathan Sidwell <nathan@acm.org>
7739 * diagnostic.c (maybe_line_and_column): New.
7740 (diagnostic_get_location_text): Use it.
7741 (diagnostic_report_current_module): Likewise.
7742 (test_diagnostic_get_location_text): Add tests.
7744 2017-11-13 Luis Machado <luis.machado@linaro.org>
7746 * doc/md.texi (Specifying processor pipeline description): Fix
7747 incorrect latency for the div instruction example.
7749 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7751 PR tree-optimization/78821
7752 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7753 that bit_not_p is the same.
7754 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7755 (split_group): Count precisely bit_not_p bits in each statement.
7756 (invert_op): New function.
7757 (imm_store_chain_info::output_merged_store): Use invert_op to
7758 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7759 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7761 2017-11-13 Martin Liska <mliska@suse.cz>
7763 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7764 (struct source_info): Likewise.
7765 (add_branch_counts): Likewise.
7766 (add_line_counts): Likewise.
7767 (function_summary): Likewise.
7768 (output_intermediate_line): Likewise.
7769 (generate_results): Likewise.
7771 2017-11-13 Martin Liska <mliska@suse.cz>
7773 * gcov.c (struct block_info): Remove typedef for block_t.
7774 (struct line_info): Likewise.
7775 (line_info::has_block): Likewise.
7776 (EXIT_BLOCK): Likewise.
7777 (unblock): Likewise.
7778 (circuit): Likewise.
7779 (get_cycles_count): Likewise.
7780 (process_file): Likewise.
7781 (read_graph_file): Likewise.
7782 (solve_flow_graph): Likewise.
7783 (find_exception_blocks): Likewise.
7784 (add_line_counts): Likewise.
7785 (accumulate_line_info): Likewise.
7786 (output_line_details): Likewise.
7788 2017-11-13 Martin Liska <mliska@suse.cz>
7790 * gcov.c (struct arc_info): Remove typedef for arc_t.
7791 (struct line_info): Likewise.
7792 (add_branch_counts): Likewise.
7793 (output_branch_count): Likewise.
7794 (function_info::~function_info): Likewise.
7795 (circuit): Likewise.
7796 (output_intermediate_line): Likewise.
7797 (read_graph_file): Likewise.
7798 (solve_flow_graph): Likewise.
7799 (find_exception_blocks): Likewise.
7800 (add_line_counts): Likewise.
7801 (accumulate_line_info): Likewise.
7802 (output_line_details): Likewise.
7803 (output_function_details): Likewise.
7805 2017-11-13 Martin Liska <mliska@suse.cz>
7807 * gcov.c (struct function_info): Remove typedef for function_t.
7808 (struct source_info): Likewise.
7809 (source_info::get_functions_at_location): Likewise.
7810 (solve_flow_graph): Likewise.
7811 (find_exception_blocks): Likewise.
7812 (add_line_counts): Likewise.
7813 (output_intermediate_file): Likewise.
7814 (process_file): Likewise.
7815 (generate_results): Likewise.
7816 (release_structures): Likewise.
7817 (read_graph_file): Likewise.
7818 (read_count_file): Likewise.
7819 (accumulate_line_counts): Likewise.
7820 (output_lines): Likewise.
7822 2017-11-13 Martin Liska <mliska@suse.cz>
7824 * gcov.c (function_info::function_info): Remove num_counts
7825 and add vector<gcov_type>.
7826 (function_info::~function_info): Use the vector.
7827 (process_file): Likewise.
7828 (read_graph_file): Likewise.
7829 (read_count_file): Likewise.
7830 (solve_flow_graph): Likewise.
7832 2017-11-13 Martin Liska <mliska@suse.cz>
7834 * gcov.c (function_info::is_artificial): New function.
7835 (process_file): Erase all artificial early.
7836 (generate_results): Skip as all artificial are already
7839 2017-11-13 Martin Liska <mliska@suse.cz>
7841 * gcov.c (read_graph_file): Store to global vector of functions.
7842 (read_count_file): Iterate the vector.
7843 (process_file): Likewise.
7844 (generate_results): Likewise.
7845 (release_structures): Likewise.
7847 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7849 PR tree-optimization/82954
7850 * gimple-ssa-store-merging.c
7851 (imm_store_chain_info::coalesce_immediate_stores): If
7852 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7854 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7856 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7857 Upddate call to ENDIAN_LANE_N.
7858 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7859 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7860 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7861 and use aarch64_endian_lane_rtx.
7862 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7864 2017-11-12 Tom de Vries <tom@codesourcery.com>
7866 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7868 2017-11-12 Tom de Vries <tom@codesourcery.com>
7870 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7872 2017-11-12 Tom de Vries <tom@codesourcery.com>
7874 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7876 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7877 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7878 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7880 2017-11-11 Martin Sebor <msebor@redhat.com>
7883 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7886 * prefic.c (translate_name): Replace strncpy with memcpy to
7887 avoid -Wstringop-truncation.
7889 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7891 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7893 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7895 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7897 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7899 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7902 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7904 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7905 enable generating XXBRH if the value is in a vector register.
7906 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7907 value is in a vector register.
7908 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7909 register to register bswap64's instead of doing the GPR sequence
7910 used on previous machines.
7911 (bswapdi2_xxbrd): New insn.
7912 (bswapdi2_reg): Disallow on ISA 3.0.
7913 (register to register bswap64 splitter): Do not split the insn on
7914 ISA 3.0 systems that use XXBRD.
7916 2017-11-10 Martin Sebor <msebor@redhat.com>
7919 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7921 (find_subframework_file): Same.
7923 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7925 * auto-profile.c (afdo_indirect_call): Drop frequency.
7926 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7927 (cgraph_node::create_edge): Drop frequency argument.
7928 (cgraph_node::create_indirect_edge): Drop frequency argument.
7929 (cgraph_edge::make_speculative): Drop frequency arguments.
7930 (cgraph_edge::resolve_speculation): Do not update frequencies
7931 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7932 (cgraph_node::dump): Check consistency in IPA mode.
7933 (cgraph_edge::maybe_hot_p): Use IPA counter.
7934 (cgraph_edge::verify_count_and_frequency): Rename to ...
7935 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7936 (cgraph_node::verify_node): Update.
7937 * cgraph.h (struct cgraph_edge): Drop frequency.
7938 (cgraph_edge::frequency): New function.
7939 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7941 (cgraph_edge::rebuild_edges): Likewise.
7942 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7943 (duplicate_thunk_for_node): Do not pass frequency.
7944 (cgraph_node::create_clone): Scale only counts.
7945 (cgraph_node::create_virtual_clone): Do not pass frequency.
7946 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7947 (cgraph_node::create_version_clone): Do not pass frequency.
7948 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7949 (cgraph_node::expand_thunk): Do not pass frequency.
7950 (cgraph_node::create_wrapper): Do not pass frequency.
7951 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7953 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7954 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7955 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7956 (gather_caller_stats): Use frequency function.
7957 (ipcp_cloning_candidate_p): Use frequency function.
7958 (ipcp_propagate_stage): Use frequency function.
7959 (get_info_about_necessary_edges): Use frequency function.
7960 (update_profiling_info): Update only IPA profile.
7961 (update_specialized_profile): Use frequency functoin.
7962 (perhaps_add_new_callers): Update only IPA profile.
7963 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7964 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7965 (dump_ipa_call_summary): Use frequency function.
7966 (estimate_edge_size_and_time): Use frequency function.
7967 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7968 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7969 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7970 (update_noncloned_counts): ... ths one; scale counts only.
7971 (clone_inlined_nodes): Do not scale frequency.
7972 (inline_call): Do not pass frequency.
7973 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7974 (compute_inlined_call_time): Use IPA profile.
7975 (want_inline_small_function_p): Use IPA profile.
7976 (want_inline_self_recursive_call_p): Use IPA profile.
7977 (edge_badness): Use IPA profile.
7978 (lookup_recursive_calls): Use IPA profile.
7979 (recursive_inlining): Do not pass frequency.
7980 (resolve_noninline_speculation): Do not update frequency.
7981 (inline_small_functions): Collect max of IPA profile.
7982 (dump_overall_stats): Dump IPA porfile.
7983 (dump_inline_stats): Dump IPA porfile.
7984 (ipa_inline): Collect IPA stats.
7985 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7986 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7987 (ipa_propagate_frequency): Use frequency function.
7988 (ipa_profile): Cleanup.
7989 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7990 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7991 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7992 (input_node): Do not stream frequency.
7993 (input_edge): Do not stream frequency.
7994 (merge_profile_summaries): Scale only IPA profiles.
7995 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7996 * predict.c (drop_profile): Do not recompute frequency.
7997 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7998 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7999 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
8000 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
8001 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
8002 * tree-inline.c (copy_bb): Do not scale frequency.
8003 (expand_call_inline): Do not scale frequency.
8004 (tree_function_versioning): Do not scale frequency.
8005 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8007 2017-11-10 Julia Koval <julia.koval@intel.com>
8009 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8010 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8011 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8012 (_mm256_maskz_gf2p8affine_epi64_epi8)
8013 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8014 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8015 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8016 (__builtin_ia32_vgf2p8affineqb_v32qi)
8017 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8018 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8020 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8024 (arm_option_override): Refactor.
8025 (arm_option_reconfigure_globals): New.
8026 (arm_options_perform_arch_sanity_checks): New.
8027 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8029 (arm_options_perform_arch_sanity_checks): Likewise
8031 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8033 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8034 (power9-qpmul): New.
8035 * rs6000/rs6000.md ("type" attr): Add qmul.
8036 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8037 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8038 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8040 2017-11-10 Martin Sebor <msebor@redhat.com>
8043 * builtins.c (compute_objsize): Handle arrays that
8044 compute_builtin_object_size likes to fail for. Make extern.
8045 * builtins.h (compute_objsize): Declare.
8046 (check_strncpy_sizes): New function.
8047 (expand_builtin_strncpy): Call check_strncpy_sizes.
8048 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8049 -Wstringop-truncation.
8050 (gimple_fold_builtin_strncat): Same.
8051 * gimple.c (gimple_build_call_from_tree): Set call location.
8052 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8053 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8054 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8055 (handle_builtin_strlen): Use strlen_to_stridx.
8056 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8058 Use strlen_to_stridx.
8059 (pass_strlen::execute): Release strlen_to_stridx.
8060 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8061 (-Wstringop-truncation): Document new option.
8063 2017-11-10 Martin Liska <mliska@suse.cz>
8065 PR gcov-profile/82702
8066 * gcov.c (main): Handle intermediate files in a different
8068 (get_gcov_intermediate_filename): New function.
8069 (output_gcov_file): Remove support of intermediate files.
8070 (generate_results): Allocate intermediate file.
8071 (release_structures): Clean-up properly fn_end.
8072 (output_intermediate_file): Start iterating with line 1.
8074 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8076 PR tree-optimization/82929
8077 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8078 ops_swapped_p non-static data member.
8079 (store_immediate_info::store_immediate_info): Clear it.
8080 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8081 ops set ops_swapped_p.
8082 (count_multiple_uses): Handle ops_swapped_p.
8084 2017-11-10 Martin Liska <mliska@suse.cz>
8086 * coverage.c (coverage_init): Stream information about
8087 support of has_unexecuted_blocks.
8088 * doc/gcov.texi: Document that.
8089 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8090 * gcov.c (read_graph_file): Likewise.
8091 (output_line_beginning): Fix a small issue with
8094 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8096 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8097 reference of trivial component.
8099 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8102 * gimple-ssa-store-merging.c
8103 (pass_store_merging::terminate_all_aliasing_chains): For
8104 gimple_store_p stmts also call refs_output_dependent_p.
8106 PR rtl-optimization/82913
8107 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8110 2017-11-09 Jeff Law <law@redhat.com>
8112 * vr-values.h: New file with vr_values class.
8113 * tree-vrp.c: Include vr-values.h
8114 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8115 data objects into the vr_values class.
8116 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8117 (get_value_range): Make it a member function within vr_values class.
8118 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8119 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8120 (op_with_constant_singleton_value_range): Likewise.
8121 (extract_range_for_var_from_comparison_expr): Likewise.
8122 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8123 (extract_range_from_binary_expr): Likewise.
8124 (extract_range_from_unary_expr): Likewise.
8125 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8126 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8127 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8128 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8129 (compare_name_with_value, compare_names): Likewise.
8130 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8131 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8132 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8133 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8134 (extract_range_from_phi_node): Likewise.
8135 (simplify_truth_ops_using_ranges): Likewise.
8136 (simplify_div_or_mod_using_ranges): Likewise.
8137 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8138 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8139 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8140 (simplify_float_conversion_using_ranges): Likewise.
8141 (simplify_internal_call_using_ranges): Likewise.
8142 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8143 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8144 poitner via x_vr_values for calls into gimple folder.
8145 (vrp_initialize_lattice): Make this the vr_values ctor.
8146 (vrp_free_lattice): Make this the vr_values dtor.
8147 (set_vr_value): New function.
8148 (class vrp_prop): Add vr_values data member. Add various member
8149 functions as well as member functions that delegate to vr_values.
8150 (check_array_ref): Make a member function within vrp_prop class.
8151 (search_for_addr_array, vrp_initialize): Likewise.
8152 (vrp_finalize): Likewise. Revamp to avoid direct access to
8153 vr_value, values_propagated, etc.
8154 (check_array_bounds): Extract vrp_prop class instance pointer from
8155 walk info structure. Use it to call member functions.
8156 (check_all_array_refs): Make a member function within vrp_prop class.
8157 Smuggle class instance pointer via walk info structure.
8158 (x_vr_values): New local static.
8159 (vrp_valueize): Use x_vr_values to get class instance.
8160 (vr_valueize_1): Likewise.
8161 (class vrp_folder): Add vr_values data member. Add various member
8162 functions as well as member functions that delegate to vr_values.
8163 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8164 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8165 class instance from vr_values. Use it to call member functions.
8166 (vrp_dom_walker): Add vr_values data member.
8167 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8168 instance via x_vr_values.
8169 (identify_jump_threads): Accept vr_values as argument. Store
8170 it into the walker structure.
8171 (evrp_dom_walker): Add vr_values class data member. Add various
8173 (evrp_dom_walker::try_find_new_range): Use vr_values data
8174 member to access the memory allocator.
8175 (evrp_dom_walker::before_dom_children): Store vr_values class
8176 instance into the vrp_folder class.
8177 (evrp_dom_walker::push_value_range): Rework to avoid direct
8178 access to num_vr_values and vr_value.
8179 (evrp_dom_walker::pop_value_range): Likewise.
8180 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8181 Use vr_values to get to dump_all_value_ranges member function.
8182 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8183 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8184 Pass vr_values class instance down to identify_jump_threads.
8185 Remove call to vrp_free_lattice.
8186 (debug_all_value_ranges): Remove.
8188 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8189 (vrp_folder): Likewise.
8191 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8192 Get it from the existing bitmap instead.
8193 (vrp_intersect_ranges_1): Likewise.
8195 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8197 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8199 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8200 set bit_not_p to it.
8201 (imm_store_chain_info::coalesce_immediate_stores): Break group
8202 if bit_not_p is different.
8203 (count_multiple_uses, split_group,
8204 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8205 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8206 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8207 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8208 to store_immediate_info ctor.
8210 2017-11-09 Jim Wilson <jimw@sifive.com>
8212 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8213 (scan_prog_file): Likewise.
8215 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8217 * bb-reorder.c (max_entry_frequency): Remove.
8218 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8219 connect_traces, push_to_next_round_p): Remove prototypes.
8220 (find_traces_1_round): Use counts only.
8221 (push_to_next_round_p): Likewise.
8222 (find_traces): Likewise.
8223 (rotate_loop): Likewise.
8224 (find_traces_1_round): Likewise.
8225 (connect_traces): Likewise.
8226 (edge_order): Likewise.
8228 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8230 * config/arm/arm.c (output_return_instruction): Add comments to
8231 indicate requirement for cmse_nonsecure_entry return to account
8232 for the size of clearing instruction output here.
8233 (thumb_exit): Likewise.
8234 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8235 return in hardfloat mode.
8237 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8239 * config/rs6000/rs6000.c (machine_function): Add a bool,
8240 "toc_is_wrapped_separately".
8241 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8242 if it wasn't explicitly set or unset, we are optimizing for speed, and
8243 doing separate shrink-wrapping.
8244 (rs6000_get_separate_components): Enable the TOC component if
8245 saving the TOC register in the prologue.
8246 (rs6000_components_for_bb): Handle the TOC component.
8247 (rs6000_emit_prologue_components): Store the TOC register where needed.
8248 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8249 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8251 2017-11-09 Martin Jambor <mjambor@suse.cz>
8253 * ipa-param-manipulation.c: New file.
8254 * ipa-param-manipulation.h: Likewise.
8255 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8256 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8257 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8258 (ipa_parm_adjustment): Likewise.
8259 (ipa_parm_adjustment_vec): Likewise.
8260 (ipa_get_vector_of_formal_parms): Moved declaration to
8261 ipa-param-manipulation.h.
8262 (ipa_get_vector_of_formal_parm_types): Likewise.
8263 (ipa_modify_formal_parameters): Likewise.
8264 (ipa_modify_call_arguments): Likewise.
8265 (ipa_combine_adjustments): Likewise.
8266 (ipa_dump_param_adjustments): Likewise.
8267 (ipa_modify_expr): Likewise.
8268 (ipa_get_adjustment_candidate): Likewise.
8269 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8270 ipa-param-manipulation.c.
8271 (ipa_get_vector_of_formal_parm_types): Likewise.
8272 (ipa_modify_formal_parameters): Likewise.
8273 (ipa_modify_call_arguments): Likewise.
8274 (ipa_modify_expr): Likewise.
8275 (get_ssa_base_param): Likewise.
8276 (ipa_get_adjustment_candidate): Likewise.
8277 (index_in_adjustments_multiple_times_p): Likewise.
8278 (ipa_combine_adjustments): Likewise.
8279 (ipa_dump_param_adjustments): Likewise.
8280 * tree-sra.c: Also include ipa-param-manipulation.h
8281 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8284 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8285 Alan Hayward <alan.hayward@arm.com>
8286 David Sherwood <david.sherwood@arm.com>
8288 * doc/sourcebuild.texi (vect_masked_store): Document.
8290 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8291 Alan Hayward <alan.hayward@arm.com>
8292 David Sherwood <david.sherwood@arm.com>
8294 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8296 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8297 Alan Hayward <alan.hayward@arm.com>
8298 David Sherwood <david.sherwood@arm.com>
8300 * doc/sourcebuild.texi (vect_variable_length): Document.
8302 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8303 Alan Hayward <alan.hayward@arm.com>
8304 David Sherwood <david.sherwood@arm.com>
8306 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8308 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8309 Alan Hayward <alan.hayward@arm.com>
8310 David Sherwood <david.sherwood@arm.com>
8312 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8314 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8315 Alan Hayward <alan.hayward@arm.com>
8316 David Sherwood <david.sherwood@arm.com>
8318 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8319 previously undocumented selectors.
8320 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8322 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8324 * doc/rtl.texi (const_vector): Say that elements can be
8325 const_wide_ints too.
8326 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8327 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8328 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8329 * optabs.c (expand_vector_broadcast): Likewise.
8331 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8332 Alan Hayward <alan.hayward@arm.com>
8333 David Sherwood <david.sherwood@arm.com>
8335 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8336 scaled index even if the unscaled address was invalid.
8337 Don't increase the complexity of using a scale in that case.
8339 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8340 Alan Hayward <alan.hayward@arm.com>
8341 David Sherwood <david.sherwood@arm.com>
8343 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8344 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8345 UNITS_PER_WORD bytes.
8346 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8347 based on REGMODE_NATURAL_SIZE of the inner mode.
8348 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8349 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8350 * expmed.c (lowpart_bit_field_p): Divide the value up into
8351 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8352 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8353 whether something is likely to occupy more than one register.
8355 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8358 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8359 profile_count::adjust_for_ipa_scaling.
8360 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8361 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8363 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8365 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8367 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8368 (split_group): Add total_orig and total_new arguments, estimate the
8369 number of statements related to the store group without store merging
8370 and with store merging.
8371 (imm_store_chain_info::output_merged_store): Adjust split_group
8372 callers, punt if estimated number of statements with store merging
8373 is not smaller than estimated number of statements without it.
8375 (handled_load): Remove has_single_use checks.
8376 (pass_store_merging::process_store): Likewise.
8378 2017-11-09 Richard Biener <rguenther@suse.de>
8380 PR tree-optimization/82902
8381 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8383 2017-11-09 Martin Liska <mliska@suse.cz>
8386 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8389 2017-11-09 Martin Liska <mliska@suse.cz>
8391 PR tree-optimization/82669
8392 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8394 2017-11-09 Martin Liska <mliska@suse.cz>
8396 PR gcov-profile/48463
8397 * coverage.c (coverage_begin_function): Output also end locus
8398 of a function and information whether the function is
8400 * gcov-dump.c (tag_function): Parse and print the information.
8401 * gcov.c (INCLUDE_MAP): Add include.
8402 (INCLUDE_SET): Likewise.
8403 (struct line_info): Move earlier in the source file because
8404 of vector<line_info> in function_info structure.
8405 (line_info::line_info): Likewise.
8406 (line_info::has_block): Likewise.
8407 (struct source_info): Add new member index.
8408 (source_info::get_functions_at_location): New function.
8409 (function_info::group_line_p): New function.
8410 (output_intermediate_line): New function.
8411 (output_intermediate_file): Use the mentioned function.
8412 (struct function_start): New.
8413 (struct function_start_pair_hash): Likewise.
8414 (process_file): Add code that identifies group functions.
8415 Assign lines either to global or function scope.
8416 (generate_results): Skip artificial functions.
8417 (find_source): Assign index for each source file.
8418 (read_graph_file): Read new flag artificial and end_line.
8419 (add_line_counts): Assign it either to global of function scope.
8420 (accumulate_line_counts): Isolate core of the function to
8421 accumulate_line_info and call it for both function and global
8423 (accumulate_line_info): New function.
8424 (output_line_beginning): Fix GNU coding style.
8425 (print_source_line): New function.
8426 (output_line_details): Likewise.
8427 (output_function_details): Likewise.
8428 (output_lines): Iterate both source (global) scope and function
8430 (struct function_line_start_cmp): New class.
8431 * doc/gcov.texi: Reflect changes in documentation.
8433 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8436 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8437 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8438 and similarly for not instead of neg.
8440 2017-11-08 Andi Kleen <ak@linux.intel.com>
8442 * config/i386/i386.opt: Add -mforce-indirect-call.
8443 * config/i386/predicates.md: Check for flag_force_indirect_call.
8444 * doc/invoke.texi: Document -mforce-indirect-call
8446 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8448 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8450 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8451 config/riscv/riscv.c (predict.h): New include.
8452 (riscv_slow_unaligned_access_p): No longer static.
8453 (riscv_block_move_straight): Add require.
8454 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8456 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8459 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8460 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8461 nonimmediate_operand predicate for operand 1 instead of
8464 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8466 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8468 * config/aarch64/constraints.md (Uml): New constraint.
8469 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8472 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8474 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8475 of two vec_duplicates into a vec_concat.
8477 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8479 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8480 Simplify vec_merge of vec_duplicate and vec_concat.
8481 * config/aarch64/constraints.md (Utq): New constraint.
8482 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8485 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8487 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8488 Simplify vec_merge of vec_duplicate and const_vector.
8489 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8491 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8492 mode iterator. Update predicate on operand 1 to
8493 handle non-const_vec constants. Delete constraints.
8494 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8496 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8498 PR tree-optimization/78821
8499 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8501 (store_operand_info::store_operand_info): Initialize it to false.
8502 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8503 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8504 store in the group, and if chain_info is non-NULL, to ignore altogether
8506 (compatible_load_p): Fail if bit_not_p does not match.
8507 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8508 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8509 (pass_store_merging::process_store): Adjust
8510 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8511 call terminate_all_aliasing_chains newly when adding a store into
8512 a chain with non-NULL chain_info.
8514 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8516 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8518 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8520 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8522 (aarch64_layout_frame): Initialise emit_frame_chain.
8523 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8524 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8526 2017-11-08 Martin Liska <mliska@suse.cz>
8528 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8529 of gimple_call_internal_p.
8531 2017-11-07 Tom de Vries <tom@codesourcery.com>
8533 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8535 2017-11-07 Tom de Vries <tom@codesourcery.com>
8537 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8540 2017-11-08 Martin Liska <mliska@suse.cz>
8543 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8545 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8547 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8549 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8552 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8553 (*cmp<mode>_ccz_1): New insn with $k alternative.
8556 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8557 RTX_COMM_COMPARE as commutative as well.
8558 (ix86_binary_operator_ok): Formatting fix.
8559 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8560 *<code><mode>3<mask_name><round_saeonly_name>,
8561 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8562 *<s>mul<mode>3_highpart<mask_name>,
8563 *vec_widen_umult_even_v16si<mask_name>,
8564 *vec_widen_umult_even_v8si<mask_name>,
8565 *vec_widen_umult_even_v4si<mask_name>,
8566 *vec_widen_smult_even_v16si<mask_name>,
8567 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8568 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8569 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8570 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8571 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8572 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8573 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8574 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8575 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8576 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8577 ix86_binary_operator_ok. Formatting fixes.
8578 (*<plusminus_insn><mode>3<mask_name><round_name>,
8579 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8582 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8584 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8585 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8586 eq and ne if TARGET_ISEL.
8587 (cmp): New code_iterator.
8588 (UNS, UNSU_, UNSIK): New code_attrs.
8589 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8590 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8592 ("eq<mode>3"): ... this.
8593 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8595 ("ne<mode>3"): ... this.
8597 2017-11-07 Julia Koval <julia.koval@intel.com>
8600 * common/config/i386/i386-common.c
8601 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8602 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8603 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8604 * config/i386/i386.opt: Ditto.
8605 * config/i386/i386.c (ix86_target_string): Ditto.
8606 (ix86_option_override_internal): Ditto.
8607 (ix86_init_mpx_builtins): Move MPX to args2.
8608 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8609 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8610 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8611 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8612 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8613 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8614 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8616 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8619 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8620 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8621 (zero-extendsidi peephole2): Remove peephole.
8623 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8626 * stor-layout.c: Include attribs.h.
8627 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8628 explicit lookup of "aligned" attribute.
8630 2017-11-07 Andrew Waterman <andrew@sifive.com>
8632 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8633 (riscv_expand_block_move): Likewise.
8634 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8636 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8637 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8638 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8639 (riscv_adjust_block_mem): Likewise.
8640 (riscv_block_move_loop): Likewise.
8641 (riscv_expand_block_move): Likewise.
8642 * config/riscv/riscv.md (movmemsi): New pattern.
8644 2017-11-07 Michael Clark <michaeljclark@mac.com>
8646 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8647 (MUSL_DYNAMIC_LINKER): Likewise.
8649 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8651 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8652 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8655 2017-11-07 Richard Biener <rguenther@suse.de>
8657 * match.pd: Fix build.
8659 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8660 Jackson Woodruff <jackson.woodruff@arm.com>
8662 PR tree-optimization/71026
8663 * match.pd: Canonicalize negate in division.
8665 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8668 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8670 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8672 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8673 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8675 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8677 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8678 non-scalar integral types.
8679 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8680 (-(A-B), -(~A)): New transformations.
8682 2017-11-07 Tom de Vries <tom@codesourcery.com>
8684 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8685 semicolon after "do {} while (0)".
8686 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8687 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8688 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8689 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8690 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8692 2017-11-07 Tom de Vries <tom@codesourcery.com>
8694 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8695 after "do {} while (0)".
8696 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8697 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8698 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8699 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8700 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8702 2017-11-07 Tom de Vries <tom@codesourcery.com>
8705 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8707 (arm_rtx_costs_internal): Add missing semicolon after
8708 HANDLE_NARROW_SHIFT_ARITH call.
8710 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8712 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8713 disable isel if it was not set explicitly.
8715 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8717 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8718 (add_type_attribute) likewise.
8720 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8722 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8723 of struct ix86_frame.
8724 (ix86_initial_elimination_offset): Likewise.
8725 (ix86_expand_split_stack_prologue): Likewise.
8727 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8729 * tree-vrp.h (enum value_range_type): Update stale comment.
8731 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8732 Alan Hayward <alan.hayward@arm.com>
8733 David Sherwood <david.sherwood@arm.com>
8735 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8736 (aarch64_expand_vec_perm_const): Take the number of units too.
8737 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8738 (aarch64_expand_vec_perm_const): Likewise.
8739 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8740 (vec_perm<mode>): Update accordingly.
8742 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8743 Alan Hayward <alan.hayward@arm.com>
8744 David Sherwood <david.sherwood@arm.com>
8746 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8747 Take the number of units too.
8748 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8749 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8750 but check for a vector mode before rather than after the call.
8751 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8752 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8753 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8754 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8755 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8756 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8757 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8758 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8759 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8760 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8761 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8762 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8763 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8764 (aarch64_sqdmull2_n<mode>): Update accordingly.
8766 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8767 Alan Hayward <alan.hayward@arm.com>
8768 David Sherwood <david.sherwood@arm.com>
8770 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8771 the number of units too.
8772 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8773 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8774 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8775 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8776 (vec_store_lanesxi<mode>): Update accordingly.
8778 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8779 Alan Hayward <alan.hayward@arm.com>
8780 David Sherwood <david.sherwood@arm.com>
8782 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8783 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8784 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8785 of units rather than the mode.
8786 * config/aarch64/iterators.md (nunits): New mode attribute.
8787 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8788 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8789 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8790 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8791 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8792 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8793 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8794 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8795 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8796 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8797 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8798 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8799 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8800 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8801 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8802 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8803 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8804 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8805 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8806 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8807 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8808 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8809 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8810 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8811 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8812 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8813 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8814 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8815 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8816 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8817 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8818 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8819 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8820 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8821 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8822 (aarch64_simd_vec_setv2di): Likewise.
8824 2017-11-06 Carl Love <cel@us.ibm.com>
8826 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8828 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8829 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8831 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8833 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8834 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8835 Add power 8 macro expansions.
8836 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8837 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8838 power 8 instructions. (VSX_XXBR): Add iterator.
8840 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8842 * config/arm/arm.md (predicable_short_it): Change default to "no",
8843 improve documentation, remove uses that are identical to the default.
8844 (enabled_for_depr_it): Rename to enabled_for_short_it.
8845 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8846 Remove default uses.
8847 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8848 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8849 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8850 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8852 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8855 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8856 float128 helper macros, which are no longer used after deleting
8857 the old 'q' built-in functions, and moving the round to odd
8858 built-in functions to being special built-in functions.
8859 (BU_FLOAT128_2): Likewise.
8860 (BU_FLOAT128_1_HW): Likewise.
8861 (BU_FLOAT128_2_HW): Likewise.
8862 (BU_FLOAT128_3_HW): Likewise.
8863 (FABSQ): Delete old 'q' built-in functions.
8864 (COPYSIGNQ): Likewise.
8865 (SQRTF128_ODD): Move round to odd built-in functions to be
8866 special built-in functions, so that we can handle
8867 -mabi=ieeelongdouble.
8868 (TRUNCF128_ODD): Likewise.
8869 (ADDF128_ODD): Likewise.
8870 (SUBF128_ODD): Likewise.
8871 (MULF128_ODD): Likewise.
8872 (DIVF128_ODD): Likewise.
8873 (FMAF128_ODD): Likewise.
8874 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8875 built-in names to 'f128'.
8876 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8877 old 'q' built-in functions, as the machine independent code for
8878 'f128' built-in functions handles this.
8879 (rs6000_expand_builtin): Add expansion for float128 round to odd
8880 functions, keying off on -mabi=ieeelongdouble of whether to use
8881 the KFmode or TFmode variant.
8882 (rs6000_init_builtins): Initialize the _Float128 round to odd
8884 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8885 _Float128 'q' built-in functions are now mapped into the new
8886 'f128' built-in functions.
8888 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8890 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8891 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8893 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8896 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8897 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8898 * toplev.c: Include "ipa-fnsummary.h".
8899 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8901 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8903 PR tree-optimization/82838
8904 * gimple-ssa-store-merging.c
8905 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8906 on a separate gimple_seq which is then appended to seq.
8908 2017-11-06 Jeff Law <law@redhat.com>
8911 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8912 (get_probe_interval): New functions.
8913 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8914 (ix86_adjust_stack_and_probe): Likewise.
8915 (output_adjust_stack_and_probe): Likewise.
8916 (ix86_emit_probe_stack_range): Likewise.
8917 (ix86_expand_prologue): Likewise.
8919 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8921 PR tree-optimization/82816
8922 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8923 if the modes of the two types are the same.
8924 (convert_plusminus_to_widen): Likewise.
8926 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8928 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8930 (usadv16qi): New define_expand.
8931 (usadv8hi): New define_expand.
8933 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8936 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8938 (inline_transform): Likewise.
8939 * predict.c (counts_to_freqs): Remove useless conditional.
8940 * profile-count.h (profile_count::apply_scale): Move sanity check.
8941 * tree-inline.c (copy_bb): Always scale.
8942 (copy_cfg_body): Likewise.
8944 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8947 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8949 (*cmp_ite0): Add enabled_for_depr_it attribute.
8950 (*cmp_ite1): Likewise.
8952 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8954 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8957 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8959 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8960 return the value by pointer.
8961 (extract_range_from_multiplicative_op_1): Update accordingly.
8962 Return as soon as an operation fails.
8964 2017-11-05 Tom de Vries <tom@codesourcery.com>
8967 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8968 (DEF_SANITIZER_BUILTIN): ... here.
8969 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8970 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8972 2017-11-05 Tom de Vries <tom@codesourcery.com>
8975 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8977 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8978 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8979 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8981 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8982 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8983 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8985 2017-11-05 Tom de Vries <tom@codesourcery.com>
8988 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8991 2017-11-04 Michael Clark <michaeljclark@mac.com>
8993 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8994 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8996 (*addsi3_extended): Likewise.
8997 (*addsi3_extended2): Likewise.
8998 (<optab>si3): Likewise.
8999 (<optab>di3): Likewise.
9000 (<optab><mode>3): Likewise.
9001 (<*optabe>si3_internal): Likewise.
9002 (zero_extendqi<SUPERQI:mode>2): Likewise.
9003 (*add<mode>hi3): Likewise.
9004 (*xor<mode>hi3): Likewise.
9005 (<optab>di3): Likewise.
9006 (*<optab>si3_extend): Likewise.
9007 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9008 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9009 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9011 2017-11-04 Andrew Waterman <andrew@sifive.com>
9013 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9014 TARGET_STRICT_ALIGN based upon -mtune argument.
9016 2017-11-04 Andrew Waterman <andrew@sifive.com>
9018 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9020 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9022 * config/i386/i386.c (choose_basereg): Use optional scratch
9023 register and add assertion.
9024 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9025 needed, and don't allocate stack.
9026 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9027 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9028 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9030 2017-11-03 Jeff Law <law@redhat.com>
9032 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9033 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9034 to probe at the start of a noreturn function.
9036 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9038 PR tree-optimization/78821
9039 * gimple-ssa-store-merging.c: Update the file comment.
9040 (MAX_STORE_ALIAS_CHECKS): Define.
9041 (struct store_operand_info): New type.
9042 (store_operand_info::store_operand_info): New constructor.
9043 (struct store_immediate_info): Add rhs_code and ops data members.
9044 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9045 and op1r arguments to the ctor, initialize corresponding data members.
9046 (struct merged_store_group): Add load_align_base and load_align
9048 (merged_store_group::merged_store_group): Initialize them.
9049 (merged_store_group::do_merge): Update them.
9050 (merged_store_group::apply_stores): Pick the constant for
9051 encode_tree_to_bitpos from one of the two operands, or skip
9052 encode_tree_to_bitpos if neither operand is a constant.
9053 (class pass_store_merging): Add process_store method decl. Remove
9054 bool argument from terminate_all_aliasing_chains method decl.
9055 (pass_store_merging::terminate_all_aliasing_chains): Remove
9056 var_offset_p argument and corresponding handling.
9057 (stmts_may_clobber_ref_p): New function.
9058 (compatible_load_p): New function.
9059 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9060 if there is overlap and rhs_code is not INTEGER_CST. For
9061 non-overlapping stores terminate group if rhs is not mergeable.
9062 (get_alias_type_for_stmts): Change first argument from
9063 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9064 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9065 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9067 (get_location_for_stmts): Change first argument from
9068 auto_vec<gimple *> & to vec<gimple *> &.
9069 (struct split_store): Remove orig_stmts data member, add orig_stores.
9070 (split_store::split_store): Create orig_stores rather than orig_stmts.
9071 (find_constituent_stmts): Renamed to ...
9072 (find_constituent_stores): ... this. Change second argument from
9073 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9074 to info structures rather than the statements.
9075 (split_group): Rename ALLOW_UNALIGNED argument to
9076 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9077 it. Adjust find_constituent_stores caller.
9078 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9079 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9080 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9081 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9082 (mem_valid_for_store_merging): New function.
9083 (handled_load): New function.
9084 (pass_store_merging::process_store): New method.
9085 (pass_store_merging::execute): Use process_store method. Adjust
9086 terminate_all_aliasing_chains caller.
9088 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9090 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9091 Return true for more constants, symbols and label references.
9092 (aarch64_valid_floating_const): Remove unused function.
9094 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9097 * config/aarch64/aarch64.c (aarch64_layout_frame):
9098 Undo forcing of LR at bottom of frame.
9100 2017-11-03 Jeff Law <law@redhat.com>
9103 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9104 for int_registers_saved.
9106 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9107 extracted from tree-ssa-dom.c.
9108 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9109 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9110 (record_equivalences_from_incoming_edge): Add additional argument
9111 to single_pred_edge_ignoring_loop_edges call.
9112 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9113 (uncprop_dom_walker::before_dom_children): Add additional argument
9114 to single_pred_edge_ignoring_loop_edges call.
9115 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9116 single_pred_edge_ignoring_loop_edges rather than open coding.
9117 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9119 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9121 * match.pd (-(-A)): Rewrite.
9123 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9125 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9126 (rs6000_emit_int_cmove): New declaration.
9127 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9128 (rs6000_emit_sISEL): Delete.
9129 (rs6000_emit_int_cmove): Make non-static.
9130 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9131 instead of rs6000_emit_sISEL.
9133 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9135 * asan.c (create_cond_insert_point): Maintain profile.
9136 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9138 * basic-block.h (struct basic_block_def): Remove frequency.
9139 (EDGE_FREQUENCY): Use to_frequency
9140 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9142 (find_traces): Update to use to_frequency.
9143 (find_traces_1_round): Likewise; use only IPA counts.
9144 (bb_to_key): Likewise.
9145 (connect_traces): Use IPA counts only.
9146 (copy_bb_p): Update to use to_frequency.
9147 (fix_up_crossing_landing_pad): Likewise.
9148 (sanitize_hot_paths): Likewise.
9149 * bt-load.c (basic_block_freq): Likewise.
9150 * cfg.c (init_flow): Set count_max to uninitialized.
9151 (check_bb_profile): Remove frequencies; check counts.
9152 (dump_bb_info): Do not dump frequencies.
9153 (update_bb_profile_for_threading): Update counts only.
9154 (scale_bbs_frequencies_int): Likewise.
9155 (MAX_SAFE_MULTIPLIER): Remove.
9156 (scale_bbs_frequencies_gcov_type): Update counts only.
9157 (scale_bbs_frequencies_profile_count): Update counts only.
9158 (scale_bbs_frequencies): Update counts only.
9159 * cfg.h (struct control_flow_graph): Add count-max.
9160 (update_bb_profile_for_threading): Update prototype.
9161 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9162 (find_many_sub_basic_blocks): Likewise.
9163 * cfgcleanup.c (try_forward_edges): Likewise.
9164 (try_crossjump_to_edge): Likewise.
9165 * cfgexpand.c (expand_gimple_cond): Likewise.
9166 (expand_gimple_tailcall): Likewise.
9167 (construct_init_block): Likewise.
9168 (construct_exit_block): Likewise.
9169 * cfghooks.c (verify_flow_info): Check consistency of counts.
9170 (dump_bb_for_graph): Do not dump frequencies.
9171 (split_block_1): Do not update frequencies.
9172 (split_edge): Do not update frequencies.
9173 (make_forwarder_block): Do not update frequencies.
9174 (duplicate_block): Do not update frequencies.
9175 (account_profile_record): Do not update frequencies.
9176 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9177 for global heuristics.
9178 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9179 (expected_loop_iterations_unbounded): Use counts only.
9180 * cfgloopmanip.c (scale_loop_profile): Simplify.
9181 (create_empty_loop_on_edge): Simplify
9183 (duplicate_loop_to_header_edge): Simplify
9184 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9185 (update_br_prob_note): Take care of removing note when profile
9187 (relink_block_chain): Do not dump frequency.
9188 (rtl_account_profile_record): Use to_frequency.
9189 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9190 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9191 (cgraph_update_edges_for_call_stmt_node): Likewise.
9192 (cgraph_edge::verify_count_and_frequency): Update.
9193 (cgraph_node::verify_node): Temporarily disable frequency verification.
9194 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9195 to_cgraph_frequency.
9196 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9197 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9199 (cgraph_node::expand_thunk): Update profile.
9200 * except.c (dw2_build_landing_pads): Do not update frequency.
9201 * final.c (compute_alignments): Use to_frequency.
9202 (dump_basic_block_info): Do not dump frequency.
9203 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9204 (dump_gimple_bb_header): Do not dump frequency.
9205 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9207 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9208 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9209 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9210 (init_before_recovery): Do not update frequency.
9211 (sched_create_recovery_edges): Do not update frequency.
9212 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9213 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9214 (ipa_cp_c_finalize): Set max_count to uninitialized.
9215 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9216 (param_change_prob): Use counts.
9217 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9219 * ipa-split.c (consider_split): Use to_frequency.
9220 (split_function): Use to_frequency.
9221 * ira-build.c (loop_compare_func): Likewise.
9222 (mark_loops_for_removal): Likewise.
9223 (mark_all_loops_for_removal): Likewise.
9224 * loop-doloop.c (doloop_modify): Do not update frequency.
9225 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9227 * lto-streamer-in.c (input_function): Update count_max.
9228 * omp-expand.c (expand_omp_taskreg): Update count_max.
9229 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9230 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9231 (maybe_hot_count_p): Use ipa counts only.
9232 (maybe_hot_bb_p): Simplify.
9233 (maybe_hot_edge_p): Simplify.
9234 (probably_never_executed): Do not take frequency argument.
9235 (probably_never_executed_bb_p): Do not pass frequency.
9236 (probably_never_executed_edge_p): Likewise.
9237 (combine_predictions_for_bb): Check that profile is nonzero.
9238 (propagate_freq): Do not set frequency.
9239 (drop_profile): Simplify.
9240 (counts_to_freqs): Simplify.
9241 (expensive_function_p): Use to_frequency.
9242 (propagate_unlikely_bbs_forward): Simplify.
9243 (determine_unlikely_bbs): Simplify.
9244 (estimate_bb_frequencies): Add hack to silence graphite issues.
9245 (compute_function_frequency): Use ipa counts.
9246 (pass_profile::execute): Update.
9247 (rebuild_frequencies): Use counts only.
9248 (force_edge_cold): Use counts only.
9249 * profile-count.c (profile_count::dump): Dump new count types.
9250 (profile_count::differs_from_p): Check compatiblity.
9251 (profile_count::to_frequency): New function.
9252 (profile_count::to_cgraph_frequency): New function.
9253 * profile-count.h (struct function): Declare.
9254 (enum profile_quality): Add profile_guessed_local and
9255 profile_guessed_global0.
9256 (class profile_proability): Decrease number of bits to 29;
9257 update from_reg_br_prob_note and to_reg_br_prob_note.
9258 (class profile_count: Update comment; decrease number of bits
9259 to 61. Check compatibility.
9260 (profile_count::compatible_p): New private member function.
9261 (profile_count::ipa_p): New member function.
9262 (profile_count::operator<): Handle global zero correctly.
9263 (profile_count::operator>): Handle global zero correctly.
9264 (profile_count::operator<=): Handle global zero correctly.
9265 (profile_count::operator>=): Handle global zero correctly.
9266 (profile_count::nonzero_p): New member function.
9267 (profile_count::force_nonzero): New member function.
9268 (profile_count::max): New member function.
9269 (profile_count::apply_scale): Handle IPA scalling.
9270 (profile_count::guessed_local): New member function.
9271 (profile_count::global0): New member function.
9272 (profile_count::ipa): New member function.
9273 (profile_count::to_frequency): Declare.
9274 (profile_count::to_cgraph_frequency): Declare.
9275 * profile.c (OVERLAP_BASE): Delete.
9276 (compute_frequency_overlap): Delete.
9277 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9278 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9279 * sched-ebb.c (rank): Use counts only.
9280 * shrink-wrap.c (handle_simple_exit): Use counts only.
9281 (try_shrink_wrapping): Use counts only.
9282 (place_prologue_for_one_component): Use counts only.
9283 * tracer.c (find_best_predecessor): Use to_frequency.
9284 (find_trace): Use to_frequency.
9285 (tail_duplicate): Use to_frequency.
9286 * trans-mem.c (expand_transaction): Do not update frequency.
9287 * tree-call-cdce.c: Do not update frequency.
9288 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9289 (gimple_merge_blocks): Likewise.
9290 (gimple_split_edge): Likewise.
9291 (gimple_duplicate_sese_region): Likewise.
9292 (gimple_duplicate_sese_tail): Likewise.
9293 (move_sese_region_to_fn): Likewise.
9294 (gimple_account_profile_record): Likewise.
9295 (insert_cond_bb): Likewise.
9296 * tree-complex.c (expand_complex_div_wide): Likewise.
9297 * tree-eh.c (lower_resx): Update profile.
9298 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9300 (initialize_cfun): Do not initialize frequencies
9301 (freqs_to_counts): Delete.
9302 (copy_cfg_body): Ignore count parameter.
9303 (copy_body): Update.
9304 (expand_call_inline): Update count_max.
9305 (optimize_inline_calls): Update count_max.
9306 (tree_function_versioning): Update count_max.
9307 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9308 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9310 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9311 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9312 (try_peel_loop): Likewise.
9313 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9315 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9316 (tree_transform_and_unroll_loop): Do not use frequencies
9317 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9318 Use reliable prediction only.
9319 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9320 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9321 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9322 probability scaling.
9323 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9324 not update frequency
9325 (any_remaining_duplicated_blocks): Likewise.
9326 (update_profile): Likewise.
9327 (estimated_freqs_path): Delete.
9328 (freqs_to_counts_path): Delete.
9329 (clear_counts_path): Delete.
9330 (ssa_fix_duplicate_block_edges): Likewise.
9331 (duplicate_thread_path): Likewise.
9332 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9333 * tree-tailcall.c (decrease_profile): Do not update frequency.
9334 (eliminate_tail_call): Likewise.
9335 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9336 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9337 (optimize_mask_stores): Likewise.
9338 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9339 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9340 (ubsan_expand_ptr_ifn): Update profile.
9341 * value-prof.c (gimple_ic): Simplify.
9342 * value-prof.h (gimple_ic): Update prototype.
9343 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9344 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9346 (want_inline_self_recursive_call_p): Likewise.
9347 (resolve_noninline_speculation): Only cummulate defined counts.
9348 (inline_small_functions): Use nonzero_p.
9349 (ipa_inline): Do not access freed node.
9351 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9353 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9354 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9356 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9358 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9359 non-legitimate address.
9361 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9363 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9364 (*lt0_<mode>di, *lt0_<mode>si): New.
9366 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9368 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9369 TARGET_PAIRED_FLOAT.
9371 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9372 Jim Wilson <jim.wilson@linaro.org>
9374 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9375 * config/aarch64/aarch64-tune.md: Regenerated.
9376 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9377 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9379 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9381 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9382 (arc_expand_prologue): Restore blink for millicode.
9383 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9385 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9388 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9389 gen_vec_duplicate after forcing the scalar into a register.
9391 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9393 * combine (try_combine): Print the insns input to try_combine to the
9396 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9399 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9400 Remove second argument from aarch64_process_target_attr call.
9401 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9403 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9405 (aarch64_handle_attr_arch): Remove second argument.
9406 (aarch64_handle_attr_cpu): Ditto.
9407 (aarch64_handle_attr_tune): Ditto.
9408 (aarch64_handle_attr_isa_flags): Ditto.
9409 (aarch64_process_one_target_attr): Ditto.
9410 (aarch64_process_target_attr): Ditto.
9411 (aarch64_option_valid_attribute_p): Remove second argument.
9412 on aarch64_process_target_attr call.
9414 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9416 * diagnostic.c: Include "selftest-diagnostic.h".
9417 (selftest::assert_location_text): New function.
9418 (selftest::test_diagnostic_get_location_text): New function.
9419 (selftest::diagnostic_c_tests): Call it.
9421 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9423 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9424 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9425 (class selftest::test_diagnostic_context): Move to...
9426 * selftest-diagnostic.c: New file.
9427 * selftest-diagnostic.h: New file.
9429 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9431 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9432 offset range for FT32B.
9433 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9434 * config/ft32/ft32.md: Add TARGET_NOPM.
9435 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9436 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9438 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9440 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9442 2017-11-02 Jeff Law <law@redhat.com>
9444 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9445 virtual keyword on FINAL OVERRIDE members.
9447 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9448 virtuals together. Add virtual destructor.
9449 (substitute_and_fold_engine): Similarly.
9451 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9453 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9455 2017-11-02 Richard Biener <rguenther@suse.de>
9457 PR tree-optimization/82795
9458 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9460 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9462 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9464 (gcc_GAS_CHECK_FEATURE): Remove.
9465 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9466 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9467 * configure: Regenerate.
9469 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9471 * config/arc/arc.c (hwloop_optimize): Account for empty
9474 2017-11-02 Richard Biener <rguenther@suse.de>
9477 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9478 Truncate ARRAY_REF index and element size.
9480 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9482 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9484 2017-11-01 Jeff Law <law@redhat.com>
9486 * tree-ssa-ccp.c (ccp_folder): New class derived from
9487 substitute_and_fold_engine.
9488 (ccp_folder::get_value): New member function.
9489 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9490 (ccp_fold_stmt): Remove prototype.
9491 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9492 * tree-ssa-copy.c (copy_folder): New class derived from
9493 substitute_and_fold_engine.
9494 (copy_folder::get_value): Renamed from get_value.
9495 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9496 * tree-vrp.c (vrp_folder): New class derived from
9497 substitute_and_fold_engine.
9498 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9499 (vrp_folder::get_value): New member function.
9500 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9501 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9502 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9503 provide a class interface to folder/substitute routines.
9504 (ssa_prop_fold_stmt_fn): Remove typedef.
9505 (ssa_prop_get_value_fn): Likewise.
9506 (subsitute_and_fold): Remove prototype.
9507 (replace_uses_in): Likewise.
9508 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9509 Renamed from replace_uses_in. Call the virtual member function
9510 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9511 (substitute_and_fold_dom_walker): Remove initialization of
9512 data member entries for calbacks. Add substitute_and_fold_engine
9513 member and initialize it.
9514 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9515 member functions for get_value, replace_phi_args_in c
9516 replace_uses_in, and fold_stmt calls.
9517 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9518 substitute_and_fold. Remove assert. Update ctor call.
9520 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9521 (ssa_prop_visit_phi_fn): Likewise.
9522 (class ssa_propagation_engine): New class to provide an interface
9524 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9526 (ssa_prop_visit_phi): Likewise.
9527 (ssa_propagation_engine::simulate_stmt): Moved into class.
9528 Call visit_phi/visit_stmt from the class rather than via
9529 file scoped static variables.
9530 (ssa_propagation_engine::simulate_block): Moved into class.
9531 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9532 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9533 set file scoped statics for the visit_stmt/visit_phi callbacks.
9534 * tree-complex.c (complex_propagate): New class derived from
9535 ssa_propagation_engine.
9536 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9537 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9538 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9540 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9541 ssa_propagation_engine.
9542 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9543 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9544 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9545 * tree-ssa-copy.c (copy_prop): New class derived from
9546 ssa_propagation_engine.
9547 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9548 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9549 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9550 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9551 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9552 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9553 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9555 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9557 PR rtl-optimization/82778
9558 PR rtl-optimization/82597
9559 * compare-elim.c (struct comparison): Add in_a_setter field.
9560 (find_comparison_dom_walker::before_dom_children): Remove killed
9561 bitmap and df_simulate_find_defs call, instead walk the defs.
9562 Compute last_setter and initialize in_a_setter. Merge definitions
9563 with first initialization for a few variables.
9564 (try_validate_parallel): Use insn_invalid_p instead of
9565 recog_memoized. Return insn rather than just the pattern.
9566 (try_merge_compare): Fix up comment. Don't uselessly test if
9567 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9569 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9572 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9573 Alan Hayward <alan.hayward@arm.com>
9574 David Sherwood <david.sherwood@arm.com>
9576 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9577 aarch64_hard_regno_nregs to get the number of registers
9580 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9581 Alan Hayward <alan.hayward@arm.com>
9582 David Sherwood <david.sherwood@arm.com>
9584 * config/aarch64/constraints.md (Upl): Rename to...
9586 * config/aarch64/aarch64.md
9587 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9590 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9591 Alan Hayward <alan.hayward@arm.com>
9592 David Sherwood <david.sherwood@arm.com>
9594 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9595 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9598 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9599 Alan Hayward <alan.hayward@arm.com>
9600 David Sherwood <david.sherwood@arm.com>
9602 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9603 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9604 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9606 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9607 of a single element.
9608 * config/aarch64/iterators.md: Add a comment above the permute
9609 unspecs to say that they are generated directly by
9610 aarch64_expand_vec_perm_const.
9611 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9613 2017-11-01 Nathan Sidwell <nathan@acm.org>
9615 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9617 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9619 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9620 and medany code models, and describe what they do.
9622 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9624 Revert accidental duplicate:
9626 * combine.c (can_change_dest_mode): Reject changes in
9627 REGMODE_NATURAL_SIZE.
9629 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9631 PR rtl-optimization/64682
9632 PR rtl-optimization/69567
9633 PR rtl-optimization/69737
9634 PR rtl-optimization/82683
9635 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9636 register mentioned in the note, drop the note, unless it came from I3,
9637 in which case it should go to I3 again.
9639 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9641 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9642 and return false if not.
9643 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9645 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9647 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9648 range is known to be empty.
9650 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9651 Alan Hayward <alan.hayward@arm.com>
9652 David Sherwood <david.sherwood@arm.com>
9654 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9655 and CONST_VECTOR_NUNITS instead of computing the number of units from
9656 the byte sizes of the vector and element.
9657 (simplify_binary_operation_1): Likewise.
9658 (simplify_const_binary_operation): Likewise.
9659 (simplify_ternary_operation): Likewise.
9661 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9662 Alan Hayward <alan.hayward@arm.com>
9663 David Sherwood <david.sherwood@arm.com>
9665 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9666 (int_mem_offset): ...this new function.
9667 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9668 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9669 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9672 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9673 Alan Hayward <alan.hayward@arm.com>
9674 David Sherwood <david.sherwood@arm.com>
9676 * lower-subreg.c (interesting_mode_p): New function.
9677 (compute_costs, find_decomposable_subregs, decompose_register)
9678 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9679 (resolve_clobber, dump_choices): Use it.
9681 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9682 Alan Hayward <alan.hayward@arm.com>
9683 David Sherwood <david.sherwood@arm.com>
9685 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9687 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9688 Alan Hayward <alan.hayward@arm.com>
9689 David Sherwood <david.sherwood@arm.com>
9691 * alias.c (find_base_value, find_base_term): Only process integer
9692 truncations. Check the precision rather than the size.
9694 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9695 Alan Hayward <alan.hayward@arm.com>
9696 David Sherwood <david.sherwood@arm.com>
9698 * machmode.h (is_narrower_int_mode): New function
9699 * optabs.c (expand_float, expand_fix): Use it.
9700 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9702 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9703 Alan Hayward <alan.hayward@arm.com>
9704 David Sherwood <david.sherwood@arm.com>
9706 * rtl.h (narrower_subreg_mode): New function.
9707 * ira-color.c (update_costs_from_allocno): Use it.
9709 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9710 Alan Hayward <alan.hayward@arm.com>
9711 David Sherwood <david.sherwood@arm.com>
9713 * optabs-query.h (convert_optab_p): New function, split out from...
9714 (convert_optab_handler): ...here.
9715 (widening_optab_handler): Delete.
9716 (find_widening_optab_handler): Remove permit_non_widening parameter.
9717 (find_widening_optab_handler_and_mode): Likewise. Provide an
9718 override that operates on mode class wrappers.
9719 * optabs-query.c (widening_optab_handler): Delete.
9720 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9721 parameter. Assert that the two modes are the same class and that
9722 the "from" mode is narrower than the "to" mode. Use
9723 convert_optab_handler instead of widening_optab_handler.
9724 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9725 instead of widening_optab_handler.
9726 * expr.c (expand_expr_real_2): Update calls to
9727 find_widening_optab_handler.
9728 * optabs.c (expand_widen_pattern_expr): Likewise.
9729 (expand_binop_directly): Take the insn_code as a parameter.
9730 (expand_binop): Only call find_widening_optab_handler for
9731 conversion optabs; use optab_handler otherwise. Update calls
9732 to find_widening_optab_handler and expand_binop_directly.
9733 Use convert_optab_handler instead of widening_optab_handler.
9734 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9735 find_widening_optab_handler and use scalar_mode rather than
9737 (convert_plusminus_to_widen): Likewise.
9739 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9740 Alan Hayward <alan.hayward@arm.com>
9741 David Sherwood <david.sherwood@arm.com>
9743 * machmode.h (fixed_size_mode): New class.
9744 * rtl.h (get_pool_mode): Return fixed_size_mode.
9745 * gengtype.c (main): Add fixed_size_mode.
9746 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9747 (get_raw_arg_mode): Likewise.
9748 * doc/tm.texi: Regenerate.
9749 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9750 * targhooks.c (default_get_reg_raw_mode): Likewise.
9751 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9752 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9753 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9754 (msp430_get_raw_result_mode): Likewise.
9755 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9756 * dbxout.c (dbxout_parms): Require fixed-size modes.
9757 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9758 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9759 * omp-low.c (lower_oacc_reductions): Likewise.
9760 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9761 (simplify_subreg): Update accordingly.
9762 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9763 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9764 that aren't fixed-size.
9765 (get_pool_mode): Return a fixed_size_mode.
9766 (output_constant_pool_2): Take a fixed_size_mode.
9768 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9769 Alan Hayward <alan.hayward@arm.com>
9770 David Sherwood <david.sherwood@arm.com>
9772 * doc/rtl.texi (vec_series): Document.
9773 (const): Say that the operand can be a vec_series.
9774 * rtl.def (VEC_SERIES): New rtx code.
9775 * rtl.h (const_vec_series_p_1): Declare.
9776 (const_vec_series_p): New function.
9777 * emit-rtl.h (gen_const_vec_series): Declare.
9778 (gen_vec_series): Likewise.
9779 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9780 (gen_vec_series): Likewise.
9781 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9782 * simplify-rtx.c (simplify_unary_operation): Handle negations
9784 (simplify_binary_operation_series): New function.
9785 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9786 (test_vector_ops_series): New function.
9787 (test_vector_ops): Call it.
9788 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9789 gen_const_vec_series.
9790 (altivec_lvsr): Likewise.
9791 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9793 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9794 Alan Hayward <alan.hayward@arm.com>
9795 David Sherwood <david.sherwood@arm.com>
9797 * doc/rtl.texi (const): Update description of address constants.
9798 Say that vector constants are allowed too.
9799 * common.md (E, F): Use CONSTANT_P instead of checking for
9801 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9802 checking for CONST_VECTOR.
9803 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9805 * expr.c (expand_expr_real_2): Check for vector modes instead
9806 of checking for CONST_VECTOR.
9807 * rtl.h (const_vec_p): New function.
9808 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9809 (unwrap_const_vec_duplicate): Handle them here too.
9811 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9812 David Malcolm <dmalcolm@redhat.com>
9813 Alan Hayward <alan.hayward@arm.com>
9814 David Sherwood <david.sherwood@arm.com>
9816 * rtl.h (vec_duplicate_p): New function.
9817 * selftest-rtl.c (assert_rtx_eq_at): New function.
9818 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9819 (assert_rtx_eq_at): Declare.
9820 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9821 * selftest-run-tests.c (selftest::run_tests): Call it.
9822 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9823 (simplify_unary_operation_1): Recursively handle vector duplicates.
9824 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9826 (simplify_subreg): Handle subregs of vector duplicates.
9827 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9828 (selftest::simplify_rtx_c_tests): New functions.
9830 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9831 Alan Hayward <alan.hayward@arm.com>
9832 David Sherwood <david.sherwood@arm.com>
9834 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9835 (gen_vec_duplicate): Likewise.
9836 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9838 (gen_const_vector): ...here.
9839 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9840 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9841 whose elements are all equal.
9842 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9843 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9844 (simplify_relational_operation): Likewise.
9845 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9847 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9848 (aarch64_expand_vector_init): Likewise.
9849 * config/arm/arm.c (neon_vdup_constant): Likewise.
9850 (neon_expand_vector_init): Likewise.
9851 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9852 (arm_block_set_unaligned_vect): Likewise.
9853 (arm_block_set_aligned_vect): Likewise.
9854 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9855 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9856 (expand_vec_perm_even_odd_pack): Likewise.
9857 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9858 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9859 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9860 gen_const_vec_duplicate.
9861 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9862 * config/mips/mips.c (mips_gen_const_int_vector): Use
9863 gen_const_vec_duplicate.
9864 (mips_expand_vector_init): Use CONST0_RTX.
9865 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9866 (define_split): Use gen_const_vec_duplicate.
9867 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9868 (define_split): Use gen_const_vec_duplicate.
9869 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9870 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9871 * config/spu/spu.c (spu_const): Likewise.
9873 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9874 Alan Hayward <alan.hayward@arm.com>
9875 David Sherwood <david.sherwood@arm.com>
9877 * combine.c (can_change_dest_mode): Reject changes in
9878 REGMODE_NATURAL_SIZE.
9880 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9882 * configure.ac (--enable-libssp): New.
9883 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9884 trying to determine target-specific default. Adjust indentation.
9885 * configure: Regenerated.
9886 * doc/install.texi (Configuration): Expand --disable-libssp
9889 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9891 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9894 2017-10-31 Martin Jambor <mjambor@suse.cz>
9897 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9899 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9901 * auto-profile.c (autofdo_source_profile::read): Use
9902 UNKNOWN_LOCATION rather than 0.
9903 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9904 (warning_at): ...this overload.
9905 (warning_at_rich_loc_n): Rename to...
9906 (warning_n): ...this overload.
9907 (error_at_rich_loc): Rename to...
9908 (error_at): ...this overload.
9909 (pedwarn_at_rich_loc): Rename to...
9910 (pedwarn): ...this overload.
9911 (permerror_at_rich_loc): Rename to...
9912 (permerror): ...this overload.
9913 (inform_at_rich_loc): Rename to...
9914 (inform): ...this overload.
9915 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9916 (diagnostic_n_impl_richloc): Rename to...
9917 (diagnostic_n_impl): ...this rich_location *-based decl.
9918 (inform_at_rich_loc): Rename to...
9919 (inform): ...this, and add an assertion.
9920 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9921 (warning_at_rich_loc): Rename to...
9922 (warning_at): ...this, and add an assertion.
9923 (warning_at_rich_loc_n): Rename to...
9924 (warning_n): ...this, and add an assertion.
9925 (warning_n): Update location_t-based implementation for removal of
9926 location_t-based diagnostic_n_impl.
9927 (pedwarn_at_rich_loc): Rename to...
9928 (pedwarn): ...this, and add an assertion.
9929 (permerror_at_rich_loc): Rename to...
9930 (permerror): ...this, and add an assertion.
9931 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9932 (error_at_rich_loc): Rename to...
9933 (error_at): ...this, and add an assertion.
9934 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9935 (driver::do_spec_on_infiles): Likewise.
9936 * substring-locations.c (format_warning_va): Update for renaming
9937 of inform_at_rich_loc.
9939 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9941 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9942 _Float<N>X built-in functions so that the variant without the
9943 "__builtin_" prefix is only enabled for the GNU C and Objective C
9944 languages when they are in non-strict ANSI/ISO mode.
9945 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9946 * target.def (floatn_builtin_p): Add a target hook to control
9947 whether _Float<N> and _Float<N>X built-in functions without the
9948 "__builtin_" prefix are enabled, and return true for C and
9949 Objective C in the default hook. Include langhooks.h in
9951 * targhooks.h (default_floatn_builtin_p): Likewise.
9952 * targhooks.c (default_floatn_builtin_p): Likewise.
9953 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9954 floatn_builtin_p target hook.
9955 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9957 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9958 Eric Botcazou <ebotcazou@adacore.com>
9960 PR rtl-optimization/81803
9961 * lra-constraints.c (curr_insn_transform): Also reload the whole
9962 register for a strict subreg no wider than a word if this is for
9963 a WORD_REGISTER_OPERATIONS target.
9965 2017-10-31 Jason Merrill <jason@redhat.com>
9967 * gdbinit.in: Skip over inlines from timevar.h.
9969 2017-10-31 Martin Liska <mliska@suse.cz>
9971 * doc/gcov.texi: Document new option.
9972 * gcov.c (print_usage): Likewise print it.
9973 (process_args): Support the argument.
9974 (format_count): New function.
9975 (format_gcov): Use the function.
9977 2017-10-31 Martin Liska <mliska@suse.cz>
9979 * gcov.c (struct name_map): do not use typedef.
9980 Define operator== and operator<.
9981 (name_search): Remove.
9982 (name_sort): Remove.
9983 (main): Do not allocate names.
9984 (process_file): Add vertical space.
9985 (generate_results): Use std::find.
9986 (release_structures): Do not release memory.
9987 (find_source): Use std::find.
9989 2017-10-31 Martin Liska <mliska@suse.cz>
9991 * gcov.c (struct line_info): Remove it's typedef.
9992 (line_info::line_info): Add proper ctor.
9993 (line_info::has_block): Do not use a typedef.
9994 (struct source_info): Do not use typedef.
9995 (circuit): Likewise.
9996 (get_cycles_count): Likewise.
9997 (output_intermediate_file): Iterate via vector iterator.
9998 (add_line_counts): Use std::vector methods.
9999 (accumulate_line_counts): Likewise.
10000 (output_lines): Likewise.
10002 2017-10-31 Martin Liska <mliska@suse.cz>
10004 * gcov.c (struct source_info): Remove typedef.
10005 (source_info::source_info): Add proper ctor.
10006 (accumulate_line_counts): Use struct, not it's typedef.
10007 (output_gcov_file): Likewise.
10008 (output_lines): Likewise.
10009 (main): Do not allocate an array.
10010 (output_intermediate_file): Use size of vector container.
10011 (process_file): Resize the vector.
10012 (generate_results): Do not preallocate, use newly added vector
10014 (release_structures): Do not release sources.
10015 (find_source): Use vector methods.
10016 (add_line_counts): Do not use typedef.
10018 2017-10-31 Martin Liska <mliska@suse.cz>
10020 * doc/gcov.texi: Document that.
10021 * gcov.c (add_line_counts): Mark lines with a non-executed
10023 (output_line_beginning): Handle such lines.
10024 (output_lines): Pass new argument.
10025 (output_intermediate_file): Print it in intermediate format.
10027 2017-10-31 Martin Liska <mliska@suse.cz>
10029 * color-macros.h: New file.
10030 * diagnostic-color.c: Factor out color related to macros to
10032 * doc/gcov.texi: Document -k option.
10033 * gcov.c (INCLUDE_STRING): Include string.h.
10034 (print_usage): Add -k option.
10035 (process_args): Parse it.
10036 (pad_count_string): New function.
10037 (output_line_beginning): Likewise.
10038 (DEFAULT_LINE_START): New macro.
10039 (output_lines): Support color output.
10041 2017-10-31 Martin Liska <mliska@suse.cz>
10043 PR gcov-profile/82633
10044 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10045 their interaction with GCOV infrastructure.
10046 * configure.ac: Add -fkeep-{inline,static}-functions to
10048 * configure: Regenerate.
10050 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10053 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10055 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10058 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10059 into a register if it does not fit into an immediate offset field.
10061 2017-10-31 Olivier Hainque <hainque@adacore.com>
10063 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10065 2017-10-31 Julia Koval <julia.koval@intel.com>
10067 * config.gcc: Add gfniintrin.h.
10068 * config/i386/gfniintrin.h: New.
10069 * config/i386/i386-builtin-types.def
10070 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10071 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10072 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10073 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10074 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10075 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10076 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10077 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10078 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10079 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10080 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10081 * config/i386/immintrin.h: Include gfniintrin.h.
10082 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10084 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10086 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10088 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10090 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10091 (arm_ashldi3_1bit): Remove pattern.
10092 (ashrdi3): Remove shift by 1 expansion.
10093 (arm_ashrdi3_1bit): Remove pattern.
10094 (lshrdi3): Remove shift by 1 expansion.
10095 (arm_lshrdi3_1bit): Remove pattern.
10096 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10097 cost of ashldi3 by 1.
10098 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10099 (<shift>di3_neon): Likewise.
10101 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10103 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10104 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10105 and (*aarch64_simd_mov<VQ:mode>).
10106 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10107 pattern alternative.
10108 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10109 attributes to match pattern alternative.
10111 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10113 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10114 * config/rs6000/emmintrin.h: New file.
10115 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10117 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10119 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10120 * (movdi_vfp_cortexa8): Remove pattern.
10122 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10124 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10126 (Specific, alpha*-dec-osf5.1): Remove.
10127 (Specific, mips-sgi-irix5): Remove.
10128 (Specific, mips-sgi-irix6): Remove.
10130 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10132 PR middle-end/22141
10133 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10134 arguments to clear_bit_region_be.
10136 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10138 * gimplify.c: Include memmodel.h.
10140 2017-10-30 Martin Jambor <mjambor@suse.cz>
10142 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10143 condition whether loop should be executed at all.
10145 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10147 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10148 gimple folding of vec_madd() intrinsics.
10149 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10150 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10151 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10153 2017-10-30 Richard Biener <rguenther@suse.de>
10155 PR tree-optimization/82762
10157 2017-10-23 Richard Biener <rguenther@suse.de>
10159 PR tree-optimization/82129
10161 2017-08-01 Richard Biener <rguenther@suse.de>
10163 PR tree-optimization/81181
10164 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10165 (compute_antic): ... end of iteration here.
10167 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10169 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10170 -std=iso9899:2017 and -std=gnu17.
10171 * doc/standards.texi (C Language): Document C17 support.
10172 * doc/cpp.texi (Overview): Mention -std=c17.
10173 (Standard Predefined Macros): Document C11 and C17 values of
10174 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10175 * doc/extend.texi (Inline): Do not list individual options for
10176 standards newer than C99.
10177 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10179 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10182 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10184 * asan.c (asan_finish_file): Align asan globals array by shadow
10187 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10189 PR middle-end/22141
10190 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10191 (struct store_immediate_info): Add bitregion_start and bitregion_end
10193 (store_immediate_info::store_immediate_info): Add brs and bre
10194 arguments and initialize bitregion_{start,end} from those.
10195 (struct merged_store_group): Add bitregion_start, bitregion_end,
10196 align_base and mask fields. Drop unnecessary struct keyword from
10197 struct store_immediate_info. Add do_merge method.
10198 (clear_bit_region_be): Use memset instead of loop storing zeros.
10199 (merged_store_group::do_merge): New method.
10200 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10201 stores as long as the surrounding bitregions have no gaps.
10202 (merged_store_group::merge_overlapping): Use do_merge.
10203 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10204 is byte aligned, rather than requiring that start and width are
10205 byte aligned. Drop unnecessary struct keyword from
10206 struct store_immediate_info. Allocate and populate also mask array.
10207 Make start of the arrays relative to bitregion_start rather than
10208 start and size them according to bitregion_{end,start} difference.
10209 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10210 struct store_immediate_info.
10211 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10212 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10213 struct keyword from struct store_immediate_info.
10214 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10215 between stores as long as the surrounding bitregions have no gaps.
10217 (struct split_store): Add orig non-static data member.
10218 (split_store::split_store): Initialize orig to false.
10219 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10220 if there is exactly a single original stmt. Change stmts argument
10221 to pointer from reference, if NULL, don't push anything to it. Add
10222 first argument, use it to optimize skipping over orig stmts that
10223 are known to be before bitpos already. Simplify.
10224 (split_group): Return unsigned int count how many stores are or
10225 would be needed rather than a bool. Add allow_unaligned argument.
10226 Change split_stores argument from reference to pointer, if NULL,
10227 only do a dry run computing how many stores would be produced.
10228 Rewritten algorithm to use both alignment and misalign if
10229 !allow_unaligned and handle bitfield stores with gaps.
10230 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10231 from bitregion_start instead of start. Compute allow_unaligned
10232 here, if true, do 2 split_group dry runs to compute which one
10233 produces fewer stores and prefer aligned if equal. Punt if
10234 new count is bigger or equal than original before emitting any
10235 statements, rather than during that. Remove no longer needed
10236 new_ssa_names tracking. Replace num_stmts with
10237 split_stores.length (). Use 32-bit stack allocated entries
10238 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10239 if possible. Handle bitfields with gaps.
10240 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10241 Compute bitregion_{start,end} for the stores and construct
10242 store_immediate_info with that. Formatting fixes.
10244 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10247 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10248 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10250 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10252 * gimplify.c: Include tm_p.h.
10254 * common.opt (gcoff): Re-add as ignored option.
10255 (gcoff1, gcoff2, gcoff3): Likewise.
10257 * Makefile.in (OBJS): Delete sdbout.o.
10258 (GTFILES): Delete $(srcdir)/sdbout.c.
10259 * debug.h: Delete sdb_debug_hooks.
10260 * final.c: Delete sdbout.h include.
10261 (final_scan_insn): Delete SDB_DEBUG check.
10262 (rest_of_clean_state): Likewise.
10263 * output.h: Delete sdb_begin_function_line.
10264 * sdbout.c: Delete.
10265 * sdbout.h: Delete.
10266 * toplev.c: Delete sdbout.h include.
10267 (process_options): Delete SDB_DEBUG check.
10268 * tree-core.h (tree_type_common): Delete pointer field of
10270 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10271 TYPE_SYMTAB_POINTER.
10272 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10273 (TYPE_SYMTAB_IS_POINTER): Delete.
10274 (TYPE_SYMTAB_IS_DIE): Renumber.
10275 * xcoffout.c: Refer to former sdbout.c file.
10276 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10278 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10279 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10280 (Debugging Options): Delete -gcoff.
10281 (-gstabs): Delete SDB reference.
10283 (-gcoff@var{level}): Delete.
10284 * doc/passes.texi (Debugging information output): Delete SDB and
10285 sdbout.c references.
10286 * doc/tm.texi: Regenerate.
10287 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10288 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10290 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10291 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10292 (SDB_DEBUGGING_INFO): Delete.
10293 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10294 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10295 * target.def (output_source_filename): Delete COFF reference.
10297 * common.opt (gcoff): Delete.
10298 (gxcoff+): Update Negative chain.
10299 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10301 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10302 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10303 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10304 and SDB references.
10305 (expand_function_start): Change sdb reference to past tense.
10306 (expand_function_end): Change sdb reference to past tense.
10307 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10308 * opts.c (debug_type_names): Delete coff entry.
10309 (common_handle_option): Delete OPT_gcoff case.
10310 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10312 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10313 * config/cris/cris.h: Delete SDB reference in comment.
10314 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10315 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10316 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10317 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10319 (ix86_expand_prologue): Likewise.
10320 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10321 * config/ia64/ia64.h: Likewise.
10322 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10323 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10325 * config/mmix/mmix.h: Likewise.
10326 * config/nds32/nds32.c: Likewise.
10327 * config/stormy/storym16.h: Likewise.
10328 * config/visium/visium.h: Likewise.
10329 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10331 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10333 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10334 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10335 FRAME_POINTER_REGNUM point at high end of local var area.
10337 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10339 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10340 Move comment around. Do not reset best_edge for a copiable
10341 destination if the copy would cause a partition change.
10342 (better_edge_p): Remove redundant check.
10344 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10346 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10348 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10350 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10351 for math functions that have _Float<N> and _Float<N>X variants.
10352 (mathfn_built_in_2): Add support for math functions that have
10353 _Float<N> and _Float<N>X variants.
10354 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10355 (expand_builtin_mathfn_ternary): Add support for fma with
10356 _Float<N> and _Float<N>X variants.
10357 (expand_builtin): Likewise.
10358 (fold_builtin_3): Likewise.
10359 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10360 create math function _Float<N> and _Float<N>X variants as external
10362 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10363 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10364 the __builtin_ prefix and if not strict ansi, without the prefix.
10365 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10366 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10367 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10368 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10369 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10370 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10371 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10372 function signatures for fma _Float<N> and _Float<N>X variants.
10373 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10374 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10375 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10376 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10377 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10378 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10379 * gencfn-macros.c (print_case_cfn): Add support for math functions
10380 that have _Float<N> and _Float<N>X variants.
10381 (print_define_operator_list): Likewise.
10382 (fltfn_suffixes): Likewise.
10384 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10385 for math functions that have _Float<N> and _Float<N>X variants.
10386 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10387 and _Float<N>X variants.
10388 (COPYSIGN): Likewise.
10391 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10392 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10394 (integer_valued_read_call_p): Likewise.
10395 * fold-const-call.c (fold_const_call_ss): Likewise.
10396 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10397 _Float<N> and _Float<N>X variants.
10398 (fold_const_call_ssss): Add support for fma _Float<N> and
10399 _Float<N>X variants.
10400 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10401 support for copysign and fma _Float<N> and _Float<N>X variants.
10402 (backprop::process_builtin_call_use): Likewise.
10403 * tree-call-cdce.c (can_test_argument_range); Add support for
10404 sqrt _Float<N> and _Float<N>X variants.
10405 (edom_only_function): Likewise.
10406 (get_no_error_domain): Likewise.
10407 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10408 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10409 copysign _Float<N> and _Float<N>X variants.
10410 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10411 handled by machine independent code.
10412 (FMAF128): Likewise.
10413 * doc/cpp.texi (Common Predefined Macros): Document defining
10414 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10415 fma _Float<N> and _Float<N>X variants.
10417 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10420 * config/i386/i386-modes.def (CCFPU): Remove definition.
10421 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10422 (ix86_cc_modes_compatible): Ditto.
10423 (ix86_expand_carry_flag_compare): Ditto.
10424 (ix86_expand_int_movcc): Ditto.
10425 (ix86_expand_int_addcc): Ditto.
10426 (ix86_reverse_condition): Ditto.
10427 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10428 Return true/false for unordered/ordered fp comparisons.
10429 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10430 (ix86_prepare_fp_compare_args): Update for rename.
10431 (ix86_expand_fp_compare): Update for rename. Generate unordered
10432 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10433 (ix86_expand_sse_compare_and_jump): Ditto.
10434 * config/i386/predicates.md (fcmov_comparison_operator):
10435 Remove CCFPU mode handling.
10436 (ix86_comparison_operator): Ditto.
10437 (ix86_carry_flag_operator): Ditto.
10438 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10439 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10440 (*cmpu<mode>_cc_i387): Ditto.
10441 (FPCMP): Remove mode iterator.
10442 (unord): Remove mode attribute.
10443 (unord_subst): New define_subst transformation
10444 (unord): New define_subst attribute.
10445 (unordered): Ditto.
10446 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10447 (*cmpi<unord>xf_i387): Ditto.
10448 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10449 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10450 using unord_subst transformation.
10451 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10452 (round_saeonly): Also handle CCFP mode.
10453 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10454 Remove UNSPEC_SAHF unspec handling.
10456 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10458 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10460 2017-10-27 Jeff Law <law@redhat.com>
10462 * gimple-ssa-sprintf.c: Include domwalk.h.
10463 (class sprintf_dom_walker): New class, derived from dom_walker.
10464 (sprintf_dom_walker::before_dom_children): New function.
10465 (struct call_info): Moved into sprintf_dom_walker class
10466 (compute_formath_length, handle_gimple_call): Likewise.
10467 (sprintf_length::execute): Call the dominator walker rather
10468 than walking the statements.
10470 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10471 gimple statement locations.
10472 (check_array_bounds): Corresponding changes. Get the statement's
10473 location directly from wi->stmt.
10475 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10478 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10480 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10482 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10483 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10485 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10488 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10489 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10490 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10491 maybe_get_pool_constant.
10492 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10495 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10497 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10498 2.26 caveat. Update gas and gld versions.
10499 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10502 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10504 * cgraph.h (set_malloc_flag): Declare.
10505 * cgraph.c (set_malloc_flag_1): New function.
10506 (set_malloc_flag): Likewise.
10507 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10508 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10510 (read_ipa_call_summary): Add support for reading is_return_callee.
10511 (write_ipa_call_summary): Stream is_return_callee.
10512 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10513 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10514 ipa-prop.h, ipa-fnsummary.h.
10515 (pure_const_names): Change to static.
10516 (malloc_state_e): Define.
10517 (malloc_state_names): Define.
10518 (funct_state_d): Add field malloc_state.
10519 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10520 (check_retval_uses): New function.
10521 (malloc_candidate_p): Likewise.
10522 (analyze_function): Add support for malloc attribute.
10523 (pure_const_write_summary): Stream malloc_state.
10524 (pure_const_read_summary): Add support for reading malloc_state.
10525 (dump_malloc_lattice): New function.
10526 (propagate_malloc): New function.
10527 (warn_function_malloc): New function.
10528 (ipa_pure_const::execute): Call propagate_malloc and
10529 ipa_free_fn_summary.
10530 (pass_local_pure_const::execute): Add support for malloc attribute.
10531 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10532 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10534 2017-10-27 Martin Liska <mliska@suse.cz>
10536 PR gcov-profile/82457
10537 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10538 for fork-like functions to be properly instrumented.
10540 2017-10-27 Richard Biener <rguenther@suse.de>
10542 PR middle-end/81659
10543 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10544 info when we redirected EH.
10546 2017-10-26 Michael Collison <michael.collison@arm.com>
10548 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10550 (<optab>_trunchf<GPI:mode>2: New pattern.
10551 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10552 * config/aarch64/iterators.md (wv): New mode attribute.
10553 (vf, VF): New mode attributes.
10554 (vgp, VGP): New mode attributes.
10555 (s): Update attribute with SImode and DImode prefixes.
10557 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10559 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10560 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10561 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10562 (nios2_option_overide): Initialize it. Don't allow R0-relative
10563 addressing with PIC.
10564 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10565 (nios2_symbolic_constant_p): Likewise.
10566 (nios2_legitimate_address_p): Likewise.
10567 (nios2_r0rel_section_name_p): New.
10568 (nios2_symbol_ref_in_r0rel_data_p): New.
10569 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10570 (r0rel_constant_p): New.
10571 (nios2_print_operand_address): Handle r0rel_constant_p.
10572 (nios2_cdx_narrow_form_p): Likewise.
10573 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10574 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10575 (Nios II Options): Document -mr0rel-sec.
10577 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10579 * config/nios2/nios2.c: Include xregex.h.
10580 (nios2_gprel_sec_regex): New.
10581 (nios2_option_overide): Initialize it. Don't allow GP-relative
10582 addressing with PIC.
10583 (nios2_small_section_name_p): Check for regex match.
10584 * config/nios2/nios2.opt (mgprel-sec=): New option.
10585 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10586 (Nios II Options): Document -mgprel-sec.
10588 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10590 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10592 2017-10-26 Tom de Vries <tom@codesourcery.com>
10594 PR tree-optimization/82707
10595 * gimple.c (gimple_copy): Fix unsharing of
10596 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10598 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10600 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10601 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10602 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10603 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10604 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10605 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10606 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10607 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10608 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10609 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10610 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10611 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10612 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10613 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10614 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10615 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10616 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10618 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10620 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10622 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10623 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10624 warning to rs6000.c. Remove the Undocumented flag, since it has
10626 (-mabi=ibmlongdouble): Likewise.
10627 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10628 already set, set the default format for long double.
10629 (rs6000_debug_reg_global): Print whether long double is IBM or
10631 (rs6000_option_override_internal): Rework setting long double
10632 format. Only warn if the user is changing the long double default
10633 and they did not use -Wno-psabi.
10634 * doc/invoke.texi (PowerPC options): Update the documentation for
10635 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10637 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10638 Alan Hayward <alan.hayward@arm.com>
10639 David Sherwood <david.sherwood@arm.com>
10641 * rtl.h (wider_subreg_mode): New function.
10642 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10643 rather than an unsigned int *.
10644 * ira-color.c (regno_max_ref_width): Replace with...
10645 (regno_max_ref_mode): ...this new variable.
10646 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10647 Use wider_subreg_mode.
10648 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10649 rather than an unsigned int *.
10650 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10651 (process_alt_operands): Likewise.
10652 (invariant_p): Likewise.
10653 * lra-spills.c (assign_mem_slot): Likewise.
10654 (add_pseudo_to_slot): Likewise.
10655 * lra.c (collect_non_operand_hard_regs): Likewise.
10656 (add_regs_to_insn_regno_info): Likewise.
10657 * reload1.c (regno_max_ref_width): Replace with...
10658 (regno_max_ref_mode): ...this new variable.
10659 (reload): Update accordingly. Update call to
10660 ira_sort_regnos_for_alter_reg.
10661 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10662 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10663 (scan_paradoxical_subregs): Likewise.
10665 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10667 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10668 (aarch64_frame): Add emit_frame_chain boolean.
10669 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10670 Move eh_return case to aarch64_layout_frame.
10671 (aarch64_layout_frame): Initialize emit_frame_chain.
10672 (aarch64_expand_prologue): Use emit_frame_chain.
10674 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10676 * config/aarch64/aarch64.c (aarch64_layout_frame):
10677 Ensure LR is always stored at the bottom of the callee-saves.
10678 Remove rarely used frame layout which saves callee-saves at top of
10679 frame, so the store of LR can be used as a valid probe in all cases.
10681 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10683 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10684 Improve unaligned TImode/TFmode base/offset split.
10686 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10687 Alan Hayward <alan.hayward@arm.com>
10688 David Sherwood <david.sherwood@arm.com>
10690 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10691 * combine.c (find_single_use_1): Likewise.
10692 (expand_field_assignment): Likewise.
10693 (move_deaths): Likewise.
10694 * lra-constraints.c (simplify_operand_subreg): Likewise.
10695 (curr_insn_transform): Likewise.
10696 * lra.c (collect_non_operand_hard_regs): Likewise.
10697 (add_regs_to_insn_regno_info): Likewise.
10698 * rtlanal.c (reg_referenced_p): Likewise.
10699 (covers_regno_no_parallel_p): Likewise.
10701 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10703 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10704 Don't print any bits outside the precision of the value.
10705 * wide-int.cc (test_printing): Add some new tests.
10707 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10709 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10710 supports -xbrace_comment option.
10711 * configure: Regenerate.
10712 * config.in: Regenerate.
10713 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10714 (ASM_CPU_SPEC): Use it.
10716 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10718 * target.def (static_rtx_alignment): New hook.
10719 * targhooks.h (default_static_rtx_alignment): Declare.
10720 * targhooks.c (default_static_rtx_alignment): New function.
10721 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10722 * doc/tm.texi: Regenerate.
10723 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10724 instead of targetm.constant_alignment. Remove call to
10725 set_mem_attributes.
10726 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10727 (cris_preferred_mininum_alignment): New function, split out from...
10728 (cris_constant_alignment): ...here.
10729 (cris_static_rtx_alignment): New function.
10730 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10732 (ix86_constant_alignment): ...here.
10733 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10734 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10735 (mmix_static_rtx_alignment): New function.
10736 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10737 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10739 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10742 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10743 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10745 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10747 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10748 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10749 mark as a sign-extending load.
10750 (local_pic_load_u): Define.
10752 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10754 PR middle-end/82062
10755 * fold-const.c (operand_equal_for_comparison_p): Also return true
10756 if ARG0 is a simple variant of ARG1 with narrower precision.
10757 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10759 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10761 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10763 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10764 scatter_static, scatter_per_elt.
10765 * x86-tune-costs.h: Add new cost entries.
10767 2017-10-25 Richard Biener <rguenther@suse.de>
10769 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10770 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10771 class pass_fre): Move to ...
10772 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10773 class pass_fre): ... here and adjust for statistics.
10775 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10778 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10779 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10780 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10781 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10784 2017-10-25 Richard Biener <rguenther@suse.de>
10786 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10787 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10788 eliminate_push_avail, eliminate_insert): Move inside...
10789 (class eliminate_dom_walker): ... this class in preparation
10791 (fini_eliminate): Remove by merging with ...
10792 (eliminate): ... this function. Adjust for class changes.
10793 (pass_pre::execute): Remove fini_eliminate call.
10794 (pass_fre::execute): Likewise.
10796 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10799 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10800 (VPERMI2, VPERMI2I): New mode iterators.
10801 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10802 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10804 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10805 mode iterator. Remove 3 old define_insn patterns.
10806 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10807 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10808 VPERMI2 mode iterator, remove the other two expanders.
10809 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10810 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10811 instructions, remove the other two patterns.
10812 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10813 mode iterator, remove the other two patterns.
10814 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10815 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10816 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10817 and adjust argument order accordingly.
10818 (ix86_expand_vec_perm): Adjust caller.
10819 (expand_vec_perm_1): Likewise.
10820 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10821 (expand_vec_perm_vpermt2_vpshub2): ... this.
10822 (ix86_expand_vec_perm_const_1): Adjust caller.
10823 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10826 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10827 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10828 (vec_shl_<mode>): Remove unused expander.
10829 (avx512bw_<shift_insn><mode>3): New define_insn.
10830 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10831 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10833 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10836 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10839 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10841 PR rtl-optimization/82396
10842 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10843 (autopref_multipass_init): Simplify initialization.
10844 (autopref_rank_data): Simplify sort order.
10845 * gcc/sched-int.h (autopref_multipass_data_): Remove
10846 multi_mem_insn_p, min_offset and max_offset.
10848 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10850 PR middle-end/60580
10851 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10852 Check special value of flag_omit_frame_pointer.
10853 (aarch64_can_eliminate): Likewise.
10854 (aarch64_override_options_after_change_1): Simplify handling of
10855 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10857 2017-10-24 Richard Biener <rguenther@suse.de>
10859 PR tree-optimization/82697
10860 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10861 zero for conditional load and unconditional store.
10863 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10865 * doc/install.texi: Document bootstrap-cet.
10867 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10870 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10871 ENDBR instruction at function entrance if function is only
10874 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10877 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10878 patterns to better describe from which operation the CF is computed.
10879 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10880 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10881 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10882 is 0, use _0 suffixed expanders instead of emitting a comparison
10885 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10887 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10888 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10890 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10892 PR middle-end/82569
10893 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10894 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10895 * loop-iv.c (iv_get_reaching_def): Likewise.
10896 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10897 variable is promoted and the partition contains undefined values.
10899 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10901 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10903 (nios2_address_cost): Define.
10904 (nios2_legitimize_address): Recognize (exp + constant) directly.
10905 (TARGET_ADDRESS_COST): Define.
10907 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10909 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10910 (nios2_symbolic_memory_operand_p): Declare.
10911 (nios2_split_large_constant): Declare.
10912 (nios2_split_symbolic_memory_operand): Declare.
10913 * config/nios2/nios2.c: Adjust includes.
10914 (nios2_symbolic_constant_allowed): New.
10915 (nios2_symbolic_constant_p): New.
10916 (nios2_plus_symbolic_constant_p): New.
10917 (nios2_valid_addr_expr_p): Recognize addresses involving
10918 symbolic constants.
10919 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10920 (nios2_symbolic_memory_operand_p): New.
10921 (nios2_large_constant_p): New.
10922 (nios2_split_large_constant): New.
10923 (nios2_split_plus_large_constant): New.
10924 (nios2_split_symbolic_memory_operand): New.
10925 (nios2_legitimize_address): Code refactoring. Handle addresses
10926 involving symbolic constants.
10927 (nios2_emit_move_sequence): Likewise.
10928 (nios2_print_operand): Improve error output.
10929 (nios2_print_operand_address): Handle LO_SUM.
10930 (nios2_cdx_narrow_form_p): Likewise.
10931 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10932 operands involving symbolic constants.
10933 (movhi_internal, movsi_internal): Likewise.
10934 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10935 (extendhisi2, extendqi<mode>2): Likewise.
10937 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10939 * tree-pass.h (PROP_rtl_split_insns): Define.
10940 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10942 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10944 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10946 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10949 * target.def (const_not_ok_for_debug_p): Default to
10950 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10951 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10952 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10953 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10954 which targetm.const_not_ok_for_debug_p returned true.
10955 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10957 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10959 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10960 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10961 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10963 (ix86_const_not_ok_for_debug_p): New function.
10964 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10965 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10967 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10970 * system.h: Conditionally include "unique-ptr.h" if
10971 INCLUDE_UNIQUE_PTR is defined.
10972 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10973 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10975 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10977 * config/rl78/rl78.md: New define_expand "subdi3".
10979 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10982 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10983 DF_REF_INSN if DF_REF_INSN_INFO is false.
10985 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10987 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10988 xmm_move instead of sse_move.
10989 (sse_store_index): New function.
10990 (ix86_register_move_cost): Be more sensible about mismatch stall;
10991 model AVX moves correctly; make difference between sse->integer and
10993 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10994 moves; make difference between SSE and AVX.
10995 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10996 and zmm_move. Increase size of sse load and store tables;
10997 add unaligned load and store tables; add ssemmx_to_integer.
10998 * x86-tune-costs.h: Update all entries according to real
10999 move latencies from Agner Fog's manual and chip documentation.
11001 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11004 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
11005 * config/i386/constraints.md (Wf): New constraint.
11006 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11007 (cmp<dwi>_doubleword): Removed.
11008 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11009 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11010 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11011 expand with cmp<dwi>_doubleword. For LTU and GEU use
11012 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11014 * common.opt (gcolumn-info): Enable by default.
11015 * doc/invoke.texi (gcolumn-info): Document new default.
11017 2017-10-23 Richard Biener <rguenther@suse.de>
11019 PR tree-optimization/82672
11020 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11021 Fold the stmt if we propagated into it.
11023 2017-10-23 Richard Biener <rguenther@suse.de>
11025 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11026 (bitmap_remove_expr_from_set): ... this. All callers call this
11027 for non-constant values.
11028 (bitmap_set_subtract): Rename to...
11029 (bitmap_set_subtract_expressions): ... this. Adjust and
11031 (bitmap_set_contains_value): Remove superfluous check.
11032 (bitmap_set_replace_value): Inline into single caller ...
11033 (bitmap_value_replace_in_set): ... here and simplify.
11034 (dependent_clean): Merge into ...
11035 (clean): ... this using an overload. Adjust.
11036 (prune_clobbered_mems): Adjust.
11037 (compute_antic_aux): Likewise.
11038 (compute_partial_antic_aux): Likewise.
11040 2017-10-23 Richard Biener <rguenther@suse.de>
11042 PR tree-optimization/82129
11044 2017-08-01 Richard Biener <rguenther@suse.de>
11046 PR tree-optimization/81181
11047 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11048 (compute_antic): ... end of iteration here.
11050 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11052 * target.def (starting_frame_offset): New hook.
11053 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11054 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11055 * doc/tm.texi.in: Regenerate.
11056 * hooks.h (hook_hwi_void_0): Declare.
11057 * hooks.c (hook_hwi_void_0): New function.
11058 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11059 STARTING_FRAME_OFFSET.
11060 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11061 * reload1.c (reload): Likewise.
11062 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11063 instead of STARTING_FRAME_OFFSET.
11064 * function.c (try_fit_stack_local): Likewise.
11065 (assign_stack_local_1): Likewise
11066 (instantiate_virtual_regs): Likewise.
11067 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11068 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11069 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11070 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11071 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11072 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11073 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11074 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11075 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11076 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11077 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11078 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11079 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11080 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11081 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11082 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11083 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11084 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11085 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11086 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11087 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11088 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11089 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11090 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11091 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11092 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11093 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11094 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11095 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11096 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11097 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11098 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11099 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11100 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11101 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11102 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11103 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11104 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11105 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11106 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11107 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11108 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11109 return a HOST_WIDE_INT.
11110 (avr_builtin_setjmp_frame_value): Use it instead of
11111 STARTING_FRAME_OFFSET.
11112 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11113 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11114 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11116 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11117 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11118 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11119 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11120 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11121 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11122 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11123 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11124 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11125 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11126 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11127 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11129 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11130 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11131 * config/mips/mips.c (mips_compute_frame_info): Refer to
11132 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11133 (mips_starting_frame_offset): New function.
11134 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11135 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11136 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11137 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11138 and return a HOST_WIDE_INT.
11139 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11140 (mmix_initial_elimination_offset): Refer to
11141 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11142 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11143 * config/pa/pa.c (pa_starting_frame_offset): New function.
11144 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11145 (pa_expand_prologue): Likewise.
11146 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11147 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11148 !FRAME_GROWS_DOWNWARD handling to...
11149 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11150 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11151 !FRAME_GROWS_DOWNWARD handling to...
11152 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11153 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11154 !FRAME_GROWS_DOWNWARD handling to...
11155 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11156 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11158 (rs6000_starting_frame_offset): New function.
11159 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11160 !FRAME_GROWS_DOWNWARD handling to...
11161 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11162 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11163 !FRAME_GROWS_DOWNWARD handling to...
11164 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11165 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11166 !FRAME_GROWS_DOWNWARD handling to...
11167 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11168 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11169 (rs6000_starting_frame_offset): New function.
11170 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11171 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11172 * config/vax/vax.c (vax_starting_frame_offset): New function.
11173 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11174 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11175 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11176 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11177 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11178 * system.h (STARTING_FRAME_OFFSET): Poison.
11180 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11182 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11183 SCALAR_TYPE_MODE instead of TYPE_MODE.
11185 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11186 Alan Hayward <alan.hayward@arm.com>
11187 David Sherwood <david.sherwood@arm.com>
11189 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11191 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11192 Alan Hayward <alan.hayward@arm.com>
11193 David Sherwood <david.sherwood@arm.com>
11195 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11197 2017-10-23 Richard Biener <rguenther@suse.de>
11199 PR tree-optimization/82129
11200 * tree-ssa-pre.c (bitmap_set_and): Remove.
11201 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11202 canonicalizing expressions in the set to those with lowest
11203 ID rather than taking that from the first edge.
11205 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11207 * combine.c (rtx_equal_for_field_assignment_p): Use
11208 byte_lowpart_offset.
11210 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11211 Alan Hayward <alan.hayward@arm.com>
11212 David Sherwood <david.sherwood@arm.com>
11214 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11215 to a SUBREG_PROMOTED_VAR.
11217 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11218 Alan Hayward <alan.hayward@arm.com>
11219 David Sherwood <david.sherwood@arm.com>
11221 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11222 (expand_debug_source_expr): Likewise.
11223 * combine.c (combine_simplify_rtx): Likewise.
11224 * cse.c (fold_rtx): Likewise.
11225 * optabs.c (expand_float): Likewise.
11226 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11227 (simplify_binary_operation_1): Likewise.
11229 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11230 Alan Hayward <alan.hayward@arm.com>
11231 David Sherwood <david.sherwood@arm.com>
11233 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11234 (record_promoted_value): Likewise.
11235 * expr.c (expand_expr_real_2): Likewise.
11236 * ree.c (update_reg_equal_equiv_notes): Likewise.
11237 (combine_set_extension): Likewise.
11238 * rtlanal.c (low_bitmask_len): Likewise.
11239 * simplify-rtx.c (neg_const_int): Likewise.
11240 (simplify_binary_operation_1): Likewise.
11242 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11243 Alan Hayward <alan.hayward@arm.com>
11244 David Sherwood <david.sherwood@arm.com>
11246 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11247 * regcprop.c (maybe_mode_change): Likewise.
11248 * reload1.c (alter_reg): Likewise.
11250 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11252 * inchash.h (inchash::hash::add_wide_int): New function.
11253 * lto-streamer-out.c (hash_tree): Use it.
11255 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11257 * inchash.h (inchash::hash::add_wide_int): Rename to...
11258 (inchash::hash::add_hwi): ...this.
11259 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11260 (polymorphic_call_target_hasher::hash): Likewise.
11261 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11262 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11263 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11264 * lto-streamer-out.c (hash_tree): Likewise.
11265 * optc-save-gen.awk: Likewise.
11266 * tree.c (add_expr): Likewise.
11268 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11271 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11272 for ordered inequality comparisons even with TARGET_IEEE_FP.
11274 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11277 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11278 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11279 Expand with cmp<dwi>_doubleword.
11281 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11283 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11284 List CET intrinsics.
11285 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11286 specific to -fcf-protection option.
11288 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11290 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11291 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11292 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11293 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11294 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11295 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11296 (extra_objs): Add cet.o for Linux/x86 targets.
11297 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11298 * config/i386/cet.c: New file.
11299 * config/i386/cetintrin.h: Likewise.
11300 * config/i386/t-cet: Likewise.
11301 * config/i386/cpuid.h (bit_SHSTK): New.
11302 (bit_IBT): Likewise.
11303 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11304 pass IBT and SHSTK bits.
11305 * config/i386/i386-builtin-types.def
11306 (VOID_FTYPE_UNSIGNED_PVOID): New.
11307 (VOID_FTYPE_UINT64_PVOID): Likewise.
11308 * config/i386/i386-builtin.def: Add CET intrinsics.
11309 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11310 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11311 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11312 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11314 * config/i386/i386.c (rest_of_insert_endbranch): New.
11315 (pass_data_insert_endbranch): Likewise.
11316 (pass_insert_endbranch): Likewise.
11317 (make_pass_insert_endbranch): Likewise.
11318 (ix86_notrack_prefixed_insn_p): Likewise.
11319 (ix86_target_string): Add -mibt, -mshstk flags.
11320 (ix86_option_override_internal): Add flag_cf_protection
11322 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11323 (ix86_print_operand): Add 'notrack' prefix output.
11324 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11325 (ix86_expand_builtin): Expand CET intrinsics.
11326 (x86_output_mi_thunk): Add 'endbranch' instruction.
11327 * config/i386/i386.h (TARGET_IBT): New.
11328 (TARGET_IBT_P): Likewise.
11329 (TARGET_SHSTK): Likewise.
11330 (TARGET_SHSTK_P): Likewise.
11331 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11332 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11333 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11334 (builtin_setjmp_setup): New pattern.
11335 (builtin_longjmp): Likewise.
11336 (rdssp<mode>): Likewise.
11337 (incssp<mode>): Likewise.
11338 (saveprevssp): Likewise.
11339 (rstorssp): Likewise.
11340 (wrss<mode>): Likewise.
11341 (wruss<mode>): Likewise.
11342 (setssbsy): Likewise.
11343 (clrssbsy): Likewise.
11344 (nop_endbr): Likewise.
11345 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11347 * config/i386/immintrin.h: Include <cetintrin.h>.
11348 * config/i386/linux-common.h
11349 (file_end_indicate_exec_stack_and_cet): New prototype.
11350 (TARGET_ASM_FILE_END): New.
11352 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11354 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11355 latencies instead of having separate table; make difference between
11356 integer and float costs.
11357 * i386.h (processor_costs): Remove scalar_stmt_cost,
11358 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11359 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11361 * x86-tune-costs.h: Remove entries which has been removed in
11362 procesor_costs from all tables; make cond_taken_branch_cost
11363 and cond_not_taken_branch_cost COST_N_INSNS based.
11365 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11367 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11369 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11371 * config/i386/i386.md (isa): Remove fma_avx512f.
11372 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11373 <avx512>_fmadd_<mode>_mask3<round_name>,
11374 <avx512>_fmsub_<mode>_mask<round_name>,
11375 <avx512>_fmsub_<mode>_mask3<round_name>,
11376 <avx512>_fnmadd_<mode>_mask<round_name>,
11377 <avx512>_fnmadd_<mode>_mask3<round_name>,
11378 <avx512>_fnmsub_<mode>_mask<round_name>,
11379 <avx512>_fnmsub_<mode>_mask3<round_name>,
11380 <avx512>_fmaddsub_<mode>_mask<round_name>,
11381 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11382 <avx512>_fmsubadd_<mode>_mask<round_name>,
11383 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11384 (*vec_widen_umult_even_v16si<mask_name>,
11385 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11386 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11388 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11390 * extend.texi: Add 'nocf_check' documentation.
11391 * gimple.texi: Add second parameter to
11392 gimple_build_call_from_tree.
11393 * invoke.texi: Add -fcf-protection documentation.
11394 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11396 2017-10-20 Richard Biener <rguenther@suse.de>
11398 PR tree-optimization/82473
11399 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11400 the largest input type.
11402 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11404 * c-attribs.c (handle_nocf_check_attribute): New function.
11405 (c_common_attribute_table): Add 'nocf_check' handling.
11406 * gimple-parser.c: Add second argument NULL to
11407 gimple_build_call_from_tree.
11408 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11409 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11411 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11412 * common.opt: Add fcf-protection flag.
11413 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11414 * flag-types.h: Add enum cf_protection_level.
11415 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11416 Add 'nocf_check' attribute propagation to gimple call.
11417 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11418 (gimple_build_call_from_tree): Update prototype.
11419 (gimple_call_nocf_check_p): New function.
11420 (gimple_call_set_nocf_check): Likewise.
11421 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11422 * ipa-icf.c: Add nocf_check attribute in statement hash.
11423 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11424 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11425 * toplev.c (process_options): Add flag_cf_protection handling.
11427 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11429 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11431 2017-10-20 Richard Biener <rguenther@suse.de>
11433 PR tree-optimization/82603
11434 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11435 remove false predicated stores.
11437 2017-10-20 Richard Biener <rguenther@suse.de>
11439 * graphite-isl-ast-to-gimple.c
11440 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11441 Remove return value and simplify, dump copied stmt after lhs
11443 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11444 Reduce dump verbosity.
11445 (gsi_insert_earliest): Likewise.
11446 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11447 * graphite.c (print_global_statistics): Adjust dumping.
11448 (print_graphite_scop_statistics): Likewise.
11449 (print_graphite_statistics): Do not dump loops here.
11450 (graphite_transform_loops): But here.
11452 2017-10-20 Nicolas Roche <roche@adacore.com>
11454 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11455 * configure: Regenerate.
11457 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11460 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11461 functions when optimizing replace GIMPLE_RETURN stmts with
11462 calls to __builtin_unreachable ().
11465 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11466 for -fsanitize=thread link of executables.
11467 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11468 link of executables.
11471 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11472 New mode iterators.
11473 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11474 define_insns for logical vector shifts to use VI248_AVX512BW
11475 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11476 condition, useless isa and prefix attributes. Change the first
11477 2 of these define_insns to ...
11478 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11479 define_insn for avx512vl.
11480 (<shift_insn><mode>3): ... and this, new define_insn without
11481 masking for non-avx512vl.
11484 * config/i386/sse.md (*andnot<mode>3,
11485 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11486 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11487 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11488 not applied use empty suffix even for TARGET_AVX512VL.
11489 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11490 is applied, supply evex,evex or evex,evex,evex instead of just
11493 2017-10-20 Julia Koval <julia.koval@intel.com>
11495 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11496 (OPTION_MASK_ISA_GFNI_UNSET): New.
11497 (ix86_handle_option): Handle OPT_mgfni.
11498 * config/i386/cpuid.h (bit_GFNI): New.
11499 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11500 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11501 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11502 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11503 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11504 * config/i386/i386.opt: Add mgfni.
11506 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11508 * config/msp430/msp430.c (msp430_option_override): Disable
11509 -fdelete-null-pointer-checks.
11510 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11512 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11514 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11515 of x87 and SSE instructions.
11517 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11519 * asan.c (create_cond_insert_point): Do not update edge count.
11520 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11521 (afdo_propagate_circuit): Likewise.
11522 (afdo_calculate_branch_prob): Likewise.
11523 (afdo_annotate_cfg): Likewise.
11524 * basic-block.h (struct edge_def): Remove count.
11525 (edge_def::count): New accessor.
11526 * bb-reorder.c (rotate_loop): Update.
11527 (find_traces_1_round): Update.
11528 (connect_traces): Update.
11529 (sanitize_hot_paths): Update.
11530 * cfg.c (unchecked_make_edge): Update.
11531 (make_single_succ_edge): Update.
11532 (check_bb_profile): Update.
11533 (dump_edge_info): Update.
11534 (update_bb_profile_for_threading): Update.
11535 (scale_bbs_frequencies_int): Update.
11536 (scale_bbs_frequencies_gcov_type): Update.
11537 (scale_bbs_frequencies_profile_count): Update.
11538 (scale_bbs_frequencies): Update.
11539 * cfganal.c (connect_infinite_loops_to_exit): Update.
11540 * cfgbuild.c (compute_outgoing_frequencies): Update.
11541 (find_many_sub_basic_blocks): Update.
11542 * cfgcleanup.c (try_forward_edges): Update.
11543 (try_crossjump_to_edge): Update
11544 * cfgexpand.c (expand_gimple_cond): Update
11545 (expand_gimple_tailcall): Update
11546 (construct_exit_block): Update
11547 * cfghooks.c (verify_flow_info): Update
11548 (redirect_edge_succ_nodup): Update
11549 (split_edge): Update
11550 (make_forwarder_block): Update
11551 (duplicate_block): Update
11552 (account_profile_record): Update
11553 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11554 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11555 * cfgloopmanip.c (scale_loop_profile): Update.
11557 (lv_adjust_loop_entry_edge): Update.
11558 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11559 (force_nonfallthru_and_redirect): Update.
11560 (purge_dead_edges): Update.
11561 (rtl_flow_call_edges_add): Update.
11562 * cgraphunit.c (init_lowered_empty_function): Update.
11563 (cgraph_node::expand_thunk): Update.
11564 * gimple-pretty-print.c (dump_probability): Update.
11565 (dump_edge_probability): Update.
11566 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11567 * haifa-sched.c (sched_create_recovery_edges): Update.
11568 * hsa-gen.c (convert_switch_statements): Update.
11569 * ifcvt.c (dead_or_predicable): Update.
11570 * ipa-inline-transform.c (inline_transform): Update.
11571 * ipa-split.c (split_function): Update.
11572 * ipa-utils.c (ipa_merge_profiles): Update.
11573 * loop-doloop.c (add_test): Update.
11574 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11575 * lto-streamer-in.c (input_cfg): Update.
11576 (input_function): Update.
11577 * lto-streamer-out.c (output_cfg): Update.
11578 * modulo-sched.c (sms_schedule): Update.
11579 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11580 * predict.c (maybe_hot_edge_p): Update.
11581 (unlikely_executed_edge_p): Update.
11582 (probably_never_executed_edge_p): Update.
11583 (dump_prediction): Update.
11584 (drop_profile): Update.
11585 (propagate_unlikely_bbs_forward): Update.
11586 (determine_unlikely_bbs): Update.
11587 (force_edge_cold): Update.
11588 * profile.c (compute_branch_probabilities): Update.
11589 * reg-stack.c (better_edge): Update.
11590 * shrink-wrap.c (handle_simple_exit): Update.
11591 * tracer.c (better_p): Update.
11592 * trans-mem.c (expand_transaction): Update.
11593 (split_bb_make_tm_edge): Update.
11594 * tree-call-cdce.c: Update.
11595 * tree-cfg.c (gimple_find_sub_bbs): Update.
11596 (gimple_split_edge): Update.
11597 (gimple_duplicate_sese_region): Update.
11598 (gimple_duplicate_sese_tail): Update.
11599 (gimple_flow_call_edges_add): Update.
11600 (insert_cond_bb): Update.
11601 (execute_fixup_cfg): Update.
11602 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11603 * tree-complex.c (expand_complex_div_wide): Update.
11604 * tree-eh.c (lower_resx): Update.
11605 (unsplit_eh): Update.
11606 (cleanup_empty_eh_move_lp): Update.
11607 * tree-inline.c (copy_edges_for_bb): Update.
11608 (freqs_to_counts): Update.
11609 (copy_cfg_body): Update.
11610 * tree-ssa-dce.c (remove_dead_stmt): Update.
11611 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11612 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11613 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11614 (unloop_loops): Update.
11615 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11616 * tree-ssa-loop-split.c (connect_loops): Update.
11617 (split_loop): Update.
11618 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11619 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11620 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11621 * tree-ssa-reassoc.c (branch_fixup): Update.
11622 * tree-ssa-tail-merge.c (replace_block_by): Update.
11623 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11624 (compute_path_counts): Update.
11625 (update_profile): Update.
11626 (recompute_probabilities): Update.
11627 (update_joiner_offpath_counts): Update.
11628 (estimated_freqs_path): Update.
11629 (freqs_to_counts_path): Update.
11630 (clear_counts_path): Update.
11631 (ssa_fix_duplicate_block_edges): Update.
11632 (duplicate_thread_path): Update.
11633 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11634 (case_bit_test_cmp): Update.
11635 (collect_switch_conv_info): Update.
11636 (gen_inbound_check): Update.
11637 (do_jump_if_equal): Update.
11638 (emit_cmp_and_jump_insns): Update.
11639 * tree-tailcall.c (decrease_profile): Update.
11640 (eliminate_tail_call): Update.
11641 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11642 (vect_do_peeling): Update.
11643 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11644 * ubsan.c (ubsan_expand_null_ifn): Update.
11645 (ubsan_expand_ptr_ifn): Update.
11646 * value-prof.c (gimple_divmod_fixed_value): Update.
11647 (gimple_mod_pow2): Update.
11648 (gimple_mod_subtract): Update.
11649 (gimple_ic): Update.
11650 (gimple_stringop_fixed_value): Update.
11652 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11655 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11657 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11659 PR rtl-optimization/82395
11660 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11661 based on non_spilled_static_chain_regno_p.
11663 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11665 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11666 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11667 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11669 2017-10-19 Martin Sebor <msebor@redhat.com>
11671 PR tree-optimization/82596
11672 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11674 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11676 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11677 (get_mem_refs_of_builtin_call): Likewise.
11678 * builtins.c (expand_builtin_apply): Adjust call to
11679 allocate_dynamic_stack_space.
11680 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11681 the third argument to allocate_dynamic_stack_space, otherwise -1.
11682 (expand_builtin): Deal with all alloca variants.
11683 (is_inexpensive_builtin): Likewise.
11684 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11685 * calls.c (special_function_p): Deal with all alloca variants.
11686 (initialize_argument_information): Adjust call to
11687 allocate_dynamic_stack_space.
11688 (expand_call): Likewise.
11689 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11690 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11691 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11692 use it for the stack usage computation.
11693 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11694 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11695 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11696 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11697 (in_loop_p): Remove first argument and useless check.
11698 (pass_walloca::execute): Remove useless test and adjust call to above.
11699 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11700 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11701 (gimplify_call_expr): Deal with all alloca variants.
11702 * hsa-gen.c (gen_hsa_alloca): Likewise.
11703 (gen_hsa_insns_for_call): Likewise.
11704 * ipa-pure-const.c (special_builtin_state): Likewise.
11705 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11706 * tree-object-size.c (alloc_object_size): Likewise.
11707 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11708 (call_may_clobber_ref_p_1): Likewise.
11709 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11710 (ccp_fold_stmt): Likewise.
11711 (optimize_stack_restore): Likewise.
11712 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11713 (mark_all_reaching_defs_necessary_1): Likewise.
11714 (propagate_necessity): Likewise.
11715 (eliminate_unnecessary_stmts): Likewise.
11716 * tree.c (build_common_builtin_nodes): Build
11717 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11718 (build_alloca_call_expr): New function.
11719 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11720 (CASE_BUILT_IN_ALLOCA): Likewise.
11721 (build_alloca_call_expr): Declare.
11722 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11724 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11727 * dwarf2out.c (new_die_raw): New static inline function.
11728 (new_die): Use it to create the DIE.
11729 (add_AT_external_die_ref): Likewise.
11730 (clone_die): Likewise.
11731 (clone_as_declaration): Likewise.
11732 (dwarf2out_vms_debug_main_pointer): Likewise.
11733 (base_type_die): Likewise. Remove early return for corner cases.
11734 Do not call add_pubtype on the DIE here.
11735 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11736 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11737 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11738 native order exists for base types, attach the DIE manually and call
11739 add_pubtype on it. Do not equate a reverse order DIE to the type.
11741 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11743 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11744 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11746 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11748 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11750 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11751 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11752 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11753 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11754 * opts.c (sanitizer_opts): Add builtin.
11755 * ubsan.c (instrument_builtin): New function.
11756 (pass_ubsan::execute): Call it.
11757 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11758 * doc/invoke.texi: Document -fsanitize=builtin.
11760 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11761 builtins, store max (log2 (align), 0) into uchar field instead of
11762 align into uptr field.
11763 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11764 store uchar 0 field instead of uptr 0 field.
11765 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11766 instead of passing one address of struct with 2 locations pass
11767 two addresses of structs with 1 location each.
11768 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11769 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11770 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11771 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11772 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11773 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11774 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11775 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11777 2017-10-19 Martin Liska <mliska@suse.cz>
11780 * file-find.c (remove_prefix): Remove.
11781 * file-find.h (remove_prefix): Likewise.
11782 * gcc-ar.c: Remove smartness of lookup.
11784 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11786 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11787 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11788 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11790 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11793 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11794 (setcc + and to xor + setcc): New peephole2.
11796 2017-10-19 Tom de Vries <tom@codesourcery.com>
11798 * doc/sourcebuild.texi (Test Directives, Variants of
11799 dg-require-support): Add dg-require-stack-size.
11801 2017-10-19 Martin Liska <mliska@suse.cz>
11804 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11805 that have a large alignment.
11806 (gimplify_target_expr): Likewise.
11808 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11810 PR rtl-optimization/82602
11811 * ira.c (rtx_moveable_p): Return false for volatile asm.
11813 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11816 * config/i386/i386-modes.def (CCGZ): New CC mode.
11817 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11818 * config/i386/predicates.md (ix86_comparison_operator):
11820 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11821 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11822 with double-word subtraction.
11823 (put_condition_code): Handle CCGZmode.
11825 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11827 * wide-int.cc (debug (const wide_int &)): New.
11828 (debug (const wide_int *)): New.
11829 (debug (const widest_int &)): New.
11830 (debug (const widest_int *)): New.
11832 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11834 PR middle-end/82556
11835 * lra-constraints.c (curr_insn_transform): Use non-input operand
11836 instead of output one for matched reload.
11838 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11840 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11841 (tree-ssa-loop-ivopts.h): New header file.
11842 (struct builtin_info): New fields.
11843 (classify_builtin_1): Compute and record base and offset parts for
11844 memset builtin partition by calling strip_offset.
11845 (offset_cmp, fuse_memset_builtins): New functions.
11846 (finalize_partitions): Fuse adjacent memset partitions by calling
11848 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11849 Expose the interface.
11850 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11852 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11854 PR tree-optimization/82574
11855 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11856 that data reference must be executed exactly once per iteration
11857 against the outermost loop in nest.
11858 (classify_partition): Update call to above function.
11860 2017-10-18 Richard Biener <rguenther@suse.de>
11862 PR tree-optimization/82591
11863 * graphite.c (graphite_transform_loops): Move code gen message
11865 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11866 Here. Handle scop_to_isl_ast failing.
11867 (scop_to_isl_ast): Limit the number of ISL operations.
11869 2017-10-18 Richard Biener <rguenther@suse.de>
11871 * graphite-isl-ast-to-gimple.c
11872 (translate_isl_ast_to_gimple::set_rename): Simplify.
11873 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11874 (graphite_copy_stmts_from_block): ... here.
11875 (copy_bb_and_scalar_dependences): Simplify.
11876 (add_parameters_to_ivs_params): Canonicalize.
11877 (generate_entry_out_of_ssa_copies): Simplify.
11878 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11879 by passing in ISL dimension.
11880 (parameter_index_in_region_1): Rename to ...
11881 (parameter_index_in_region): ... this.
11882 (extract_affine): Adjust assert, pass down parameter index.
11883 (add_param_constraints): Use range-info when available.
11884 (build_scop_context): Adjust.
11885 * sese.c (new_sese_info): Adjust.
11886 (free_sese_info): Likewise.
11887 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11888 Remove unused typedefs.
11889 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11891 2017-10-18 Martin Liska <mliska@suse.cz>
11893 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11895 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11897 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11898 (s390_sched_init): Do not reset s390_sched_state if we entered the
11899 current basic block via a fallthru edge and all others are unlikely.
11901 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11903 * config/s390/s390.c (NUM_SIDES): New variable.
11904 (LONGRUNNING_THRESHOLD): New variable.
11905 (LATENCY_FACTOR): New variable.
11906 (s390_sched_score): Decrease score for long-running instructions on
11908 (s390_sched_variable_issue): Perform bookkeeping for long-running
11911 2017-10-18 Richard Biener <rguenther@suse.de>
11913 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11914 Simplify with removal of the parameter rename map.
11915 (set_rename): Likewise.
11916 (should_copy_to_new_region): Likewise.
11917 (graphite_copy_stmts_from_block): Likewise.
11918 (copy_bb_and_scalar_dependences): Remove initialization of
11919 unused copied_bb_map.
11920 (copy_def): Remove.
11921 (copy_internal_parameters): Likewise.
11922 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11923 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11924 Use INTEGRAL_TYPE_P.
11925 (parameter_index_in_region_1): Rename to ...
11926 (assign_parameter_index_in_region): ... this. Assert we have
11927 a parameter we handle.
11928 (scan_tree_for_params): Adjust.
11929 * sese.h (parameter_rename_map_t): Remove.
11930 (struct sese_info_t): Remove unused parameter_rename_map and
11931 copied_bb_map members.
11932 * sese.c (new_sese_info): Adjust.
11933 (free_sese_info): Likewise.
11935 2017-10-18 Martin Liska <mliska@suse.cz>
11938 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11939 on an abnormal edge.
11941 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11943 * doc/invoke.texi (ffunction-sections and fdata-sections):
11946 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11948 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11949 the use statement can throw internally.
11951 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11953 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11954 any RTX present on the RHS of a SET.
11955 * compare-elim.c (try_eliminate_compare): Restore comment.
11957 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11959 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11960 may be also called on tcc_type nodes.
11961 * langhooks.c (lhd_tree_size): Likewise.
11963 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11965 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11966 format_warning_at_substring.
11967 (maybe_warn): Convert source_range * param to a location_t. Pass
11968 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11969 (format_directive): Remove code to extract source_ranges and
11970 source_range * in favor of just a location_t.
11971 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11973 * substring-locations.c (format_warning_va): Convert
11974 source_range * param to a location_t.
11975 (format_warning_at_substring): Likewise.
11976 * substring-locations.h (format_warning_va): Likewise.
11977 (format_warning_at_substring): Likewise.
11979 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11981 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11983 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11984 and scatter/gather ops.
11986 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11987 vec_gather_load and vec_scatter_store.
11988 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11989 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11990 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11991 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11992 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11993 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11995 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11997 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11998 Detect FTST instruction and handle its register pops. Only pop
11999 second operand if can_pop_second_op is true.
12000 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
12001 set can_pop_second_op to false in the compare_for_stack_reg call.
12003 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
12004 output_fp_compare for stack register operands.
12005 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
12006 instructions here. Do not emit stack register pops here. Assert
12007 that FCOMPP pops next to top stack register. Rewrite function.
12009 2017-10-17 Nathan Sidwell <nathan@acm.org>
12011 PR middle-end/82577
12012 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12013 use DECL_ASSEMBLER_NAME_RAW.
12015 PR middle-end/82546
12016 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12019 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12020 Wilco Dijkstra <wilco.dijkstra@arm.com>
12022 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12023 converstion to Pmode from the buf_addr.
12025 2017-10-17 Richard Biener <rguenther@suse.de>
12027 * graphite-dependences.c (scop_get_reads_and_writes): Change
12028 output parameters to references.
12030 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12032 PR 71026/tree-optimization
12033 * fold-const.c (distribute_real_division): Removed.
12034 (fold_binary_loc): Remove calls to distribute_real_divison.
12036 2017-10-17 Richard Biener <rguenther@suse.de>
12038 * graphite-scop-detection.c
12039 (scop_detection::stmt_has_simple_data_refs_p): Always use
12040 the full nest as region.
12041 (try_generate_gimple_bb): Likewise.
12042 * sese.c (scalar_evolution_in_region): Simplify now that
12043 SCEV can handle instantiation in regions.
12044 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12045 in the non-loop part of a function if requested.
12047 2017-10-17 Richard Biener <rguenther@suse.de>
12049 PR tree-optimization/82563
12050 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12052 (graphite_regenerate_ast_isl): Call it.
12053 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12055 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12057 PR tree-optimization/82549
12058 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12059 Formatting fixes. Instead of calling make_bit_field_ref with negative
12062 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12064 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12065 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12066 _mm_maskz_reduce_ss): New.
12067 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12068 __builtin_ia32_reducess_mask): Ditto..
12069 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12070 * config/i386/sse.md (reduces<mode>): Renamed to ...
12071 (reduces<mode><mask_scalar_name>): ... this.
12072 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12074 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12075 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12077 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12079 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12080 * selftest-run-tests.c (selftest::run_tests): Call
12081 selftest::unique_ptr_tests_cc_tests.
12082 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12083 * unique-ptr-tests.cc: New file.
12085 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12088 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12090 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12091 (make_hard_regno_born, make_hard_regno_dead): Update
12092 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12094 2017-10-16 Jeff Law <law@redhat.com>
12096 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12098 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12100 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12102 2017-10-16 Olivier Hainque <hainque@adacore.com>
12104 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12105 with_cpu if we were configured for an e500v2 target cpu name.
12107 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12109 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12110 * doc/invoke.texi: Document +nodsp as a valid extension for
12113 2017-10-16 Martin Liska <mliska@suse.cz>
12115 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12116 (test_set_range): Likewise.
12117 (test_range_functions): Rename to ...
12118 (test_bit_in_range): ... this.
12119 (sbitmap_c_tests): Add new test.
12121 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12123 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12125 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12126 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12128 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12130 * config/aarch64/aarch64-builtins.c
12131 (aarch64_types_quadopu_lane_qualifiers): New.
12132 (TYPES_QUADOPU_LANE): New.
12133 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12134 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12135 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12136 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12137 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12138 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12139 (Vdottype, DOTPROD): New.
12140 (sur): Add SDOT and UDOT.
12142 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12144 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12145 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12146 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12147 Add TARGET_DOTPROD.
12148 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12149 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12150 Enable TARGET_DOTPROD.
12151 (cortex-a75.cortex-a55): Likewise.
12152 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12154 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12156 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12157 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12159 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12161 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12162 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12163 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12164 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12165 * config/arm/types.md (neon_dot, neon_dot_q): New.
12166 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12168 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12170 * config/arm/arm.h (TARGET_DOTPROD): New.
12171 * config/arm/arm.c (arm_arch_dotprod): New.
12172 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12173 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12174 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12175 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12176 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12177 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12178 * doc/invoke.texi (armv8.2-a): Document dotprod
12180 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12182 * i386.c (ix86_vec_cost): New function.
12183 (ix86_rtx_costs): Handle vector operations better.
12184 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12185 * x86-tune-costs.h: Add new costs to all tables.
12187 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12189 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12191 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12192 divsd, sqrtss and sqrtsd
12193 * x86-tune-costs.h: Add new entries to all costs.
12194 (znver1_cost): Fix to match real instruction latencies.
12196 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12197 Michael Collison <michael.collison@arm.com>
12199 * compare-elim.c: Include emit-rtl.h.
12200 (can_merge_compare_into_arith): New function.
12201 (try_validate_parallel): Likewise.
12202 (try_merge_compare): Likewise.
12203 (try_eliminate_compare): Call the above when no previous clobber
12205 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12208 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12210 PR middle-end/62263
12211 PR middle-end/82498
12212 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12213 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12215 PR middle-end/62263
12216 PR middle-end/82498
12217 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12218 to be any operand_equal_p operands. For & (B - 1) require
12219 B to be power of 2. Recognize
12220 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12222 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12225 * optabs.c (expand_memory_blockage): Fix call of
12226 targetm.have_memory_blockage.
12228 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12231 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12232 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12233 objects to extra_objs instead of overwriting it.
12235 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12237 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12238 Use any_fp_register_operand as operand[3] predicate. Simplify
12239 equality test for operands[2] and operands[4] memory location.
12240 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12241 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12242 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12243 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12244 any_fp_register_operand as operand[1] predicate. Simplify
12245 equality test for operands[0] and operands[3] memory location.
12246 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12247 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12248 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12250 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12252 * target-insns.def: Add memory_blockage.
12253 * optabs.c (expand_memory_blockage): New function.
12254 (expand_asm_memory_barrier): Rename ...
12255 (expand_asm_memory_blockage): ... to this.
12256 (expand_mem_thread_fence): Call expand_memory_blockage
12257 instead of expand_asm_memory_barrier.
12258 (expand_mem_singnal_fence): Ditto.
12259 (expand_atomic_load): Ditto.
12260 (expand_atomic_store): Ditto.
12261 * doc/md.texi (Standard Pattern Names For Generation):
12262 Document memory_blockage instruction pattern.
12264 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12266 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12267 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12268 * config/rl78/rl78.md: New define_expand "adddi3".
12270 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12272 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12275 2017-10-13 Jeff Law <law@redhat.com>
12277 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12279 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12282 * internal-fn.c (expand_mul_overflow): If both operands have
12283 the same highpart of -1 or 0 and the topmost bit of lowpart
12284 is different, overflow is if res <= 0 rather than res < 0.
12286 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12288 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12289 TARGET_P9_VECTOR code for unaligned_load case.
12291 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12293 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12295 2017-10-13 Nathan Sidwell <nathan@acm.org>
12297 * tree-core.h (tree_contains_struct): Make bool.
12298 * tree.c (tree_contains_struct): Likewise.
12299 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12300 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12301 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12302 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12303 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12305 2017-10-13 Richard Biener <rguenther@suse.de>
12307 * graphite-isl-ast-to-gimple.c
12308 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12309 parameters and dominance check.
12310 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12311 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12312 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12313 Do not update SSA form here or do intermediate IL verification.
12314 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12315 (graphite_initialize): Remove check on the number of loops in
12316 the function and inline into graphite_transform_loops.
12317 (graphite_finalize): Inline into graphite_transform_loops.
12318 (graphite_transform_loops): Perform SSA update and IL verification
12320 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12322 2017-10-13 Richard Biener <rguenther@suse.de>
12324 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12325 graphite_expression_type_precision): Avoid global constructor
12327 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12328 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12329 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12330 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12331 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12332 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12334 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12337 * config/i386/i386.h (ix86_red_zone_size): New.
12338 * config/i386/i386.md (push peephole2s): Replace
12339 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12341 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12342 Alan Hayward <alan.hayward@arm.com>
12343 David Sherwood <david.sherwood@arm.com>
12345 * combine.c (can_change_dest_mode): Reject changes in
12346 REGMODE_NATURAL_SIZE.
12348 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12349 Alan Hayward <alan.hayward@arm.com>
12350 David Sherwood <david.sherwood@arm.com>
12352 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12353 (expand_debug_source_expr): Likewise.
12354 * combine.c (combine_simplify_rtx): Likewise.
12355 * cse.c (fold_rtx): Likewise.
12356 * fwprop.c (canonicalize_address): Likewise.
12357 * targhooks.c (default_shift_truncation_mask): Likewise.
12359 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12360 Alan Hayward <alan.hayward@arm.com>
12361 David Sherwood <david.sherwood@arm.com>
12363 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12364 (widened_mode): Likewise.
12365 (expand_unop): Likewise.
12366 * ree.c (transform_ifelse): Likewise.
12367 (merge_def_and_ext): Likewise.
12368 (combine_reaching_defs): Likewise.
12369 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12371 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12372 Alan Hayward <alan.hayward@arm.com>
12373 David Sherwood <david.sherwood@arm.com>
12375 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12376 * combine.c (gen_lowpart_for_combine): Likewise.
12377 * dwarf2out.c (rtl_for_decl_location): Likewise.
12378 * final.c (alter_subreg): Likewise.
12379 * rtlhooks.c (gen_lowpart_general): Likewise.
12380 (gen_lowpart_if_possible): Likewise.
12382 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12383 Alan Hayward <alan.hayward@arm.com>
12384 David Sherwood <david.sherwood@arm.com>
12386 * calls.c (expand_call): Use subreg_lowpart_offset.
12387 * cse.c (cse_insn): Likewise.
12388 * regcprop.c (copy_value): Likewise.
12389 (copyprop_hardreg_forward_1): Likewise.
12391 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12394 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12395 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12396 =Q constraints to +Q and into insn condition add check
12397 that operands[0] and operands[1] are equal.
12398 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12399 =Q constraints to +Q and into insn condition add check
12400 that operands[0] is equal to either operands[1] or operands[2].
12403 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12404 instead of handling MINUS_EXPR twice (once for each argument),
12405 canonicalize operand order and handle just once, use rtype where
12406 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12409 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12410 any values of __C while still being pattern recognizable as a simple
12411 rotate instruction.
12413 2017-10-13 Richard Biener <rguenther@suse.de>
12415 PR tree-optimization/82451
12417 2017-10-02 Richard Biener <rguenther@suse.de>
12419 PR tree-optimization/82355
12420 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12421 a mapping for the enclosing loop but avoid generating one for
12422 the loop tree root.
12423 (copy_bb_and_scalar_dependences): Remove premature codegen
12424 error on PHIs in blocks duplicated into multiple places.
12425 * graphite-scop-detection.c
12426 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12427 in the region use it as loop and nest to analyze the DR in.
12428 (try_generate_gimple_bb): Likewise.
12429 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12430 (add_loop_constraints): For blocks in a loop not in the region
12431 create a dimension with a single iteration.
12432 * sese.h (gbb_loop_at_index): Remove assert.
12434 * cfgloop.c (loop_preheader_edge): For the loop tree root
12435 return the single successor of the entry block.
12436 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12437 Reset the SCEV hashtable and niters.
12438 * graphite-scop-detection.c
12439 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12440 assert that we only have POLYNOMIAL_CHREC that vary in loops
12441 contained in the region.
12442 (scop_detection::graphite_can_represent_expr): Adjust.
12443 (scop_detection::stmt_has_simple_data_refs_p): For loops
12444 not in the region set loop to NULL. The nest is now the
12445 entry edge to the region.
12446 (try_generate_gimple_bb): Likewise.
12447 * sese.c (scalar_evolution_in_region): Adjust for
12448 instantiate_scev change.
12449 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12450 Make nest parameter the edge into the region.
12451 (create_data_ref): Likewise.
12452 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12453 entry edge into a region and adjust instantiate_scev calls.
12454 (create_data_ref): Likewise.
12455 (graphite_find_data_references_in_stmt): Likewise.
12456 (find_data_references_in_stmt): Pass the loop preheader edge
12457 from the nest argument.
12458 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12459 parameter the edge into the region.
12460 (instantiate_parameters): Use the loop preheader edge as entry.
12461 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12463 (get_instantiated_value_entry): Make instantiate_below parameter
12464 the edge into the region.
12465 (instantiate_scev_name): Likewise. Adjust dominance checks,
12466 when we cannot use loop-based instantiation instantiate by
12467 walking use-def chains.
12468 (instantiate_scev_poly): Adjust.
12469 (instantiate_scev_binary): Likewise.
12470 (instantiate_scev_convert): Likewise.
12471 (instantiate_scev_not): Likewise.
12472 (instantiate_array_ref): Remove.
12473 (instantiate_scev_3): Likewise.
12474 (instantiate_scev_2): Likewise.
12475 (instantiate_scev_1): Likewise.
12476 (instantiate_scev_r): Do not blindly handle N-operand trees.
12477 Do not instantiate array-refs. Handle all constants and invariants.
12478 (instantiate_scev): Make instantiate_below parameter
12479 the edge into the region.
12480 (resolve_mixers): Use the loop preheader edge for the region
12481 parameter to instantiate_scev_r.
12482 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12484 2017-10-13 Richard Biener <rguenther@suse.de>
12486 PR tree-optimization/82525
12487 * graphite-isl-ast-to-gimple.c
12488 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12490 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12491 Fail code generation when we cannot represent the isl integer.
12492 (binary_op_to_tree): Elide modulo operations that are no-ops
12493 in the type we code generate. Remove now superfluous code
12496 2017-10-13 Richard Biener <rguenther@suse.de>
12498 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12499 (scop_detection::harmful_loop_in_region): Remove premature
12500 IV type restriction.
12501 (scop_detection::graphite_can_represent_scev): We can handle
12502 pointer IVs just fine.
12504 2017-10-13 Alan Modra <amodra@gmail.com>
12506 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12507 "Clobbers and Scratch Registers". Add paragraph on
12508 alternative to clobbers for scratch registers and OpenBLAS
12511 2017-10-13 Alan Modra <amodra@gmail.com>
12513 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12514 example of a memory input for a string of known length. Move
12515 commentary out of table. Add a number of new examples
12516 covering array memory inputs.
12518 2017-10-12 Martin Liska <mliska@suse.cz>
12520 PR tree-optimization/82493
12521 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12522 (test_range_functions): New function.
12523 (sbitmap_c_tests): Likewise.
12524 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12525 * selftest.h (sbitmap_c_tests): New function.
12527 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12529 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12531 * config/rs6000/amo.h: Fix spacing issue.
12533 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12536 * config/i386/i386.md (*ashl<mode>3_mask_1,
12537 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12538 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12541 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12543 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12544 (profile_probability): Set max_probability
12545 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12547 * profile-count.c (profile_probability::differs_from_p): Do not
12548 rely on max_probaiblity == 10000
12550 2017-10-12 Jeff Law <law@redhat.com>
12552 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12555 2017-10-12 Martin Sebor <msebor@redhat.com>
12559 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12560 (handle_alias_pairs): Call it.
12561 * common.opt (-Wattribute-alias): New option.
12562 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12563 * doc/invoke.texi (-Wattribute-alias): Document.
12565 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12568 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12570 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12572 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12573 (make_hard_regno_born, make_hard_regno_dead): Update
12574 bb_killed_pseudos and bb_gen_pseudos.
12576 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12578 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12580 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12582 * config/alpha/alpha.c (alpha_split_conditional_move):
12583 Use std::swap instead of manually swapping.
12584 (alpha_stdarg_optimize_hook): Ditto.
12585 (alpha_canonicalize_comparison): Ditto.
12587 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12589 * tree-loop-distribution.c (struct builtin_info): New struct.
12590 (struct partition): Refactor fields into struct builtin_info.
12591 (partition_free): Free struct builtin_info.
12592 (build_size_arg_loc, build_addr_arg_loc): Delete.
12593 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12594 information from struct builtin_info.
12595 (find_single_drs): New function refactored from classify_partition.
12596 Also moved builtin validity checks to this function.
12597 (compute_access_range, alloc_builtin): New functions.
12598 (classify_builtin_st, classify_builtin_ldst): New functions.
12599 (classify_partition): Refactor code into functions find_single_drs,
12600 classify_builtin_st and classify_builtin_ldst.
12601 (distribute_loop): Don't do runtime alias check when distributing
12603 (find_seed_stmts_for_distribution): New function.
12604 (pass_loop_distribution::execute): Refactor code finding seed
12605 stmts into above function. Support distribution for the innermost
12606 two-level loop nest. Adjust dump information.
12608 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12610 * tree-loop-distribution.c: Adjust the general comment.
12611 (NUM_PARTITION_THRESHOLD): New macro.
12612 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12613 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12614 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12615 in call to build_partition_graph.
12616 (finalize_partitions): New parameter. Make loop distribution more
12617 conservative by fusing more partitions.
12618 (distribute_loop): Don't do runtime alias check in case of loop nest
12620 (find_seed_stmts_for_distribution): New function.
12621 (prepare_perfect_loop_nest): New function.
12622 (pass_loop_distribution::execute): Refactor code finding seed stmts
12623 and loop nest into above functions. Support loop nest distribution.
12624 Adjust dump information accordingly.
12626 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12628 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12629 and set PTYPE_SEQUENTIAL for merged partition.
12631 2017-10-12 Richard Biener <rguenther@suse.de>
12633 PR tree-optimization/69728
12635 2017-09-19 Richard Biener <rguenther@suse.de>
12637 PR tree-optimization/69728
12638 * graphite-sese-to-poly.c (schedule_error): New global.
12639 (add_loop_schedule): Handle empty domain by failing the
12641 (build_original_schedule): Handle schedule_error.
12643 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12644 domain by returning an unchanged schedule.
12646 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12648 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12649 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12651 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12653 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12656 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12659 * expr.c (store_field): Don't optimize away bitsize == 0 store
12660 from CALL_EXPR with addressable return type.
12662 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12664 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12665 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12666 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12667 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12668 TARGET_ISEL instead of TARGET_ISEL<sel>.
12670 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12672 * config/rs6000/rs6000.c
12673 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12675 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12677 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12679 (reg_or_cint_operand): Fix comment.
12680 (reg_or_zero_operand): New predicate.
12681 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12682 * config/rs6000/rs6000.c (output_isel): Delete.
12683 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12684 instead of reg_or_cint_operand. Output instruction directly (not via
12686 (isel_unsigned_<mode>): Ditto.
12687 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12688 gpc_reg_operand. Add an instruction alternative for this. Output
12689 instruction directly.
12690 (*isel_reversed_unsigned_<mode>): Ditto.
12692 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12694 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12695 (TARGET_CANONICALIZE_COMPARISON): Define.
12697 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12700 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12701 Check whether the dest is REG before adding REG_EQUIV note.
12703 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12706 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12708 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12709 (make_hard_regno_born, make_hard_regno_dead): Update
12710 bb_killed_pseudos and bb_gen_pseudos.
12712 2017-10-11 Nathan Sidwell <nathan@acm.org>
12714 * incpath.h (enum incpath_kind): Name enum, prefix values.
12715 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12716 * incpath.c (heads, tails): Use INC_MAX.
12717 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12718 (merge_include_chains, split_quote_chain,
12719 register_include_chains): Update incpath_kind names.
12720 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12721 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12723 (add_framework_path, darwin_register_objc_includes): Likewise.
12724 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12726 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12728 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12729 Do not use float_operator operator predicate.
12730 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12731 * config/i386/predicates.md (float_operator): Remove predicate.
12733 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12735 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12736 (*jccxf_i387): Ditto.
12737 (*jcc<mode>_i387): Ditto.
12738 (*jccu<mode>_i387): Ditto.
12739 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12740 (*jcc_*_i387 splitters): Remove.
12741 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12742 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12743 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12746 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12748 * profile-count.h (slow_safe_scale_64bit): New function.
12749 (safe_scale_64bit): New inline.
12750 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12751 * profile-count.c: Include wide-int.h
12752 (slow_safe_scale_64bit): New.
12754 2017-10-11 Nathan Sidwell <nathan@acm.org>
12756 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12757 HAS_DECL_ASSEMBLER_NAME_P.
12758 * gimple-expr.c (gimple_decl_printable_name: Check
12759 HAS_DECL_ASSEMBLER_NAME_P too.
12760 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12761 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12762 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12763 * passes.c (rest_of_decl_compilation): Check
12764 HAS_DECL_ASSEMBLER_NAME_P too.
12765 * recog.c (verify_changes): Likewise.
12766 * tree-pretty-print.c (dump_decl_name): Likewise.
12767 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12769 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12770 (SET_DECL_ASSEMBLER_NAME): Use it.
12771 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12772 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12773 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12775 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12777 * config.gcc (i386, x86_64): Add extra objects.
12778 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12779 (ix86_min_insn_size): Declare.
12780 (ix86_issue_rate): Declare.
12781 (ix86_adjust_cost): Declare.
12782 (ia32_multipass_dfa_lookahead): Declare.
12783 (ix86_macro_fusion_p): Declare.
12784 (ix86_macro_fusion_pair_p): Declare.
12785 (ix86_bd_has_dispatch): Declare.
12786 (ix86_bd_do_dispatch): Declare.
12787 (ix86_core2i7_init_hooks): Declare.
12788 (ix86_atom_sched_reorder): Declare.
12789 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12790 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12791 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12792 (rip_relative_addr_p): Rename to ...
12793 (ix86_rip_relative_addr_p): ... this one; export.
12794 (memory_address_length): Update.
12795 (ix86_issue_rate): Move to x86-tune-sched.c.
12796 (ix86_flags_dependent): Move to x86-tune-sched.c.
12797 (ix86_agi_dependent): Move to x86-tune-sched.c.
12798 (exact_dependency_1): Move to x86-tune-sched.c.
12799 (exact_store_load_dependency): Move to x86-tune-sched.c.
12800 (ix86_adjust_cost): Move to x86-tune-sched.c.
12801 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12802 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12803 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12804 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12805 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12806 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12807 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12808 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12809 (min_insn_size): Rename to ...
12810 (ix86_min_insn_size): ... this one; export.
12811 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12812 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12813 (core2i7_first_cycle_multipass_backtrack): Move to
12814 x86-tune-sched-core.c.
12815 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12816 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12817 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12818 (ix86_avoid_jump_mispredicts): Update.
12819 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12820 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12821 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12822 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12823 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12824 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12825 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12826 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12827 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12828 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12829 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12830 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12831 (BIG): Move to ix86-tune-sched-bd.c.
12832 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12833 (enum insn_path): Move to ix86-tune-sched-bd.c.
12834 (get_mem_group): Move to ix86-tune-sched-bd.c.
12835 (is_cmp): Move to ix86-tune-sched-bd.c.
12836 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12837 (is_branch): Move to ix86-tune-sched-bd.c.
12838 (is_prefetch): Move to ix86-tune-sched-bd.c.
12839 (init_window): Move to ix86-tune-sched-bd.c.
12840 (allocate_window): Move to ix86-tune-sched-bd.c.
12841 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12842 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12843 (process_end_window): Move to ix86-tune-sched-bd.c.
12844 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12845 (find_constant): Move to ix86-tune-sched-bd.c.
12846 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12847 (has_immediate): Move to ix86-tune-sched-bd.c.
12848 (get_insn_path): Move to ix86-tune-sched-bd.c.
12849 (get_insn_group): Move to ix86-tune-sched-bd.c.
12850 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12851 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12852 (add_insn_window): Move to ix86-tune-sched-bd.c.
12853 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12854 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12855 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12856 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12857 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12858 (do_dispatch): Move to ix86-tune-sched-bd.c.
12859 (has_dispatch): Move to ix86-tune-sched-bd.c.
12860 * config/i386/t-i386: Add new object files.
12861 * config/i386/x86-tune-costs.h: New file.
12862 * config/i386/x86-tune-sched-atom.c: New file.
12863 * config/i386/x86-tune-sched-bd.c: New file.
12864 * config/i386/x86-tune-sched-core.c: New file.
12865 * config/i386/x86-tune-sched.c: New file.
12867 2017-10-11 Liu Hao <lh_mouse@126.com>
12869 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12870 the generic version below instead.
12871 (should_colorize): Recognize Windows consoles as terminals
12873 * pretty-print.c [__MINGW32__] (write_all): New function.
12874 [__MINGW32__] (find_esc_head): Likewise.
12875 [__MINGW32__] (find_esc_terminator): Likewise.
12876 [__MINGW32__] (eat_esc_sequence): Likewise.
12877 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12879 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12882 2017-10-11 Richard Biener <rguenther@suse.de>
12884 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12885 Properly call analyze_scalar_evolution with the loop of the stmt.
12887 2017-10-11 Richard Biener <rguenther@suse.de>
12889 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12890 * tree-core.h (tree_base): Add chrec_var union member.
12891 * tree.h (CHREC_VAR): Remove.
12892 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12893 * tree-chrec.h (build_polynomial_chrec): Adjust.
12894 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12895 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12897 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12899 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12900 * match.pd: ... here.
12901 ((T) X == (T) Y): Relax condition.
12903 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12905 PR tree-optimization/82472
12906 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12908 (break_alias_scc_partitions): Update postorder number.
12910 2017-10-11 Martin Liska <mliska@suse.cz>
12913 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12915 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12917 2017-10-11 Martin Liska <mliska@suse.cz>
12919 * print-rtl.c (print_insn): Move declaration of idbuf
12920 to same scope as name.
12922 2017-10-11 Martin Liska <mliska@suse.cz>
12927 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12928 volatile arguments.
12930 2017-10-11 Martin Liska <mliska@suse.cz>
12933 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12934 volatile arguments.
12936 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12938 * config.gcc (default_gnu_indirect_function): Default to yes for
12939 arm*-*-linux* with glibc.
12941 2017-10-11 Richard Biener <rguenther@suse.de>
12943 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12944 default-defs and types we do not want to analyze.
12945 (interpret_loop_phi): Replace unreachable code with an assert.
12946 (compute_scalar_evolution_in_loop): Remove and inline ...
12947 (analyze_scalar_evolution_1): ... here, replacing condition with
12948 what makes the intent clearer. Remove handling of cases
12949 get_scalar_evolution now handles.
12951 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12953 PR rtl-optimization/81434
12954 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12955 comment for main loop. In sched_group_found if, also add checks for
12956 pass and min_cost_group.
12958 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12960 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12961 (rs6000_insn_cost): New function.
12962 * config/rs6000/rs6000.md (cost): New attribute.
12964 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12965 H.J. Lu <hongjiu.lu@intel.com>
12969 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12970 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12971 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12972 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12973 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12974 OPTION_MASK_ISA_MMX and some other bit set require both
12975 mmx and the ISAs without the mmx bit.
12976 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12977 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12978 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12979 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12980 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12981 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12982 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12983 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12984 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12985 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12986 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12987 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12988 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12989 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12990 Add OPTION_MASK_ISA_MMX.
12992 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12994 * config.gcc (armv7*-*-freebsd*): New target.
12995 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12997 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12999 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
13000 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
13003 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
13005 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
13006 a template parameter.
13007 (WIDE_INT_REF_FOR): Update accordingly.
13008 * tree.h (wi::int_traits <const_tree>): Delete.
13009 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13010 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13011 (wi::tree_to_wide_ref): New typedef.
13012 (wi::to_wide): New function.
13013 * calls.c (get_size_range): Use wi::to_wide when operating on
13014 trees as wide_ints.
13015 * cgraph.c (cgraph_node::create_thunk): Likewise.
13016 * config/i386/i386.c (ix86_data_alignment): Likewise.
13017 (ix86_local_alignment): Likewise.
13018 * dbxout.c (stabstr_O): Likewise.
13019 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13020 * expr.c (const_vector_from_tree): Likewise.
13021 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13022 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13023 (fold_negate_expr_1, int_const_binop_1, const_binop)
13024 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13025 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13026 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13027 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13028 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13029 (fold_not_const, round_up_loc): Likewise.
13030 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13031 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13032 (alloca_call_type): Likewise.
13033 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13034 * godump.c (go_output_typedef): Likewise.
13035 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13036 * internal-fn.c (get_min_precision): Likewise.
13037 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13038 * ipa-polymorphic-call.c
13039 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13040 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13041 (ipa_modify_call_arguments): Likewise.
13042 * match.pd: Likewise.
13043 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13044 * print-tree.c (print_node_brief, print_node): Likewise.
13045 * stmt.c (expand_case): Likewise.
13046 * stor-layout.c (layout_type): Likewise.
13047 * tree-affine.c (tree_to_aff_combination): Likewise.
13048 * tree-cfg.c (group_case_labels_stmt): Likewise.
13049 * tree-data-ref.c (dr_analyze_indices): Likewise.
13050 (prune_runtime_alias_test_list): Likewise.
13051 * tree-dump.c (dequeue_and_dump): Likewise.
13052 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13053 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13054 * tree-pretty-print.c (dump_generic_node): Likewise.
13055 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13056 (simple_iv_with_niters): Likewise.
13057 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13058 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13059 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13060 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13061 (refine_value_range_using_guard, number_of_iterations_ne_max)
13062 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13063 (get_cst_init_from_scev, record_nonwrapping_iv)
13064 (scev_var_range_cant_overflow): Likewise.
13065 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13066 * tree-ssa-pre.c (compute_avail): Likewise.
13067 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13068 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13069 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13070 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13071 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13072 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13073 (dump_case_nodes, try_switch_expansion): Likewise.
13074 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13075 (vect_do_peeling): Likewise.
13076 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13077 * tree-vect-stmts.c (vectorizable_load): Likewise.
13078 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13079 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13080 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13081 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13082 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13083 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13084 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13085 (evrp_dom_walker::before_dom_children): Likewise.
13086 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13087 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13088 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13089 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13090 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13091 (get_range_pos_neg): Likewise.
13092 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13093 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13094 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13095 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13096 * config/avr/avr.c (avr_fold_builtin): Likewise.
13097 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13098 * config/msp430/msp430.c (msp430_attr): Likewise.
13099 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13100 * config/powerpcspe/powerpcspe-c.c
13101 (altivec_resolve_overloaded_builtin): Likewise.
13102 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13103 (rs6000_expand_ternop_builtin): Likewise.
13104 * config/rs6000/rs6000-c.c
13105 (altivec_resolve_overloaded_builtin): Likewise.
13106 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13107 (rs6000_expand_ternop_builtin): Likewise.
13108 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13110 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13112 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13113 when copying loop nest with only one inner loop.
13115 2017-10-10 Richard Biener <rguenther@suse.de>
13117 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13118 blocks if SCEV is active.
13119 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13121 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13122 (scev_initialize): Assert we are not yet initialized.
13124 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13126 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13127 inner loop's exit stmt by making it always exit the loop, otherwise
13128 we would generate an infinite empty loop.
13130 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13132 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13133 renaming variables in new preheader if it's deleted.
13135 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13137 * tree-loop-distribution.c (struct partition): Remove unused field
13138 loops of the structure.
13139 (partition_alloc, partition_free): Ditto.
13140 (build_rdg_partition_for_vertex): Ditto.
13142 2017-10-09 Jeff Law <law@redhat.com>
13144 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13145 return type to match prototype and documentation.
13147 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13149 * config/rs6000/rs6000.c (processor_costs): Move to ...
13150 * config/rs6000/rs6000.h: ... here.
13151 (rs6000_cost): Declare.
13153 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13155 * except.c (setjmp_fn): New global variable.
13156 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13157 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13158 if DONT_USE_BUILTIN_SETJMP is defined.
13160 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13162 * target.def (insn_cost): New hook.
13163 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13164 * doc/tm.texi: Regenerate.
13165 * rtlanal.c (insn_cost): Use the new hook.
13167 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13169 * combine.c (combine_validate_cost): Compute the new insn_cost,
13170 not just pattern_cost.
13171 (try_combine): Adjust comment.
13173 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13175 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13177 * combine.c (uid_insn_cost): Adjust comment.
13178 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13180 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13181 * dse.c (find_shift_sequence): Ditto.
13182 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13183 (bb_valid_for_noce_process_p): Use pattern_cost.
13184 * rtl.h (insn_rtx_cost): Delete.
13185 (pattern_cost): New prototype.
13186 (insn_cost): New prototype.
13187 * rtlanal.c (insn_rtx_cost): Rename to...
13188 (pattern_cost): ... this.
13191 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13193 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13194 (*jcc<mode>_0_r_i387): Ditto.
13195 (*jccxf_r_i387): Ditto.
13196 (*jcc<mode>_r_i387): Ditto.
13197 (*jccu<mode>_r_i387): Ditto.
13198 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13199 (*jcc): Rename from *jcc_1.
13201 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13203 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13204 deferred rescans after the lvx/stvx recombination pre-pass.
13206 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13208 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13209 memory operation instruction support.
13210 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13211 (rs6000-ibm-aix[789]*): Likewise.
13212 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13213 Document new functions.
13215 2017-10-09 Richard Biener <rguenther@suse.de>
13217 PR tree-optimization/82397
13218 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13219 equality only for semantically equal trees.
13221 2017-10-09 Richard Biener <rguenther@suse.de>
13223 PR tree-optimization/82449
13224 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13225 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13226 allow constant addresses.
13227 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13230 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13232 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13235 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13238 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13241 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13244 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13246 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13249 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13250 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13252 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13254 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13255 (WI_BINARY_PREDICATE_RESULT): Likewise.
13256 (wi::binary_traits::operator_result): New type.
13257 (wi::binary_traits::predicate_result): Likewise.
13258 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13259 (generic_wide_int::operator==, generic_wide_int::operator!=)
13260 (generic_wide_int::operator&, generic_wide_int::and_not)
13261 (generic_wide_int::operator|, generic_wide_int::or_not)
13262 (generic_wide_int::operator^, generic_wide_int::operator+
13263 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13265 (operator~, unary operator-, operator==, operator!=, operator&)
13266 (operator|, operator^, operator+, binary operator-, operator*): New
13268 * expr.c (get_inner_reference): Use wi::bit_and_not.
13269 * fold-const.c (fold_binary_loc): Likewise.
13270 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13271 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13272 (bit_value_binop): Likewise.
13273 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13274 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13275 (extract_range_from_binary_expr_1): Likewise.
13276 (masked_increment): Likewise.
13277 (simplify_bit_ops_using_ranges): Likewise.
13279 2017-10-09 Martin Jambor <mjambor@suse.cz>
13282 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13283 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13284 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13286 (hsa_fixup_mov_insn_type): New function.
13287 (hsa_op_with_type::get_in_type): Use it.
13288 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13289 immediates in an assert.
13290 (hsa_op_with_type::extend_int_to_32bit): New method.
13291 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13292 types. Convert to dest type if necessary.
13293 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13294 (reg_for_gimple_ssa): Pass false as min32int to
13295 hsa_type_for_scalar_tree_type.
13296 (gen_hsa_addr): Fixup type when creating addresable temporary.
13297 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13298 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13299 necessary. Call hsa_fixup_mov_insn_type.
13300 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13301 extend operands and convert to dest type if necessary.
13302 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13303 to dest type if necessary.
13304 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13305 if conversion nt necessary and size matches.
13306 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13307 to dest type if necessary.
13308 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13309 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13311 (gen_hsa_clrsb): Likewise.
13312 (gen_hsa_ffs): Likewise.
13313 (gen_hsa_divmod): Extend operands and convert to dest type if
13315 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13317 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13319 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13320 Remove empty default arguments. Use a brace block as output
13322 (conditional return): Ditto.
13324 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13325 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13327 (group_ending_nop): Ditto.
13328 (doloop_end): Ditto.
13329 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13330 (splitters for those): Ditto.
13332 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13334 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13335 a conditional jump (and the compare for it) so that pc_rtx is the
13337 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13338 for the deleted and renamed ctr<mode>_internal[234] patterns.
13339 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13340 Delete second conditional return pattern.
13341 (ctr<mode>_internal2): Delete this second bdnz pattern.
13342 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13343 (ctr<mode>_internal4): Delete this second bdz pattern.
13345 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13347 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13348 (always_initialized_rtx_for_ssa_name_p): New predicate.
13349 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13350 (finish_out_of_ssa): Free new field of SA.
13351 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13352 * tree-ssa-coalesce.c: Include tree-ssa.h.
13353 (get_parm_default_def_partitions): Remove extern keyword.
13354 (get_undefined_value_partitions): New function.
13355 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13356 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13357 uninitialized bits.
13358 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13360 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13362 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13364 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13366 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13367 for targets that preffer 128bit.
13369 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13371 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13373 2017-10-08 Olivier Hainque <hainque@adacore.com>
13375 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13376 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13378 (thumb_set_return_address): Likewise.
13380 2017-10-08 Olivier Hainque <hainque@adacore.com>
13382 * common/config/arm/arm-common.c (arm_except_unwind_info):
13383 Handle DWARF2_UNWIND_INFO.
13385 2017-10-07 Michael Collison <michael.collison@arm.com>
13387 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13390 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13392 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13393 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13394 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13395 defined, force the creation of a new block for a dispatch label.
13397 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13399 * invoke.texi (Wsuggest-attribute=cold): Document.
13400 * common.opt (Wsuggest-attribute=cold): New
13401 * ipa-pure-const.c (warn_function_cold): New function.
13402 * predict.c (compute_function_frequency): Use it.
13403 * predict.h (warn_function_cold): Declare.
13405 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13407 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13410 2017-10-06 Martin Liska <mliska@suse.cz>
13412 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13413 keyword for member functions.
13414 (struct sanopt_tree_couple): New struct.
13415 (struct sanopt_tree_couple_hash): New function.
13416 (struct sanopt_ctx): Add new hash_map.
13417 (has_dominating_ubsan_ptr_check): New function.
13418 (record_ubsan_ptr_check_stmt): Likewise.
13419 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13420 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13421 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13423 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13426 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13427 aarch64_simd_valid_immediate on CONST_VECTORs.
13428 (aarch64_reg_or_bic_imm): Likewise.
13430 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13432 PR rtl-optimization/82396
13433 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13435 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13437 * graphite-dependences.c (scop_get_reads): Move code to...
13438 (scop_get_must_writes): Move code to...
13439 (scop_get_may_writes): Move code to...
13440 (scop_get_reads_and_writes): ... here.
13441 (scop_get_dependences): Call scop_get_reads_and_writes.
13443 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13445 PR tree-optimization/82434
13446 * fold-const.h (can_native_encode_type_p,
13447 can_native_encode_string_p): Remove.
13448 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13449 don't encode anything, just return what would be otherwise returned.
13450 (native_encode_fixed, native_encode_complex, native_encode_vector):
13452 (native_encode_string): Likewise. Inline by hand
13453 can_native_encode_string_p.
13454 (can_native_encode_type_p): Remove.
13455 (can_native_encode_string_p): Remove.
13456 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13457 STRING_CSTs using can_native_encode_string_p, test all
13458 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13459 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13460 argument from native_encode_expr.
13461 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13462 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13465 2017-10-06 Richard Biener <rguenther@suse.de>
13467 PR tree-optimization/82397
13468 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13469 operand_equal_p but rely on data_ref_compare_tree for detecting
13471 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13472 to match up with dr_group_sort_cmp.
13474 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13477 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13479 * config/s390/s390-builtin-types.def: Regenerate.
13481 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13484 * config/s390/s390-builtin-types.def: Regenerate.
13485 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13486 Change flag from B_VXE to B_VX.
13487 (s390_vec_min_dbl): Remove B_VXE flag.
13489 2017-10-06 Richard Biener <rguenther@suse.de>
13491 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13492 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13493 translate_isl_ast_to_gimple::is_valid_rename,
13494 translate_isl_ast_to_gimple::get_rename,
13495 translate_isl_ast_to_gimple::get_def_bb_for_const,
13496 translate_isl_ast_to_gimple::get_new_name,
13497 translate_isl_ast_to_gimple::collect_all_ssa_names,
13498 translate_isl_ast_to_gimple::copy_loop_phi_args,
13499 translate_isl_ast_to_gimple::collect_all_ssa_names,
13500 translate_isl_ast_to_gimple::copy_loop_phi_args,
13501 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13502 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13503 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13504 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13505 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13506 translate_isl_ast_to_gimple::copy_cond_phi_args,
13507 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13508 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13509 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13510 translate_isl_ast_to_gimple::rename_uses,
13511 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13512 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13513 (set_rename_for_each_def): Likewise.
13514 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13515 here. Handle rewriting SCEV analyzable uses here.
13516 (copy_bb_and_scalar_dependences): Generate code for PHI
13518 (graphite_regenerate_ast_isl): Adjust.
13519 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13520 (add_write, add_read): New functions.
13521 (build_cross_bb_scalars_def): Use it and simplify.
13522 (build_cross_bb_scalars_use): Likewise.
13523 (graphite_find_cross_bb_scalar_vars): Inline into...
13524 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13525 simulating out-of-SSA. Compute liveout and add dependencies.
13526 (build_scops): Force an empty entry block.
13527 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13529 (sese_build_liveouts): Declare.
13530 (sese_trivially_empty_bb_p): Likewise.
13531 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13532 compute liveout and debug_liveout.
13533 (sese_bad_liveouts_use): Remove.
13534 (sese_reset_debug_liveouts_bb): Likewise.
13535 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13536 (sese_build_liveouts): Build liveout and debug_liveout and store
13538 (new_sese_info): Adjust.
13539 (free_sese_info): Likewise.
13540 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13541 do not build liveout here.
13542 (move_sese_in_condition): Adjust region entry.
13543 (scev_analyzable_p): Match up with chrec_apply requirements.
13544 (sese_trivially_empty_bb_p): New.
13545 * tree-into-ssa.c (get_reaching_def): Properly support generating
13546 default-defs for incremental rewrite of anonymous names.
13548 2017-10-06 Richard Biener <rguenther@suse.de>
13550 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13551 precision do not perform modulo reduction.
13553 2017-10-06 Richard Biener <rguenther@suse.de>
13555 PR tree-optimization/82436
13556 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13557 conservatively choose the vectorization factor when checking
13558 whether we can perform the required load permutation.
13559 (vect_transform_slp_perm_load): Assert when we may not fail.
13561 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13563 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13564 message for incompatible -msdata=* and -mcall-* options.
13566 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13568 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13569 rate for post-reload scheduling.
13571 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13573 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13575 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13577 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13578 to improve monte carlo in scimark.
13580 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13582 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13583 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13584 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13585 pentium4_cost, nocona_cost): Set reassociation width to 1.
13586 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13587 width to 2 for fp operations and 1 otherwise.
13588 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13590 (atom_cost): Set reassociation width to 2.
13591 (slm_cost, generic_cost): Set fp reassociation width
13592 to 2 and 1 otherwise.
13593 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13594 (core_cost): Set fp reassociation width to 4 and vector to 2.
13595 (ix86_reassociation_width): Rewrite using cost table; special case
13596 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13597 and TARGET_AVX128_OPTIMAL.
13598 * config/i386/i386.h (processor_costs): Add
13599 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13600 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13601 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13602 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13603 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13604 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13606 2017-10-05 Nathan Sidwell <nathan@acm.org>
13608 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13610 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13612 * config/arm/arm.c (arm_test_fpu_data): New.
13613 (arm_run_selftests): Call arm_test_fpu_data.
13615 2017-10-04 Nathan Sidwell <nathan@acm.org>
13617 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13619 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13621 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13623 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13624 check type for aarch64_simd_valid_immediate.
13625 (aarch64_output_simd_mov_immediate): Update prototype.
13626 (aarch64_simd_valid_immediate): Update prototype.
13627 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13628 support for ORR-immediate.
13629 (and<mode>3): modified pattern to add support for BIC-immediate.
13630 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13631 now checks for valid immediate for BIC and ORR based on new enum
13633 (aarch64_output_simd_mov_immediate): Function now used to output
13634 BIC/ORR imm as well based on new enum argument.
13635 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13637 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13638 (aarch64_reg_or_bic_imm): Likewise.
13640 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13642 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13643 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13645 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13649 PR rtl-optimization/82396
13650 * haifa-sched.c (autopref_multipass_init): Simplify
13652 (autopref_rank_data): Simplify sort order.
13653 * sched-int.h (autopref_multipass_data_): Remove
13654 multi_mem_insn_p, min_offset and max_offset.
13656 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13658 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13660 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13662 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13663 vect_doubleint_cvt.
13665 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13667 * doc/sourcebuild.texi: Document vect_long_mult.
13669 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13671 PR tree-optimization/82413
13672 * fold-const.c (build_range_check): Use widest_int when comparing
13673 the maximum ETYPE value with HIGH.
13675 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13677 PR rtl-optimization/82396
13678 * haifa-sched.c (autopref_multipass_init): Simplify
13680 (autopref_rank_data): Simplify sort order.
13681 * sched-int.h (autopref_multipass_data_): Remove
13682 multi_mem_insn_p, min_offset and max_offset.
13684 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13686 PR tree-optimization/82381
13687 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13688 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13691 PR tree-optimization/82374
13692 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13693 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13694 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13695 current_function_decl to the new decl.
13697 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13699 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13700 helper macro for IEEE float128 hardware built-in functions.
13701 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13703 (TRUNCF128_ODD): Likewise.
13704 (ADDF128_ODD): Likewise.
13705 (SUBF128_ODD): Likewise.
13706 (MULF128_ODD): Likewise.
13707 (DIVF128_ODD): Likewise.
13708 (FMAF128_ODD): Likewise.
13709 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13710 UNSPEC_TRUNC_ROUND_TO_ODD.
13711 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13712 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13713 floating point round to odd instructions.
13714 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13715 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13716 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13717 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13718 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13719 (trunc<mode>sf2_hw): Change the truncate with round to odd
13720 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13721 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13722 to odd hardware instructions.
13723 (sub<mode>3_odd): Likewise.
13724 (mul<mode>3_odd): Likewise.
13725 (div<mode>3_odd): Likewise.
13726 (sqrt<mode>2_odd): Likewise.
13727 (fma<mode>4_odd): Likewise.
13728 (fms<mode>4_odd): Likewise.
13729 (nfma<mode>4_odd): Likewise.
13730 (nfms<mode>4_odd): Likewise.
13731 (trunc<mode>df2_odd): Change the truncate with round to odd
13732 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13734 * doc/extend.texi (PowerPC built-in functions): Update documentation
13735 for existing IEEE float128-bit built-in functions. Add built-in
13736 functions that generate the IEEE 128-bit floating point round to
13739 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13741 PR rtl-optimization/77729
13742 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13743 to (X&(C1&~C2))|C2 transformations.
13745 2017-10-03 Martin Jambor <mjambor@suse.cz>
13747 PR tree-optimization/82363
13748 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13749 mismatch, mark lacc written regardless of racc.
13751 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13753 PR tree-optimization/82381
13754 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13755 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13756 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13757 fallthrough into reassoc_stmt_dominates_stmt_p.
13760 * combine.c (combine_instructions): Don't combine in unreachable
13763 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13766 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13767 function to not use the have_cpu variable. Do not set cpu_index,
13768 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13769 or the default cpu.
13770 (rs6000_valid_attribute_p): Remove duplicate initializations of
13771 old_optimize and func_optimize.
13772 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13773 (rs6000_activate_target_options): Make global.
13774 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13777 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13779 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13780 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13781 Set *poffset to 0 and *psize and *pmax_size to -1 if
13782 *poffset + *psize overflows in HOST_WIDE_INT.
13784 PR tree-optimization/82387
13785 PR tree-optimization/82388
13786 PR tree-optimization/82389
13787 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13788 instead of live_bytes non-NULL.
13790 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13793 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13796 2017-10-02 Richard Biener <rguenther@suse.de>
13798 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13799 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13800 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13802 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13804 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13805 requested precision matches the type's.
13806 * calls.c (alloc_max_size): Calculate the new candidate size as
13807 a widest_int and use wi::to_widest when comparing it with the
13808 current candidate size.
13809 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13810 zero rather than integer_zero_node.
13811 * match.pd: Check for a no-op conversion before using wi::add
13812 rather than after. Use tree_to_uhwi when summing small shift
13813 counts into an unsigned int.
13815 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13816 Alan Hayward <alan.hayward@arm.com>
13817 David Sherwood <david.sherwood@arm.com>
13820 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13821 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13822 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13823 POINTER_AND_FP_REGS.
13825 2017-10-02 Richard Biener <rguenther@suse.de>
13827 PR tree-optimization/82355
13828 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13829 a mapping for the enclosing loop but avoid generating one for
13830 the loop tree root.
13831 (copy_bb_and_scalar_dependences): Remove premature codegen
13832 error on PHIs in blocks duplicated into multiple places.
13833 * graphite-scop-detection.c
13834 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13835 in the region use it as loop and nest to analyze the DR in.
13836 (try_generate_gimple_bb): Likewise.
13837 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13838 (add_loop_constraints): For blocks in a loop not in the region
13839 create a dimension with a single iteration.
13840 * sese.h (gbb_loop_at_index): Remove assert.
13842 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13844 * omp-expand.c (adjust_context_scope): New function.
13845 (expand_parallel_call): Call adjust_context_scope.
13847 2017-10-01 Jeff Law <law@redhat.com>
13849 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13850 dom_opt_dom_walker class with direct access to private members.
13851 Add comments. Call test_for_singularity.
13852 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13853 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13854 m_dummy_cond anymore.
13855 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13857 (pass_dominator:execute): Build the dummy_cond here and pass it
13858 to the dom_opt_dom_walker ctor.
13859 (test_for_singularity): New function.
13861 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13862 Maya Rashish <coypu@sdf.org>
13864 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13865 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13866 (alpha*-*-netbsd*) Use nbsd_tm_file.
13867 (arm*-*-netbsdelf*) Likewise.
13868 (i[34567]86-*-netbsdelf*) Likewise.
13869 (x86_64-*-netbsd*) Likewise.
13870 (mips*-*-netbsd*) Likewise.
13871 (powerpc-*-netbsd*) Likewise.
13872 (sh*-*-netbsd*) Likewise.
13873 (sparc-*-netbsdelf*) Likewise.
13874 (sparc64-*-netbsd*) Likewise.
13875 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13877 (vax-*-netbsdelf*) Likewise.
13878 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13879 (UINT_FAST8_TYPE) Likewise.
13880 (INT_FAST16_TYPE) Check CHAR_FAST16.
13881 (UINT_FAST16_TYPE) Likewise.
13883 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13886 * config/i386/i386.md
13887 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13888 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13889 *divmodsi4_zext_2): New define_insn_and_split.
13890 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13891 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13892 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13893 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13894 New define_insn_and_split.
13895 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13896 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13897 operands[1] having DImode when mode is SImode.
13899 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13900 always SImode for DIV and MOD in REG_EQUAL notes.
13902 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13904 PR middle-end/82319
13905 * match.pd: Fix handling of NaNs in pattern.
13907 2017-09-29 Jeff Law <law@redhat.com>
13909 * sbitmap.c (bitmap_bit_in_range_p): New function.
13910 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13911 * tree-ssa-dse.c (live_bytes_read): New function.
13912 (dse_classify_store): Ignore reads of dead bytes.
13914 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13915 typos and whitespace errors.
13916 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13917 * config/s390/s390.c (s390_emit_prologue): Likewise.
13919 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13922 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13923 with a symbol for LRA.
13925 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13927 PR rtl-optimization/82338
13928 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13930 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13932 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13933 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13934 (qsort_chk): Declare.
13935 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13936 (qsort_chk): New function.
13938 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13940 PR tree-optimization/82337
13941 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13942 phi definition if the PHI result appears in an abnormal PHI.
13943 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13944 basis appears in an abnormal PHI.
13946 2017-09-29 Richard Biener <rguenther@suse.de>
13948 * graphite-isl-ast-to-gimple.c
13949 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13950 (binary_op_to_tree): Use it.
13951 (get_rename_from_scev): Likewise.
13952 (copy_loop_phi_nodes): Likewise.
13953 (copy_bb_and_scalar_dependences): Likewise.
13954 (translate_pending_phi_nodes): Likewise.
13956 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13959 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13960 for movabsq $(i32 << shift), r64.
13962 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13964 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13965 index when encoding %esp as %rsp to avoid 0x67 prefix.
13967 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13969 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13970 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13972 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13974 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13975 Extensions with more than 16 double VFP registers.
13976 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13977 to_clear_mask and all code related to it. Replace the remaining
13978 entry by a sbitmap and adapt code accordingly.
13980 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13982 * brig-builtins.def: Change pure attributes to const.
13984 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13986 * config.gcc (default_gnu_indirect_function): Default to yes for
13987 sparc*-*-linux* with glibc.
13989 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13991 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13992 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13993 when creating .init_array and .fini_array sections with priority
13996 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13999 * config/aarch64/aarch64.c
14000 (aarch64_builtin_support_vector_misalignment): Always return false
14001 when misalignment is unknown.
14003 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14005 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
14006 this function to return false if the definition used by the swap
14007 instruction is artificial, or if the memory address from which the
14008 constant value is loaded is not represented by a base address held
14009 in a register or if the base address register is a frame or stack
14010 pointer. Additionally, return false if the base address of the
14011 loaded constant is a SYMBOL_REF but is not considered to be a
14013 (replace_swapped_load_constant): New function.
14014 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14015 loaded constant vector with a load of a swapped constant vector.
14017 2017-09-27 Carl Love <cel@us.ibm.com>
14019 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14020 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14021 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14024 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14026 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14027 first, always call autopref_rank_data otherwise.
14029 2017-09-27 Richard Biener <rguenther@suse.de>
14031 * graphite-scop-detection.c (find_scop_parameters): Move
14032 loop bound handling ...
14033 (gather_bbs::before_dom_children): ... here, avoiding the need
14034 to build scop_info->loop_nest.
14035 (record_loop_in_sese): Remove.
14036 * sese.h (sese_info_t::loop_nest): Remove.
14037 * sese.c (new_sese_info): Do not allocate loop_nest.
14038 (free_sese_info): Do not free loop_nest.
14040 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14043 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14044 lhs from calls if the lhs has addressable type.
14046 2017-09-27 Richard Biener <rguenther@suse.de>
14048 * graphite.h (scop::max_alias_set): New member.
14049 * graphite-scop-detection.c: Remove references to non-existing
14050 --param in comments.
14051 (build_alias_sets): Record the maximum alias set used for drs.
14052 (build_scops): Support zero as unlimited for
14053 --param graphite-max-arrays-per-scop.
14054 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14055 and inline into ...
14056 (build_poly_sr_1): ... here. Compute alias set based on the
14057 maximum alias set used for drs rather than
14058 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14060 2017-09-27 Richard Biener <rguenther@suse.de>
14062 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14063 --param loop-block-tile-size=0 to disable tiling.
14065 2017-09-27 Richard Biener <rguenther@suse.de>
14067 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14068 (graphite-max-nb-scop-params): Document special value zero.
14069 * domwalk.h (dom_walker::STOP): New symbolical constant.
14070 (dom_walker::dom_walker): Add optional parameter for bb to
14072 (dom_walker::~dom_walker): Declare.
14073 (dom_walker::before_dom_children): Document STOP return value.
14074 (dom_walker::m_user_bb_to_rpo): New member.
14075 (dom_walker::m_bb_to_rpo): Likewise.
14076 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14077 mapping here if not provided by the user.
14078 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14079 provided by the user.
14080 (dom_walker::STOP): Define.
14081 (dom_walker::walk): Do not compute bb to RPO mapping here.
14082 Support STOP return value from before_dom_children to stop
14084 * graphite-optimize-isl.c (optimize_isl): If the schedule
14085 is the same still generate code if -fgraphite-identity
14086 or -floop-parallelize-all are given.
14087 * graphite-scop-detection.c: Include cfganal.h.
14088 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14090 (gather_bbs::before_dom_children): Return STOP for BBs
14092 (build_scops): Compute bb to RPO mapping and pass it to
14093 the domwalk. Treat --param graphite-max-nb-scop-params=0
14094 as not limiting the number of params.
14095 * graphite.c (graphite_initialize): Remove limit on the
14096 number of basic-blocks in a function.
14097 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14098 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14099 default value of 10.
14101 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14103 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14104 Adjust code to eliminate needing to do the shift right 32-bits
14105 operation after XSCVDPSPN.
14107 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14109 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14110 ((X / Y) != 0 -> X >= Y): Likewise.
14112 2017-09-26 Carl Love <cel@us.ibm.com>
14114 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14115 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14116 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14117 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14118 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14119 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14120 definitions and overloading.
14121 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14122 statement for P9V_BUILTIN_XST_LEN_R.
14123 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14124 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14125 define_expand and define_insn for the instructions and builtins.
14126 * doc/extend.texi: Update the built-in documentation file for the new
14127 built-in functions.
14128 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14129 define_insn for the instructions
14131 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14134 * gcc/config/netbsd-protos.h: New file.
14135 * gcc/config/netbsd.c: New file.
14136 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14137 * gcc/config/t-netbsd: New file.
14138 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14139 (tmake_file) Add t-netbsd.
14140 (extra_objs) Add netbsd.o.
14142 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14146 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14148 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14150 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14151 sign extension from a vector register to a GPR by doing a 32-bit
14152 direct move and then an EXTSW.
14153 (extendsi<mode>2 splitter): Likewise.
14154 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14155 right or vector extract after doing XSCVDPSPN. Use
14156 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14158 (movdi_from_sf_zero_ext): Likewise.
14159 (reload_gpr_from_vsxsf): Likewise.
14160 (p8_mfvsrd_4_disf): Delete, no longer used.
14161 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14162 then needing to move the SFmode to a GPR to use the XSCVDPSP
14163 instruction instead of FRSP and XSCVDPSPN.
14164 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14165 it is adjacent to the other XSCVSPDP insns.
14166 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14167 SFmode to be in traditional Altivec registers.
14168 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14169 (vsx_xscvspdpn): Likewise.
14170 (vsx_xscvspdpn_scalar): Likewise.
14172 2017-09-26 Martin Jambor <mjambor@suse.cz>
14174 * tree-sra.c (compare_access_positions): Put integral types first,
14175 stabilize sorting of integral types, remove conditions putting
14176 non-full-precision integers last.
14177 (sort_and_splice_var_accesses): Disable scalarization if a
14178 non-integert would be represented by a non-full-precision integer.
14180 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14182 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14183 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14184 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14185 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14186 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14187 conditionals inside the function instead of around it. Call
14188 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14189 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14191 2017-09-26 Richard Biener <rguenther@suse.de>
14193 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14195 (scop_detection::build_scop_breadth): ... this. Removed.
14196 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14197 (scop_detection::harmful_stmt_in_bb): Likewise.
14198 (scop_detection::graphite_can_represent_stmt): Likewise.
14199 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14200 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14201 (scop_detection::can_represent_loop_1): ... this. Removed.
14202 (scop_detection::harmful_loop_in_region): Simplify after inlining
14203 the above and remove more quadraticness.
14204 (build_scops): Adjust.
14205 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14208 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14211 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14212 REGNO (base) == SP_REG if base is a REG.
14214 PR middle-end/35691
14215 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14216 if it is different SSA_NAME.
14217 (optimize_range_tests_cmp_bitwise): New function.
14218 (optimize_range_tests): Call it.
14220 2017-09-26 Richard Biener <rguenther@suse.de>
14222 PR tree-optimization/82321
14223 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14224 for the def being inside the loop.
14226 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14228 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14230 * config/s390/s390-builtins.def: Fix constraint on op4.
14232 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14234 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14235 independent expanders.
14236 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14237 ("vec_ordered", "vec_unordered"): New expanders.
14239 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14241 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14244 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14246 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14247 vec_unpacks_lo_v16qi.
14248 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14250 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14252 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14253 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14254 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14256 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14258 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14260 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14262 ("*vec_slb<mode>"): New insn pattern.
14263 ("vec_shr_<mode>"): New expander.
14264 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14265 and force the shift count operand to V16QImode.
14266 ("vec_srb<mode>"): Set shift count mode to V16QI.
14268 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14270 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14271 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14272 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14274 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14277 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14279 2017-09-26 Richard Biener <rguenther@suse.de>
14281 PR tree-optimization/82320
14282 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14285 2017-09-25 Jeff Law <law@redhat.com>
14287 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14288 prototype for new argument.
14289 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14290 mostly extracted from rs6000_emit_allocate_stack.
14291 (rs6000_emit_probe_stack_range_stack_clash): New function.
14292 (rs6000_emit_allocate_stack): Call
14293 rs6000_emit_probe_stack_range_stack_clash as needed.
14294 (rs6000_emit_probe_stack_range): Add additional argument
14295 to call to gen_probe_stack_range{si,di}.
14296 (output_probe_stack_range): New.
14297 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14298 (output_probe_stack_range_stack_clash): New.
14299 (rs6000_emit_prologue): Emit notes into dump file as requested.
14300 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14301 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14302 Add additional operand and pass it to output_probe_stack_range.
14304 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14306 PR tree-optimization/82163
14307 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14308 (checking_verify_loop_closed_ssa): New parameter.
14309 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14310 (check_loop_closed_ssa_stmt): Delete.
14311 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14312 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14313 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14316 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14318 * brig-builtins.def: Treat HSAIL barrier builtins as
14319 setjmp/longjump style functions.
14321 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14323 * target.def (constant_alignment): New hook.
14324 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14325 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14326 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14327 * doc/tm.texi: Regenerate.
14328 * targhooks.h (default_constant_alignment): Declare.
14329 (constant_alignment_word_strings): Likewise.
14330 * targhooks.c (default_constant_alignment): New function.
14331 (constant_alignment_word_strings): Likewise.
14332 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14333 instead of CONSTANT_ALIGNMENT.
14334 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14335 (force_const_mem): Likewise.
14336 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14337 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14338 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14339 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14340 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14342 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14343 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14344 constant_alignment_word_strings.
14345 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14346 (CONSTANT_ALIGNMENT): Likewise.
14347 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14348 (arm_constant_alignment): New function.
14349 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14350 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14351 constant_alignment_word_strings.
14352 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14353 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14354 constant_alignment_word_strings.
14355 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14356 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14357 (cris_constant_alignment): New function.
14358 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14359 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14360 (epiphany_constant_alignment): New function.
14361 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14362 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14363 constant_alignment_word_strings.
14364 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14365 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14366 constant_alignment_word_strings.
14367 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14368 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14369 constant_alignment_word_strings.
14370 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14371 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14372 * config/i386/i386.c (ix86_constant_alignment): Make static.
14373 Use the same interface as the target hook.
14374 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14375 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14376 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14377 constant_alignment_word_strings.
14378 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14379 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14380 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14381 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14382 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14383 constant_alignment_word_strings.
14384 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14385 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14386 constant_alignment_word_strings.
14387 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14388 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14389 constant_alignment_word_strings.
14390 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14391 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14393 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14394 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14395 * config/mips/mips.c (mips_constant_alignment): New function.
14396 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14397 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14398 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14399 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14400 (mmix_constant_alignment): Make static. Use the same interface
14401 as the target hook.
14402 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14403 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14404 constant_alignment_word_strings.
14405 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14406 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14407 constant_alignment_word_strings.
14408 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14409 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14410 constant_alignment_word_strings.
14411 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14412 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14413 (rs6000_constant_alignment): New function.
14414 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14415 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14416 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14417 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14418 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14419 (rs6000_constant_alignment): New function.
14420 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14421 * config/s390/s390.c (s390_constant_alignment): New function.
14422 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14423 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14424 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14425 constant_alignment_word_strings.
14426 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14427 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14428 (sparc_constant_alignment): New function.
14429 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14430 * config/spu/spu.c (spu_constant_alignment): New function.
14431 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14432 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14433 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14434 constant_alignment_word_strings.
14435 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14436 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14437 constant_alignment_word_strings.
14438 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14439 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14440 constant_alignment_word_strings.
14441 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14442 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14443 (visium_constant_alignment): New function.
14444 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14445 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14446 (xtensa_constant_alignment): New function.
14447 * system.h (CONSTANT_ALIGNMENT): Poison.
14449 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14451 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14452 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14453 (rs6000_builtin_valid_without_lhs): New helper function.
14454 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14455 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14457 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14459 * target.h (vec_perm_indices): Use unsigned short rather than
14461 (auto_vec_perm_indices): Likewise.
14462 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14463 Use unsigned int rather than unsigned char.
14464 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14466 2017-09-25 Richard Biener <rguenther@suse.de>
14468 * cfgloop.h (sort_sibling_loops): Declare.
14469 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14470 (sort_sibling_loops): New function sorting the sibling loop list
14472 * graphite.c (graphite_transform_loops): Sort sibling loops.
14474 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14476 * target.def (vec_perm_const_ok): Change sel parameter to
14478 * optabs-query.c (can_vec_perm_p): Update accordingly.
14479 * doc/tm.texi: Regenerate.
14480 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14481 auto_vec_perm_indices and remove separate nelt field.
14482 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14483 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14484 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14485 (aarch64_expand_vec_perm_const): Update accordingly.
14486 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14487 to vec_perm_indices.
14488 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14489 auto_vec_perm_indices and remove separate nelt field.
14490 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14491 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14492 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14494 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14495 to vec_perm_indices.
14496 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14497 sel to vec_perm_indices.
14498 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14499 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14500 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14502 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14505 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14508 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14509 on the FUNCTION_DECL function context if it has a DIE that is a
14512 2017-09-25 Richard Biener <rguenther@suse.de>
14514 PR tree-optimization/82285
14515 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14518 2017-09-25 Tom de Vries <tom@codesourcery.com>
14522 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14525 2017-09-25 Richard Biener <rguenther@suse.de>
14527 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14528 ISL errors other than isl_error_quota happen. Dump if the
14529 schedule is the same.
14530 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14531 errors instead of aborting inside ISL.
14533 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14536 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14537 of libgcc_eh for m64.
14538 * config/i386/darwin64.h: Likewise.
14540 2017-09-25 Richard Biener <rguenther@suse.de>
14542 PR middle-end/82144
14543 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14544 attribute for incomplete types nor twice for complete ones.
14546 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14549 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14550 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14553 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14556 * config/i386/i386.opt (mprefer-avx256): Use
14557 ix86_target_flags variable.
14558 * config/i386/i386.c (ix86_target_string): Move
14559 -mprefer-avx256 to flag2_opts.
14561 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14563 PR middle-end/35691
14564 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14565 and x != -1 | y != -1 into (x & y) != -1.
14567 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14569 * config.gcc: Add new case statement to set
14570 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14571 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14572 s390x-*-linux* case statements. Added aarch64 to the list of
14573 supported architectures.
14575 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14577 PR tree-optimization/82289
14578 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14579 STMT_VINFO_RELEVANT_P.
14581 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14582 Alan Hayward <alan.hayward@arm.com>
14583 David Sherwood <david.sherwood@arm.com>
14585 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14586 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14587 (extract_range_from_binary_expr_1): Don't call
14588 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14590 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14591 Alan Hayward <alan.hayward@arm.com>
14592 David Sherwood <david.sherwood@arm.com>
14594 * target.def (preferred_vector_alignment): New hook.
14595 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14597 * doc/tm.texi: Regenerate.
14598 * targhooks.h (default_preferred_vector_alignment): Declare.
14599 * targhooks.c (default_preferred_vector_alignment): New function.
14600 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14602 (DR_TARGET_ALIGNMENT): New macro.
14603 (aligned_access_p): Update commentary.
14604 (vect_known_alignment_in_bytes): New function.
14605 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14607 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14608 Calculate the misalignment based on the target alignment rather than
14610 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14611 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14612 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14613 the target alignment, rather than masking the element misalignment
14614 with the number of elements in a vector. Also use the target
14615 alignment when calculating the maximum number of peels.
14616 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14617 instead of TYPE_ALIGN_UNIT.
14618 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14619 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14620 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14621 (vect_create_data_ref_ptr): Likewise.
14622 (vect_setup_realignment): Realign by ANDing with
14623 -DR_TARGET_MISALIGNMENT.
14624 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14625 the number of peels based on DR_TARGET_ALIGNMENT.
14626 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14627 with the guaranteed alignment boundary when deciding whether
14629 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14630 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14631 (ensure_base_align): Remove stmt_info parameter. Get the
14632 target base alignment from DR_TARGET_ALIGNMENT.
14633 (vectorizable_store): Update call accordingly. Interpret
14634 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14636 (vectorizable_load): Likewise.
14638 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14639 Alan Hayward <alan.hayward@arm.com>
14640 David Sherwood <david.sherwood@arm.com>
14642 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14643 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14644 (vect_enhance_data_refs_alignment): Likewise.
14646 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14648 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14649 error. Only quit immediately if parsing is complete.
14650 (BEGIN): Initialize fatal_err and parse_done.
14651 (begin fpu, end fpu): Check number of arguments.
14652 (begin arch, end arch): Likewise.
14653 (begin cpu, end cpu): Likewise.
14654 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14655 (optalias): Likewise.
14657 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14659 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14660 * config/arm/arm-isa.h: Delete. Move definitions to ...
14661 * arm-cpus.in: ... here. Use new feature and fgroup values.
14662 * config/arm/arm.c (arm_option_override): Use lower case for feature
14664 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14665 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14666 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14668 (print_isa_bits_for): New function.
14669 (gen_isa): New function.
14670 (gen_comm_data): Use print_isa_bits_for.
14671 (define feature): New keyword.
14672 (define fgroup): New keyword.
14673 * config/arm/t-arm (TM_H): Remove.
14674 (GTM_H): Add arm-isa.h.
14675 (arm-isa.h): Add rule to generate file.
14676 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14677 case for feature bit names.
14679 2017-09-22 Richard Biener <rguenther@suse.de>
14681 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14683 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14684 print of no dependency loops ...
14685 * graphite.c (graphite_transform_loops): ... here.
14686 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14688 (same_close_phi_node, remove_duplicate_close_phi,
14689 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14690 (canonicalize_loop_closed_ssa): ... here and simplify.
14691 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14692 (optimize_isl): Use dump_printf_loc to tell when we stopped
14693 optimizing because of an ISL timeout.
14695 2017-09-22 Richard Biener <rguenther@suse.de>
14697 PR tree-optimization/82291
14698 * tree-if-conv.c (predicate_mem_writes): Make sure to
14699 remove writes in blocks predicated with false.
14701 2017-09-22 Richard Biener <rguenther@suse.de>
14703 * sese.c: Include cfganal.h.
14704 (if_region_set_false_region): Remove.
14705 (create_if_region_on_edge): Likewise.
14706 (move_sese_in_condition): Re-implement without destroying
14709 2017-09-22 Richard Biener <rguenther@suse.de>
14711 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14712 Verify both BBs contain loop PHI nodes before dispatching to
14713 copy_loop_phi_args.
14714 (graphite_regenerate_ast_isl): Do not recompute dominators,
14715 do not verify three times. Restructure for clarity.
14716 * graphite-scop-detection.c (same_close_phi_node,
14717 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14718 defined_in_loop_p, canonicalize_loop_closed_ssa,
14719 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14720 checking and SSA rewrite, move to ...
14721 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14722 (graphite_initialize): Do not pass in ctx, do not reset the
14723 SCEV cache, compute only dominators.
14724 (graphite_transform_loops): Allocate ISL ctx after
14725 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14726 Maintain post-dominators only around build_scops.
14727 * sese.c (if_region_set_false_region): Make static. Free
14728 and recompute dominators.
14729 (move_sese_in_condition): Assert we don't get called with
14730 post-dominators computed.
14731 * sese.h (if_region_set_false_region): Remove.
14733 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14735 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14736 mode attribute for TARGET_AVX512VL.
14738 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14740 * config/i386/i386.opt (mprefer-avx256): New option.
14741 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14743 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14744 for TARGET_PREFER_AVX256.
14745 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14747 2017-09-21 Jeff Law <law@redhat.com>
14749 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14750 Fix dump output if the only stack space is for pushed registers.
14752 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14754 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14757 2017-09-21 Martin Sebor <msebor@redhat.com>
14760 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14761 code (in C++) or code that triggers warnings.
14763 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14765 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14767 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14769 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14770 * sched-rgn.c: Ditto.
14771 * sel-sched-ir.c: Ditto.
14773 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14775 * toplev.h (set_random_seed): Adjust return type.
14776 * toplev.c (init_local_tick): Move eager initialization of random_seed
14777 to get_random_seed. Adjust comment.
14778 (init_random_seed): Inline to get_random_seed, delete.
14779 (get_random_seed): Initialize random_seed lazily.
14780 (set_random_seed): Do not return previous value.
14781 (print_switch_value): Do not call get_random_seed.
14783 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14785 * cgraph.c (delete_function_version): New, broken out from...
14786 (cgraph_node::delete_function_version): ...here. Rename to
14787 cgraph_node::delete_function_version_by_decl. Update all uses.
14788 (cgraph_node::remove): Call delete_function_version.
14790 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14793 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14794 VAR_DECLs to which addressable non-volatile parameters are mapped
14795 and for id->retvar after the return value assignment. Clear
14796 id->retval and id->retbnd after inlining.
14798 2017-09-21 Richard Biener <rguenther@suse.de>
14800 PR tree-optimization/82276
14801 PR tree-optimization/82244
14802 * tree-vrp.c (build_assert_expr_for): Set
14803 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14805 (remove_range_assertions): Revert earlier change.
14807 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14810 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14812 2017-09-21 Richard Biener <rguenther@suse.de>
14814 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14815 Restore valid IL after code generation errors.
14816 * graphite.c (graphite_transform_loops): Diagnose code
14817 generation issues as MSG_MISSED_OPTIMIZATION and continue
14818 with processing SCOPs.
14820 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14821 Alan Hayward <alan.hayward@arm.com>
14822 David Sherwood <david.sherwood@arm.com>
14824 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14825 rather than choosing between plus_constant and gen_rtx_<CODE>.
14826 * expr.c (emit_push_insn): Likewise.
14827 (expand_expr_real_2): Likewise.
14829 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14830 Alan Hayward <alan.hayward@arm.com>
14831 David Sherwood <david.sherwood@arm.com>
14833 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14835 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14836 Alan Hayward <alan.hayward@arm.com>
14837 David Sherwood <david.sherwood@arm.com>
14839 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14840 calling tree_to_uhwi.
14842 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14843 Alan Hayward <alan.hayward@arm.com>
14844 David Sherwood <david.sherwood@arm.com>
14846 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14847 INTEGER_CST rather than a negative test for ADDR_EXPR.
14849 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14850 Alan Hayward <alan.hayward@arm.com>
14851 David Sherwood <david.sherwood@arm.com>
14853 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14854 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14856 2017-09-21 Richard Biener <rguenther@suse.de>
14858 PR tree-optimization/71351
14859 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14860 graphite_create_new_loop_guard): Remove, fold remaining parts
14862 (translate_isl_ast_node_for): ... here and simplify.
14864 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14867 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14868 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14869 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14870 alternative always use QI mode, for -Os imov (=R,R) alternative
14871 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14874 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14875 Jeff Law <law@redhat.com>
14877 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14878 (allocate_stack_space): New function, partially extracted from
14879 s390_emit_prologue.
14880 (s390_emit_prologue): Track offset to most recent stack probe.
14881 Code to allocate space moved into allocate_stack_space.
14882 Dump actions when no stack is allocated.
14883 (s390_prologue_plus_offset): New function.
14884 (s390_emit_stack_probe): Likewise.
14886 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14888 * common.opt (Wa, Wl, Wp, g, gz=): Add
14890 (gno-column-info): Remove.
14891 (gcolumn-info): Drop RejectNegative.
14892 (gno-): New prefix.
14893 (gno-record-gcc-switches): Remove.
14894 (grecord-gcc-switches): Drop RejectNegative.
14895 (gno-split-dwarf): Remove.
14896 (gsplit-dwarf): Drop RejectNegative.
14897 (gno-strict-dwarf): Remove.
14898 (gstrict-dwarf): Drop RejectNegative.
14899 * config/darwin.opt (gfull, gused): Add RejectNegative.
14900 * dwarf2out.c (gen_producer_string): Drop
14901 gno-record-gcc-switches handler.
14902 * optc-gen.awk: Add g to prefixes with negative forms.
14903 * opts-common.c (remapping_prefix_p): New.
14904 (find_opt): Check it.
14905 (generate_canonical_option): Test g prefix.
14906 (option_map): Add -gno- mapping.
14907 (add_misspelling_candidates): Check remapping_prefix_p.
14909 2017-09-20 Jeff Law <law@redhat.com>
14911 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14912 thinko in stack clash protection support.
14914 * explow.c (compute_stack_clash_protection_loop_data): Use
14915 CONST_INT_P instead of explicit test. Verify object is a
14916 CONST_INT_P before looking at INTVAL.
14917 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14918 instead of explicit test.
14920 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14923 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14924 address instead of to r1 and r11.
14926 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14928 * config.gcc: Support "knm".
14929 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14930 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14932 * config/i386/i386.c (m_KNM): Define.
14933 (processor_target_table): Add "knm".
14935 (ix86_option_override_internal): Add "knm".
14936 (ix86_issue_rate): Add PROCESSOR_KNM.
14937 (ix86_adjust_cost): Ditto.
14938 (ia32_multipass_dfa_lookahead): Ditto.
14939 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14940 (fold_builtin_cpu): Add M_INTEL_KNM.
14941 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14942 (processor_type): Add PROCESSOR_KNM.
14943 * config/i386/x86-tune.def: Add m_KNM.
14944 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14946 2017-09-20 Richard Biener <rguenther@suse.de>
14948 PR tree-optimization/80213
14949 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14950 are allowed in empty BBs as well.
14951 (canonicalize_loop_closed_ssa): Also look for other complex
14953 (scop_detection::get_sese): Include the loop-closed PHI block
14955 (scop_detection::merge_sese): Remove code adding extra blocks.
14956 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14957 (build_scops): Assert the final returned scop is invalid.
14959 2017-09-20 Richard Biener <rguenther@suse.de>
14961 PR tree-optimization/82264
14962 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14964 (vn_phi_lookup): Likewise.
14965 (vn_phi_insert): Likewise.
14967 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14969 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14970 that fits into uhwi or shwi, add DW_AT_const_value regardless
14971 of early_dwarf without going through RTL, using add_AT_unsigned
14974 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14975 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14976 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14977 (DEBUG_MACRO_SECTION): Likewise.
14978 (DEBUG_LTO_MACRO_SECTION): Likewise.
14979 (DEBUG_STR_DWO_SECTION): Likewise.
14980 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14981 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14982 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14983 (DEBUG_LTO_LINE_STR_SECTION): Define.
14984 (init_sections_and_labels): Initialize debug_line_str_section
14985 variable. Initialize debug_loc_section for -gdwarf-5 to
14986 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14988 2017-09-20 Richard Biener <rguenther@suse.de>
14990 * graphite-sese-to-poly.c (extract_affine): Properly handle
14991 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14993 2017-09-20 Richard Biener <rguenther@suse.de>
14995 PR tree-optimization/81373
14996 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14997 Force SESE live-out defs to be handled even if they are
15000 2017-09-19 Jeff Law <law@redhat.com>
15002 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
15003 nothing for stack adjustments with REG_STACK_CHECK.
15004 * sched-deps.c (parse_add_or_inc): Reject insns with
15005 REG_STACK_CHECK from dependency breaking.
15006 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15007 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15008 * reg-notes.def (STACK_CHECK): New note.
15010 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15011 (ix86_expand_prologue): Dump stack clash info as needed.
15012 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15014 * function.c (dump_stack_clash_frame_info): New function.
15015 * function.h (dump_stack_clash_frame_info): Prototype.
15016 (enum stack_clash_probes): New enum.
15018 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15019 flag_stack_clash_protection.
15020 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15021 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15022 (arm_frame_pointer_required): Likewise.
15023 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15024 (ia64_expand_prologue): Likewise.
15025 * config/mips/mips.c (mips_expand_prologue): Likewise.
15026 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15027 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15028 (sparc_flat_expand_prologue): Likewise.
15029 * config/spu/spu.c (spu_expand_prologue): Likewise.
15031 * explow.c: Include "params.h".
15032 (anti_adjust_stack_and_probe_stack_clash): New function.
15033 (get_stack_check_protect): Likewise.
15034 (compute_stack_clash_protection_loop_data): Likewise.
15035 (emit_stack_clash_protection_loop_start): Likewise.
15036 (emit_stack_clash_protection_loop_end): Likewise.
15037 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15038 Use anti_adjust_stack_and_probe_stack_clash.
15039 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15040 (emit_stack_clash_protection_loop_start): Likewise.
15041 (emit_stack_clash_protection_loop_end): Likewise.
15042 * rtl.h (get_stack_check_protect): Prototype.
15043 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15044 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15045 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15047 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15049 * doc/tm.texi: Rebuilt.
15050 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15051 get_stack_check_protect.
15052 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15053 * config/arm/arm.c (arm_expand_prologue): Likewise.
15054 (arm_frame_pointer_required): Likewise.
15055 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15056 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15057 * config/mips/mips.c (mips_expand_prologue): Likewise.
15058 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15059 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15060 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15061 (sparc_flat_expand_prologue): Likewise.
15063 * common.opt (-fstack-clash-protection): New option.
15064 * flag-types.h (enum stack_check_type): Note difference between
15065 -fstack-check= and -fstack-clash-protection.
15066 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15067 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15068 * toplev.c (process_options): Issue warnings/errors for cases
15069 not handled with -fstack-clash-protection.
15070 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15071 (-fstack-check): Note additional problem with -fstack-check=generic.
15072 Note that -fstack-check is primarily for Ada and refer users
15073 to -fstack-clash-protection for stack-clash-protection.
15074 Document new params for stack clash protection.
15076 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15078 * config/i386/i386.c (ix86_split_long_move): Do not handle
15079 address used for LEA in a special way.
15081 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15083 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15085 2017-09-19 Martin Sebor <msebor@redhat.com>
15088 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15089 of incompatible types.
15091 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15093 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15094 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15095 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15096 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15098 2017-09-19 Richard Biener <rguenther@suse.de>
15100 PR tree-optimization/82244
15101 * tree-vrp.c (remove_range_assertions): Do not propagate
15102 a constant to abnormals but replace the assert with a copy.
15104 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15106 PR rtl-optimization/57878
15107 PR rtl-optimization/68988
15108 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15109 avoidance test involving non_reload_pseudos. Move frequency test
15110 below the general fragmentation avoidance test.
15112 2017-09-19 Richard Biener <rguenther@suse.de>
15114 PR tree-optimization/69728
15115 * graphite-sese-to-poly.c (schedule_error): New global.
15116 (add_loop_schedule): Handle empty domain by failing the
15118 (build_original_schedule): Handle schedule_error.
15120 2017-09-19 Richard Biener <rguenther@suse.de>
15122 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15123 Do not iterate to sibling loops but only to siblings of inner
15126 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15129 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15130 that the registers are different.
15132 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15134 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15135 to processor_model and "amdfam17h" to arch_names_table.
15136 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15138 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15141 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15143 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15144 Alan Hayward <alan.hayward@arm.com>
15145 David Sherwood <david.sherwood@arm.com>
15147 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15149 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15151 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15152 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15153 vect_schedule_slp_instance.
15154 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15155 Update call to vect_slp_analyze_node_operations. Simplify return
15157 (vect_slp_analyze_bb_1): Update call accordingly.
15158 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15159 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15160 (vect_schedule_slp): Update call accordingly.
15162 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15163 Alan Hayward <alan.hayward@arm.com>
15164 David Sherwood <david.sherwood@arm.com>
15166 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15167 with types that aren't in fact scalar.
15169 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15171 * tree-vect-slp.c (vect_record_max_nunits): New function,
15173 (vect_build_slp_tree_1): ...here.
15174 (vect_build_slp_tree_2): Call it for phis too.
15176 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15178 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15179 to vect_get_vec_def_for_operand when getting the mask operand.
15181 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15182 Alan Hayward <alan.hayward@arm.com>
15183 David Sherwood <david.sherwood@arm.com>
15185 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15188 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15189 Alan Hayward <alan.hayward@arm.com>
15190 David Sherwood <david.sherwood@arm.com>
15192 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15193 calculation for vector booleans.
15195 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15196 Alan Hayward <alan.hayward@arm.com>
15197 David Sherwood <david.sherwood@arm.com>
15199 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15201 (vect_transform_stmt): ...here.
15202 (vect_analyze_stmt): Use it instead of calling
15203 vectorizable_live_operation directly.
15205 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15207 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15208 non-SIMT targets in acc vector loops.
15210 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15212 * configure.ac: Add arc and check if assembler supports gdwarf2.
15213 * configure: Regenerate.
15215 2017-09-18 Richard Biener <rguenther@suse.de>
15217 PR tree-optimization/82220
15218 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15219 epilogue niters from the min_profitable_iters compute.
15221 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15224 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15225 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15226 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15228 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15231 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15232 switching to a new text section.
15234 2017-09-18 Richard Biener <rguenther@suse.de>
15236 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15238 (build_alias_set): Reject aliases with no access function.
15240 2017-09-18 Richard Biener <rguenther@suse.de>
15242 PR tree-optimization/79622
15243 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15245 (build_cross_bb_scalars_use): Likewise.
15247 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15249 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15251 2017-09-18 Alan Modra <amodra@gmail.com>
15254 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15255 stack_pointer_rtx for count 0. Update comments. Break up
15256 large rtl expression.
15258 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15260 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15261 Increase to 20 bytes.
15262 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15263 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15264 or avx version of the stub.
15266 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15269 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15270 compute the minimum stack alignment. Also update preferred stack
15271 boundary for leaf functions.
15273 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15275 PR tree-optimization/82228
15276 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15279 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15281 * common/config/nds32/nds32-common.c
15282 (nds32_option_optimization_table): Refine formatting.
15283 (nds32_option_optimization_table): Use -fsched-pressure and
15284 -fomit-frame-pointer for specific optimization level.
15286 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15288 * config/nds32/nds32.c: Refine formatting and comments.
15289 * config/nds32/nds32.h: Likewise.
15290 * config/nds32/nds32.md: Likewise.
15291 * config/nds32/nds32-cost.c: Likewise.
15292 * config/nds32/nds32-isr.c: Likewise.
15293 * config/nds32/nds32-md-auxiliary.c: Likewise.
15294 * config/nds32/nds32-multiple.md: Likewise.
15295 * config/nds32/nds32-predicates.c: Likewise.
15297 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15298 Jakub Jelinek <jakub@redhat.com>
15300 Add support for -std=c++2a.
15301 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15303 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15305 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15308 * doc/extend.texi (Common Function Attributes): Add
15309 references to ARM, AArch64, and S/390 specific attributes.
15310 (Function Specific Option Pragmas): Add AArch64 and S/390
15311 to list of back ends that support the target pragma.
15313 2017-09-15 Nathan Sidwell <nathan@acm.org>
15315 * doc/standards.texi: Fix C++17 description. Update URLs for
15318 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15320 * common.opt (Wcast-align=strict): New warning option.
15321 * doc/invoke.texi: Document -Wcast-align=strict.
15323 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15325 * cgraph.h (cgraph_thunk_info): Add comments.
15326 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15327 assert for VIRTUAL_* arguments stricter.
15329 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15331 PR tree-optimization/71026
15332 * match.pd: Move RDIV patterns from fold-const.c
15333 * fold-const.c (distribute_real_division): Removed.
15334 (fold_binary_loc): Remove calls to distribute_real_divison.
15336 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15338 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15339 c++1z and gnu++1z as deprecated. Change other references to
15340 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15341 Change -Wc++1z-compat to -Wc++17-compat.
15342 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15343 * dwarf2out.c (highest_c_language): Handle C++17.
15344 (gen_compile_unit_die): Likewise.
15346 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15348 PR rtl-optimization/82192
15349 * combine.c (make_extraction): Don't look through non-paradoxical
15350 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15353 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15354 Alan Hayward <alan.hayward@arm.com>
15355 David Sherwood <david.sherwood@arm.com>
15357 * target.def (function_arg_offset): New hook.
15358 * targhooks.h (default_function_arg_offset): Declare.
15359 * targhooks.c (default_function_arg_offset): New function.
15360 * function.c (locate_and_pad_parm): Use
15361 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15362 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15363 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15364 * doc/tm.texi: Regenerate.
15365 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15366 * config/spu/spu.c (spu_function_arg_offset): New function.
15367 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15368 * system.h (FUNCTION_ARG_OFFSET): Poison.
15370 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15371 Alan Hayard <alan.hayward@arm.com>
15372 David Sherwood <david.sherwood@arm.com>
15374 * target.def (truly_noop_truncation): New hook.
15375 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15376 than TRULY_NOOP_TRUNCATION.
15377 * hooks.h (hook_bool_uint_uint_true): Declare.
15378 * hooks.c (hook_bool_uint_uint_true): New function.
15379 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15380 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15381 * doc/tm.texi: Regenerate.
15382 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15383 rather than TRULY_NOOP_TRUNCATION in comments.
15384 (simplify_comparison): Likewise.
15385 (record_truncated_value): Likewise.
15386 * expmed.c (extract_bit_field_1): Likewise.
15387 (extract_split_bit_field): Likewise.
15388 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15389 instead of TRULY_NOOP_TRUNCATION.
15390 * function.c (assign_parm_setup_block): Likewise.
15391 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15392 * rtlhooks.c: Include target.h.
15393 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15394 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15395 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15396 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15397 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15398 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15399 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15400 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15401 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15402 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15403 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15404 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15405 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15406 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15407 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15408 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15409 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15410 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15411 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15412 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15413 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15414 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15415 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15416 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15417 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15418 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15419 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15420 rather than TRULY_NOOP_TRUNCATION in comments.
15421 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15422 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15423 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15424 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15425 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15426 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15427 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15428 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15429 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15430 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15431 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15432 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15433 rather than TRULY_NOOP_TRUNCATION in comments.
15434 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15435 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15436 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15437 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15438 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15439 TRULY_NOOP_TRUNCATION condition.
15440 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15441 (TRULY_NOOP_TRUNCATION): Delete.
15442 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15443 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15444 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15445 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15446 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15447 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15448 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15449 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15450 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15451 rather than TRULY_NOOP_TRUNCATION in comments.
15452 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15453 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15454 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15455 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15456 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15457 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15459 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15462 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15463 (*cmp_ior): Likewise.
15464 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15465 (*ior_scc_scc_cmp): Likewise.
15466 (*and_scc_scc): Likewise.
15467 (*and_scc_scc_cmp): Likewise.
15469 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15470 Alan Hayard <alan.hayward@arm.com>
15471 David Sherwood <david.sherwood@arm.com>
15473 * target.def (can_change_mode_class): New hook.
15474 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15475 (hard_regno_nregs): Likewise.
15476 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15477 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15478 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15479 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15480 (LOAD_EXTEND_OP): Update accordingly.
15481 * doc/tm.texi: Regenerate.
15482 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15483 CANNOT_CHANGE_MODE_CLASS.
15484 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15485 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15486 * combine.c (simplify_set): Update accordingly.
15487 * emit-rtl.c (validate_subreg): Likewise.
15488 * recog.c (general_operand): Likewise.
15489 * regcprop.c (mode_change_ok): Likewise.
15490 * reload1.c (choose_reload_regs): Likewise.
15491 (inherit_piecemeal_p): Likewise.
15492 * rtlanal.c (simplify_subreg_regno): Likewise.
15493 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15494 instead of CANNOT_CHANGE_MODE_CLASS.
15495 (reload_cse_simplify_operands): Likewise.
15496 * reload.c (push_reload): Use targetm.can_change_mode_class
15497 instead of CANNOT_CHANGE_MODE_CLASS.
15498 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15499 REG_CANNOT_CHANGE_MODE_P.
15500 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15501 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15502 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15503 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15504 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15505 (arm_can_change_mode_class): New function.
15506 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15507 than CANNOT_CHANGE_MODE_CLASS in comments.
15508 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15509 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15510 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15511 (ix86_can_change_mode_class): ...this new function, inverting the
15512 sense of the return value.
15513 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15514 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15515 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15516 (ia64_can_change_mode_class): New function.
15517 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15518 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15519 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15520 (m32c_can_change_mode_class): ...this new function, inverting the
15521 sense of the return value.
15522 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15523 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15524 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15525 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15526 (mips_can_change_mode_class): ...this new function, inverting the
15527 sense of the return value.
15528 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15529 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15530 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15531 (msp430_can_change_mode_class): New function.
15532 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15533 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15534 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15535 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15536 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15537 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15538 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15539 (pa_cannot_change_mode_class): Replace with...
15540 (pa_can_change_mode_class): ...this new function, inverting the
15541 sense of the return value.
15542 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15543 than CANNOT_CHANGE_MODE_CLASS in comments.
15544 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15545 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15546 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15547 (pdp11_cannot_change_mode_class): Replace with...
15548 (pdp11_can_change_mode_class): ...this new function, inverting the
15549 sense of the return value.
15550 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15551 * config/powerpcspe/powerpcspe-protos.h
15552 (rs6000_cannot_change_mode_class_ptr): Delete.
15553 * config/powerpcspe/powerpcspe.c
15554 (rs6000_cannot_change_mode_class_ptr): Delete.
15555 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15556 (rs6000_option_override_internal): Assign to
15557 targetm.can_change_mode_class instead of
15558 rs6000_cannot_change_mode_class_ptr.
15559 (rs6000_cannot_change_mode_class): Replace with...
15560 (rs6000_can_change_mode_class): ...this new function, inverting the
15561 sense of the return value.
15562 (rs6000_debug_cannot_change_mode_class): Replace with...
15563 (rs6000_debug_can_change_mode_class): ...this new function.
15564 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15565 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15566 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15567 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15568 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15570 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15571 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15572 (rs6000_option_override_internal): Assign to
15573 targetm.can_change_mode_class instead of
15574 rs6000_cannot_change_mode_class_ptr.
15575 (rs6000_cannot_change_mode_class): Replace with...
15576 (rs6000_can_change_mode_class): ...this new function, inverting the
15577 sense of the return value.
15578 (rs6000_debug_cannot_change_mode_class): Replace with...
15579 (rs6000_debug_can_change_mode_class): ...this new function.
15580 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15581 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15582 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15583 (s390_can_change_mode_class): ...this new function, inverting the
15584 sense of the return value.
15585 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15586 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15587 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15588 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15589 (sh_cannot_change_mode_class): Replace with...
15590 (sh_can_change_mode_class): ...this new function, inverting the
15591 sense of the return value.
15592 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15593 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15594 (sparc_can_change_mode_class): New function.
15595 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15596 * config/spu/spu.c (spu_can_change_mode_class): New function.
15597 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15598 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15599 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15600 (visium_can_change_mode_class): New function.
15601 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15603 2017-09-15 Richard Biener <rguenther@suse.de>
15605 PR tree-optimization/82217
15606 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15607 but not undefined case.
15609 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15612 * postreload.c (reload_cse_simplify_operands): Skip
15613 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15615 2017-09-15 Richard Biener <rguenther@suse.de>
15617 PR tree-optimization/68823
15618 * graphite-scop-detection.c (build_alias_set): If we have a
15619 possible dependence check whether we can handle them by just
15620 looking at the DRs DR_ACCESS_FNs.
15621 (build_scops): If build_alias_set fails, fail the SCOP.
15623 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15625 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15626 to support float128 built-in functions that require the ISA 3.0
15628 (BU_FLOAT128_3_HW): Likewise.
15629 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15630 built-in functions.
15631 (FMAF128): Likewise.
15633 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15634 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15635 floating point instructions.
15636 (rs6000_invalid_builtin): Likewise.
15637 (rs6000_builtin_mask_names): Likewise.
15638 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15639 (RS6000_BTM_FLOAT128_HW): Likewise.
15640 (RS6000_BTM_COMMON): Likewise.
15641 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15643 * doc/extend.texi (RS/6000 built-in functions): Document the
15644 IEEE 128-bit floating point square root and fused multiply-add
15645 built-in functions.
15647 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15649 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15650 reg (r2) isn't in the set of registers defined in the prologue.
15652 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15653 Alan Hayward <alan.hayward@arm.com>
15654 David Sherwood <david.sherwood@arm.com>
15656 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15657 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15658 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15659 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15660 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15662 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15663 max_vectorization_factor.
15664 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15666 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15667 Alan Hayward <alan.hayward@arm.com>
15668 David Sherwood <david.sherwood@arm.com>
15670 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15671 (vect_worthwhile_without_simd_p): Declare.
15672 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15673 (vectorizable_reduction): Use it.
15674 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15675 (vectorizable_operation): Likewise.
15677 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15678 Alan Hayward <alan.hayward@arm.com>
15679 David Sherwood <david.sherwood@arm.com>
15681 * tree-vectorizer.h (vect_get_num_copies): New function.
15682 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15683 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15684 (vectorizable_induction): Likewise.
15685 (vectorizable_live_operation): Likewise.
15686 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15687 (vectorizable_bswap): Likewise.
15688 (vectorizable_call): Likewise.
15689 (vectorizable_conversion): Likewise.
15690 (vectorizable_assignment): Likewise.
15691 (vectorizable_shift): Likewise.
15692 (vectorizable_operation): Likewise.
15693 (vectorizable_store): Likewise.
15694 (vectorizable_load): Likewise.
15695 (vectorizable_condition): Likewise.
15696 (vectorizable_comparison): Likewise.
15697 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15699 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15700 Alan Hayward <alan.hayward@arm.com>
15701 David Sherwood <david.sherwood@arm.com>
15703 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15704 of vect_init_vector.
15706 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15707 Alan Hayward <alan.hayward@arm.com>
15708 David Sherwood <david.sherwood@arm.com>
15710 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15711 an inline wrapper that provides a location.
15712 (gimple_build_vector): Likewise.
15713 * gimple-fold.c (gimple_build_vector_from_val): New function.
15714 (gimple_build_vector): Likewise.
15715 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15716 functions to build the initial value. Always return a gimple value.
15717 (get_initial_defs_for_reduction): Likewise. Only compute
15719 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15720 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15721 (vectorizable_induction): Use gimple_build_vector rather than
15724 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15725 Alan Hayward <alan.hayward@arm.com>
15726 David Sherwood <david.sherwood@arm.com>
15728 * target.h (vec_perm_indices): New typedef.
15729 (auto_vec_perm_indices): Likewise.
15730 * optabs-query.h: Include target.h
15731 (can_vec_perm_p): Take a vec_perm_indices *.
15732 * optabs-query.c (can_vec_perm_p): Likewise.
15733 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15734 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15735 * tree-vect-generic.c (lower_vec_perm): Likewise.
15736 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15737 (vect_grouped_load_supported): Likewise.
15738 (vect_shift_permute_load_chain): Likewise.
15739 (vect_permute_store_chain): Use auto_vec_perm_indices.
15740 (vect_permute_load_chain): Likewise.
15741 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15742 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15743 Update uses of can_vec_perm_p.
15744 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15745 mode with a number of elements. Take a vec_perm_indices *.
15746 (vect_create_epilog_for_reduction): Update accordingly.
15747 Use auto_vec_perm_indices.
15748 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15749 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15750 (vect_transform_slp_perm_load): Likewise.
15751 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15752 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15753 (vect_gen_perm_mask_checked): Likewise.
15754 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15755 (vect_gen_perm_mask_checked): Likewise.
15756 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15757 (vectorizable_store): Likewise.
15758 (vectorizable_load): Likewise.
15759 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15760 (vectorizable_bswap): Likewise.
15762 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15763 Alan Hayward <alan.hayward@arm.com>
15764 David Sherwood <david.sherwood@arm.com>
15766 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15767 * tree.c (build_vector): Likewise.
15768 (build_vector_from_ctor): Update accordingly.
15769 (build_vector_from_val): Likewise.
15770 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15771 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15772 * tree-vect-generic.c (add_rshift): Likewise.
15773 (expand_vector_divmod): Likewise.
15774 (optimize_vector_constructor): Likewise.
15775 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15776 (vect_transform_slp_perm_load): Likewise.
15777 (vect_schedule_slp_instance): Likewise.
15778 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15779 (vectorizable_call): Likewise.
15780 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15781 * expmed.c (make_tree): Likewise.
15782 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15783 a vector passed to build_vector.
15784 (fold_convert_const): Likewise.
15785 (exact_inverse): Likewise.
15786 (fold_ternary_loc): Likewise.
15787 (fold_relational_const): Likewise.
15788 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15789 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15790 (const_unop): Likewise. Store the reduction accumulator in a
15791 variable rather than an array.
15792 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15793 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15794 the new vector, rather than constructing it after the input arrays.
15795 (native_interpret_vector): Use auto_vec<tree> when building
15796 a vector passed to build_vector. Add elements in order.
15797 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15798 auto_vec<tree> when building a vector passed to build_vector.
15799 (vect_create_epilog_for_reduction): Likewise.
15800 (vectorizable_induction): Likewise.
15801 (get_initial_def_for_reduction): Likewise. Fix indentation of
15803 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15805 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15806 passed to build_vector.
15808 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15809 Alan Hayward <alan.hayward@arm.com>
15810 David Sherwood <david.sherwood@arm.com>
15812 * tree-core.h (tree_base::u): Add an "nelts" field.
15813 (tree_vector): Use VECTOR_CST_NELTS as the length.
15814 * tree.c (tree_size): Likewise.
15815 (make_vector): Initialize VECTOR_CST_NELTS.
15816 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15817 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15818 TYPE_VECTOR_SUBPARTS.
15819 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15820 as the number of units, setting it from VECTOR_CST_NELTS.
15821 (const_vector_from_tree): Likewise.
15822 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15823 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15824 (fold_negate_expr_1): Likewise.
15825 (fold_convert_const): Likewise.
15826 (const_binop): Likewise. Differentiate the number of output and
15828 (const_unop): Likewise.
15829 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15830 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15831 in cases that did the opposite.
15833 2017-09-14 Richard Biener <rguenther@suse.de>
15835 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15838 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15840 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15842 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15845 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15846 if and where to split a bb, except for splitting before debug insn
15847 sequences followed by non-label real insn. Delete debug insns
15848 in between basic blocks.
15850 * combine.c (make_compound_operation_int): Formatting fixes.
15852 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15853 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15854 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15855 * config/sol2.h (LINK_EH_SPEC): Likewise.
15856 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15857 * config/s390/linux.h (LINK_SPEC): Likewise.
15858 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15859 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15860 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15861 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15862 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15863 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15864 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15865 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15867 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15869 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15871 (ENDFILE_LINUX_SPEC): Likewise.
15872 (LINK_EH_SPEC): Likewise.
15873 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15874 (LINK_OS_LINUX_SPEC32): Likewise.
15875 (LINK_OS_LINUX_SPEC64): Likewise.
15876 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15877 (LINK_OS_LINUX_SPEC): Likewise.
15879 2017-09-13 Martin Liska <mliska@suse.cz>
15881 PR middle-end/82154
15882 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15883 CASE_HIGH is NULL_TREE.
15885 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15886 Alan Hayward <alan.hayward@arm.com>
15887 David Sherwood <david.sherwood@arm.com>
15889 * target.def (secondary_memory_needed): New hook.
15890 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15891 instead of SECONDARY_MEMORY_NEEDED.
15892 (secondary_memory_needed_mode): Likewise.
15893 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15894 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15895 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15896 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15897 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15898 * doc/tm.texi: Regenerate.
15899 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15900 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15901 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15902 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15903 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15904 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15905 mode argument first and change the reg_class arguments to reg_class_t.
15906 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15907 Make static. Update the call to inline_secondary_memory_needed.
15908 (ix86_register_move_cost): Update the call to
15909 inline_secondary_memory_needed.
15910 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15911 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15913 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15914 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15916 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15917 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15918 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15919 and match hook interface. Add comment from mips.h.
15920 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15921 * config/mmix/mmix.md (truncdfsf2): Refer to
15922 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15924 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15925 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15926 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15927 (pa_secondary_memory_needed): New function.
15928 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15929 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15930 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15931 (pdp11_secondary_memory_needed): Make static and match hook interface.
15932 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15933 * config/powerpcspe/powerpcspe-protos.h
15934 (rs6000_secondary_memory_needed_ptr): Delete.
15935 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15937 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15938 (rs6000_option_override_internal): Assign to
15939 targetm.secondary_memory_needed rather than
15940 rs6000_secondary_memory_needed_ptr.
15941 (rs6000_secondary_memory_needed): Match hook interface.
15942 (rs6000_debug_secondary_memory_needed): Likewise.
15943 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15944 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15945 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15946 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15947 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15948 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15950 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15951 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15952 (rs6000_option_override_internal): Assign to
15953 targetm.secondary_memory_needed rather than
15954 rs6000_secondary_memory_needed_ptr.
15955 (rs6000_secondary_memory_needed): Match hook interface.
15956 (rs6000_debug_secondary_memory_needed): Likewise.
15957 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15958 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15959 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15960 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15961 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15962 (sparc_secondary_memory_needed): New function.
15963 * lra-constraints.c (check_and_process_move): Refer to
15964 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15966 (curr_insn_transform): Likewise.
15967 (process_alt_operands): Use targetm.secondary_memory_needed
15968 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15969 (check_secondary_memory_needed_p): Likewise.
15970 (choose_split_class): Likewise.
15971 * reload.c: Unconditionally include code that was previously
15972 conditional on SECONDARY_MEMORY_NEEDED.
15973 (push_secondary_reload): Use targetm.secondary_memory_needed
15974 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15975 (push_reload): Likewise.
15976 * reload1.c: Unconditionally include code that was previously
15977 conditional on SECONDARY_MEMORY_NEEDED.
15978 (choose_reload_regs): Use targetm.secondary_memory_needed
15979 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15980 (gen_reload): Likewise.
15981 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15983 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15984 Alan Hayward <alan.hayward@arm.com>
15985 David Sherwood <david.sherwood@arm.com>
15987 * target.def (secondary_memory_needed_mode): New hook:
15988 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15989 * targhooks.h (default_secondary_memory_needed_mode): New function.
15990 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15991 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15992 * doc/tm.texi: Regenerate.
15993 * lra-constraints.c (check_and_process_move): Use
15994 targetm.secondary_memory_needed_mode instead of
15995 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15996 (curr_insn_transform): Likewise.
15997 * reload.c (get_secondary_mem): Likewise.
15998 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15999 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
16001 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16002 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16003 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
16004 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16005 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16007 * config/powerpcspe/powerpcspe-protos.h
16008 (rs6000_secondary_memory_needed_mode): Delete.
16009 * config/powerpcspe/powerpcspe.c
16010 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16011 (rs6000_secondary_memory_needed_mode): Make static.
16012 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16013 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16015 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16017 (rs6000_secondary_memory_needed_mode): Make static.
16018 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16019 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16020 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16021 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16022 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16024 (sparc_secondary_memory_needed_mode): New function.
16025 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16027 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16029 * config/aarch64/constraints.md (Umq): New constraint.
16030 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16032 (mov<mode>): Update condition.
16034 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16036 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16037 when bitposition is the same.
16039 2017-09-13 Richard Biener <rguenther@suse.de>
16041 * dwarf2out.c (output_die_symbol): Remove.
16042 (output_die): Do not output a DIEs symbol.
16044 2017-09-13 Richard Biener <rguenther@suse.de>
16046 PR middle-end/82128
16047 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16048 default-def to avoid breaking iterator update with the weird
16049 interaction with cgraph_update_edges_for_call_stmt_node.
16051 2017-09-13 Richard Biener <rguenther@suse.de>
16053 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16054 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16055 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16056 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16058 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16060 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16061 Disable pc relative literal load irrespective of
16062 TARGET_FIX_ERR_A53_84341 for default.
16064 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16066 * config/sparc/sparc.c (output_return): Output the source location of
16067 the insn in the delay slot, if any.
16068 (output_sibcall): Likewise.
16070 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16073 * common.opt (-static-pie): New alias.
16074 (shared): Negate static-pie.
16075 (-no-pie): Update help text.
16077 (static-pie): New option.
16078 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16079 -static-pie support.
16080 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16081 (LINK_EH_SPEC): Likewise.
16082 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16083 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16084 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16085 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16086 (init_gcc_specs): Likewise.
16087 (init_spec): Likewise.
16088 (display_help): Update help message for -pie.
16089 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16092 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16094 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16095 (movdi_aarch64): Likewise.
16096 (movti_aarch64): Likewise.
16098 2017-09-12 Simon Wright <simon@pushface.org>
16101 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16102 calculation of the minor version, always output as 0.
16104 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16107 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16108 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16109 on it early, rather than manual conversion late. For
16110 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16111 instead of performing manual conversion.
16113 2017-09-12 Carl Love <cel@us.ibm.com>
16115 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16116 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16118 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16119 VMULOSW): Add definitions.
16120 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16121 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16122 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16123 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16124 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16126 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16128 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16130 (movti_aarch64): Likewise.
16131 (movdf_aarch64): Likewise.
16132 (movtf_aarch64): Likewise.
16133 (load_pairdi): Likewise.
16134 (store_pairdi): Likewise.
16135 (load_pairdf): Likewise.
16136 (store_pairdf): Likewise.
16137 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16138 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16139 (ldr_got_small_<mode>): Likewise.
16140 (ldr_got_small_28k_<mode>): Likewise.
16141 (ldr_got_tiny): Likewise.
16142 * config/aarch64/iterators.md (ldst_sz): New.
16143 (ldpstp_sz): Likewise.
16144 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16146 (thunderx_load): Split load_8 to load_16.
16147 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16149 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16150 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16151 (xgene1_store_pair): Split store_8 to store_16.
16152 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16153 (falkor_st_0_st_sd): Split store_8 to store_16.
16155 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16157 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16158 and store1/2/3/4 to store_4/8/12/16.
16159 * config/aarch64/aarch64.md: Update for rename.
16160 * config/arm/arm.md: Likewise.: Likewise.
16161 * config/arm/arm.c: Likewise.
16162 * config/arm/thumb1.md: Likewise.
16163 * config/arm/thumb2.md: Likewise.
16164 * config/arm/vfp.md: Likewise.
16165 * config/arm/arm-generic.md: Likewise.
16166 * config/arm/arm1020e.md: Likewise.
16167 * config/arm/arm1026ejs.md: Likewise.
16168 * config/arm/arm1136jfs.md: Likewise.
16169 * config/arm/arm926ejs.md: Likewise.
16170 * config/arm/cortex-a15.md: Likewise.
16171 * config/arm/cortex-a17.md: Likewise.
16172 * config/arm/cortex-a5.md: Likewise.
16173 * config/arm/cortex-a53.md: Likewise.
16174 * config/arm/cortex-a57.md: Likewise.
16175 * config/arm/cortex-a7.md: Likewise.
16176 * config/arm/cortex-a8.md: Likewise.
16177 * config/arm/cortex-a9.md: Likewise.
16178 * config/arm/cortex-m4.md: Likewise.
16179 * config/arm/cortex-m7.md: Likewise.
16180 * config/arm/cortex-r4.md: Likewise.
16181 * config/arm/exynos-m1.md: Likewise.
16182 * config/arm/fa526.md: Likewise.
16183 * config/arm/fa606te.md: Likewise.
16184 * config/arm/fa626te.md: Likewise.
16185 * config/arm/fa726te.md: Likewise.
16186 * config/arm/fmp626.md: Likewise.
16187 * config/arm/iwmmxt.md: Likewise.
16188 * config/arm/ldmstm.md: Likewise.
16189 * config/arm/marvell-pj4.md: Likewise.
16190 * config/arm/xgene1.md: Likewise.
16191 * config/aarch64/thunderx.md: Likewise.
16192 * config/aarch64/thunderx2t99.md: Likewise.
16193 * config/aarch64/falkor.md: Likewise.
16195 2017-09-12 Martin Liska <mliska@suse.cz>
16197 * attribs.c (private_lookup_attribute): New function.
16198 * attribs.h (private_lookup_attribute): Declared here.
16199 (lookup_attribute): Called from this place.
16201 2017-09-12 Richard Biener <rguenther@suse.de>
16203 PR tree-optimization/82157
16204 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16205 stmts with side-effects.
16207 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16208 Alan Hayward <alan.hayward@arm.com>
16209 David Sherwood <david.sherwood@arm.com>
16211 * target.def (hard_regno_nregs): New hook.
16212 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16213 * targhooks.h (default_hard_regno_nregs): Declare.
16214 * targhooks.c (default_hard_regno_nregs): New function.
16215 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16216 (TARGET_HARD_REGNO_NREGS): ...this hook.
16217 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16218 (CLASS_MAX_NREGS): Likewise.
16219 * doc/tm.texi: Regenerate.
16220 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16221 instead of HARD_REGNO_NREGS.
16222 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16223 HARD_REGNO_NREGS in the comment.
16224 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16225 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16226 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16227 Return an unsigned int.
16228 (TARGET_HARD_REGNO_NREGS): Redefine.
16229 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16230 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16231 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16232 (arc_hard_regno_nregs): New function.
16233 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16234 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16235 (arm_hard_regno_nregs): New function.
16236 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16237 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16238 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16239 (TARGET_HARD_REGNO_NREGS): Redefine.
16240 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16241 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16242 (HARD_REGNO_NREGS): Delete.
16243 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16244 (cr16_hard_regno_nregs): New function.
16245 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16246 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16247 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16248 (cris_hard_regno_nregs): New function.
16249 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16250 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16251 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16252 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16253 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16254 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16255 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16256 (frv_hard_regno_nregs): Make static. Take and return an
16258 (frv_class_max_nregs): Remove outdated copy of documentation.
16259 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16260 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16261 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16262 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16263 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16264 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16265 (TARGET_HARD_REGNO_NREGS): Redefine.
16266 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16267 (CLASS_MAX_NREGS): Update comment.
16268 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16269 (ia64_hard_regno_nregs): New function.
16270 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16271 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16272 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16273 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16274 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16276 (m32c_hard_regno_nregs): Likewise. Make static.
16277 (TARGET_HARD_REGNO_NREGS): Redefine.
16278 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16279 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16280 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16281 (m68k_hard_regno_nregs): New function.
16282 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16283 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16284 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16285 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16286 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16287 Take and return an unsigned int.
16288 (TARGET_HARD_REGNO_NREGS): Redefine.
16289 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16290 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16291 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16292 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16293 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16294 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16295 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16296 (msp430_hard_regno_nregs): Make static. Take and return an
16298 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16299 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16300 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16301 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16302 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16303 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16304 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16305 (TARGET_HARD_REGNO_NREGS): Redefine.
16306 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16307 (PA_HARD_REGNO_NREGS): ...this.
16308 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16309 (PA_HARD_REGNO_NREGS): ...this.
16310 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16311 (pa_hard_regno_nregs): New function.
16312 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16313 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16314 (pdp11_hard_regno_nregs): New function.
16315 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16316 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16317 (rs6000_hard_regno_nregs_hook): New function.
16318 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16319 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16320 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16321 Take and return an unsigned int. Move earlier in file.
16322 (TARGET_HARD_REGNO_NREGS): Redefine.
16323 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16324 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16325 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16326 (rl78_hard_regno_nregs): Make static. Take and return an
16328 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16329 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16330 (rs6000_hard_regno_nregs_hook): New function.
16331 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16332 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16333 (TARGET_HARD_REGNO_NREGS): Redefine.
16334 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16335 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16336 instead of HARD_REGNO_NREGS.
16337 (s390_hard_regno_nregs): New function.
16338 (s390_hard_regno_mode_ok): Add comment from s390.h.
16339 (TARGET_HARD_REGNO_NREGS): Redefine.
16340 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16341 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16342 (sh_hard_regno_nregs): New function.
16343 (sh_pass_in_reg_p): Use it.
16344 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16345 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16346 (sparc_hard_regno_nregs): New function.
16347 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16348 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16349 (spu_function_arg_advance): Use it, supplying a valid register number.
16350 (TARGET_HARD_REGNO_NREGS): Redefine.
16351 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16352 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16353 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16354 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16355 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16356 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16357 (CLASS_MAX_NREGS): Remove copy of old documentation.
16358 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16359 (visium_hard_regno_nregs): New function.
16360 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16361 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16362 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16363 xtensa_hard_regno_nregs): New function.
16364 * system.h (HARD_REGNO_NREGS): Poison.
16366 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16368 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16369 hard_regno_nregs instead of HARD_REGNO_NREGS.
16370 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16371 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16372 (c6x_expand_epilogue): Likewise.
16373 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16374 (frv_read_iacc_argument): Likewise.
16375 * config/sh/sh.c: Include regs.h.
16376 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16377 (regs_used): Likewise.
16378 (output_stack_adjust): Likewise.
16379 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16380 * expmed.c: Include regs.h.
16381 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16382 * ree.c: Include regs.h.
16383 (combine_reaching_defs): Use hard_regno_nregs instead of
16385 (add_removable_extension): Likewise.
16387 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16389 * regs.h (hard_regno_nregs): Turn into a function.
16390 (end_hard_regno): Update accordingly.
16391 * caller-save.c (setup_save_areas): Likewise.
16392 (save_call_clobbered_regs): Likewise.
16393 (replace_reg_with_saved_mem): Likewise.
16394 (insert_restore): Likewise.
16395 (insert_save): Likewise.
16396 * combine.c (can_change_dest_mode): Likewise.
16397 (move_deaths): Likewise.
16398 (distribute_notes): Likewise.
16399 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16400 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16401 (rs6000_split_multireg_move): Likewise.
16402 (rs6000_register_move_cost): Likewise.
16403 (rs6000_memory_move_cost): Likewise.
16404 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16405 (rs6000_split_multireg_move): Likewise.
16406 (rs6000_register_move_cost): Likewise.
16407 (rs6000_memory_move_cost): Likewise.
16408 * cselib.c (cselib_reset_table): Likewise.
16409 (cselib_lookup_1): Likewise.
16410 * emit-rtl.c (set_mode_and_regno): Likewise.
16411 * function.c (aggregate_value_p): Likewise.
16412 * ira-color.c (setup_profitable_hard_regs): Likewise.
16413 (check_hard_reg_p): Likewise.
16414 (calculate_saved_nregs): Likewise.
16415 (assign_hard_reg): Likewise.
16416 (improve_allocation): Likewise.
16417 (calculate_spill_cost): Likewise.
16418 * ira-emit.c (modify_move_list): Likewise.
16419 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16420 (ira_hard_reg_in_set_p): Likewise.
16421 * ira.c (setup_reg_mode_hard_regset): Likewise.
16422 (clarify_prohibited_class_mode_regs): Likewise.
16423 (check_allocation): Likewise.
16424 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16425 (lra_setup_reg_renumber): Likewise.
16426 (setup_try_hard_regno_pseudos): Likewise.
16427 (spill_for): Likewise.
16428 (assign_hard_regno): Likewise.
16429 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16430 * lra-constraints.c (in_class_p): Likewise.
16431 (lra_constraint_offset): Likewise.
16432 (simplify_operand_subreg): Likewise.
16433 (lra_constraints): Likewise.
16434 (split_reg): Likewise.
16435 (split_if_necessary): Likewise.
16436 (invariant_p): Likewise.
16437 (inherit_in_ebb): Likewise.
16438 * lra-lives.c (process_bb_lives): Likewise.
16439 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16440 (get_hard_regs): Likewise.
16441 (do_remat): Likewise.
16442 * lra-spills.c (assign_spill_hard_regs): Likewise.
16443 * mode-switching.c (create_pre_exit): Likewise.
16444 * postreload.c (reload_combine_recognize_pattern): Likewise.
16445 * recog.c (peep2_find_free_register): Likewise.
16446 * regcprop.c (kill_value_regno): Likewise.
16447 (set_value_regno): Likewise.
16448 (copy_value): Likewise.
16449 (maybe_mode_change): Likewise.
16450 (find_oldest_value_reg): Likewise.
16451 (copyprop_hardreg_forward_1): Likewise.
16452 * regrename.c (check_new_reg_p): Likewise.
16453 (regrename_do_replace): Likewise.
16454 * reload.c (push_reload): Likewise.
16455 (combine_reloads): Likewise.
16456 (find_dummy_reload): Likewise.
16457 (operands_match_p): Likewise.
16458 (find_reloads): Likewise.
16459 (find_equiv_reg): Likewise.
16460 (reload_adjust_reg_for_mode): Likewise.
16461 * reload1.c (count_pseudo): Likewise.
16462 (count_spilled_pseudo): Likewise.
16463 (find_reg): Likewise.
16464 (clear_reload_reg_in_use): Likewise.
16465 (free_for_value_p): Likewise.
16466 (allocate_reload_reg): Likewise.
16467 (choose_reload_regs): Likewise.
16468 (reload_adjust_reg_for_temp): Likewise.
16469 (emit_reload_insns): Likewise.
16470 (delete_output_reload): Likewise.
16471 * rtlanal.c (subreg_get_info): Likewise.
16472 * sched-deps.c (sched_analyze_reg): Likewise.
16473 * sel-sched.c (init_regs_for_mode): Likewise.
16474 (mark_unavailable_hard_regs): Likewise.
16475 (choose_best_reg_1): Likewise.
16476 (verify_target_availability): Likewise.
16477 * valtrack.c (dead_debug_insert_temp): Likewise.
16478 * var-tracking.c (track_loc_p): Likewise.
16479 (emit_note_insn_var_location): Likewise.
16480 * varasm.c (make_decl_rtl): Likewise.
16481 * reginfo.c (choose_hard_reg_mode): Likewise.
16482 (init_reg_modes_target): Refer directly to
16483 this_target_regs->x_hard_regno_nregs.
16485 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16487 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16488 instead of hard_regno_nregs.
16490 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16492 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16493 end_hard_regno instead of hard_regno_nregs.
16494 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16495 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16496 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16497 * ira-color.c (improve_allocation): Likewise.
16498 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16499 * lra-lives.c (mark_regno_live): Likewise.
16500 (mark_regno_dead): Likewise.
16501 * lra-remat.c (operand_to_remat): Likewise.
16502 * lra.c (collect_non_operand_hard_regs): Likewise.
16503 * postreload.c (reload_combine_note_store): Likewise.
16504 (move2add_valid_value_p): Likewise.
16505 * reload.c (regno_clobbered_p): Likewise.
16507 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16509 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16511 * config/v850/v850.c (v850_reorg): Likewise.
16512 * reload.c (refers_to_regno_for_reload_p): Likewise.
16513 (find_equiv_reg): Likewise.
16514 * reload1.c (reload_reg_reaches_end_p): Likewise.
16516 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16518 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16520 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16521 * config/arm/arm.c (output_move_neon): Likewise.
16522 (arm_attr_length_move_neon): Likewise.
16523 (neon_split_vcombine): Likewise.
16524 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16525 (c6x_mark_reg_written): Likewise.
16526 (c6x_dwarf_register_span): Likewise.
16527 * config/i386/i386.c (ix86_save_reg): Likewise.
16528 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16529 (rws_access_reg): Likewise.
16530 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16531 * mode-switching.c (create_pre_exit): Likewise.
16532 * ree.c (combine_reaching_defs): Likewise.
16533 (add_removable_extension): Likewise.
16534 * regcprop.c (find_oldest_value_reg): Likewise.
16535 (copyprop_hardreg_forward_1): Likewise.
16536 * reload.c (reload_inner_reg_of_subreg): Likewise.
16537 (push_reload): Likewise.
16538 (combine_reloads): Likewise.
16539 (find_dummy_reload): Likewise.
16540 (reload_adjust_reg_for_mode): Likewise.
16541 * reload1.c (find_reload_regs): Likewise.
16542 (forget_old_reloads_1): Likewise.
16543 (reload_reg_free_for_value_p): Likewise.
16544 (reload_adjust_reg_for_temp): Likewise.
16545 (emit_reload_insns): Likewise.
16546 (delete_output_reload): Likewise.
16547 * sel-sched.c (choose_best_reg_1): Likewise.
16548 (choose_best_pseudo_reg): Likewise.
16550 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16551 Alan Hayward <alan.hayward@arm.com>
16552 David Sherwood <david.sherwood@arm.com>
16554 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16555 * target.def (slow_unaligned_access): New hook.
16556 * targhooks.h (default_slow_unaligned_access): Declare.
16557 * targhooks.c (default_slow_unaligned_access): New function.
16558 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16559 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16560 * doc/tm.texi: Regenerate.
16561 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16562 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16563 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16565 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16566 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16568 (rs6000_slow_unaligned_access): New function.
16569 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16570 (expand_block_compare): Likewise.
16571 (expand_strn_compare): Likewise.
16572 (rs6000_rtx_costs): Likewise.
16573 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16574 (riscv_slow_unaligned_access): Likewise.
16575 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16576 (riscv_slow_unaligned_access_p): ...this and make static.
16577 (riscv_option_override): Update accordingly.
16578 (riscv_slow_unaligned_access): New function.
16579 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16580 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16581 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16582 (rs6000_slow_unaligned_access): New function.
16583 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16584 (rs6000_rtx_costs): Likewise.
16585 * config/rs6000/rs6000-string.c (expand_block_compare)
16586 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16587 of SLOW_UNALIGNED_ACCESS.
16588 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16589 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16590 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16591 of SLOW_UNALIGNED_ACCESS.
16592 * expmed.c (simple_mem_bitfield_p): Likewise.
16593 * expr.c (alignment_for_piecewise_move): Likewise.
16594 (emit_group_load_1): Likewise.
16595 (emit_group_store): Likewise.
16596 (copy_blkmode_from_reg): Likewise.
16597 (emit_push_insn): Likewise.
16598 (expand_assignment): Likewise.
16599 (store_field): Likewise.
16600 (expand_expr_real_1): Likewise.
16601 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16602 * lra-constraints.c (simplify_operand_subreg): Likewise.
16603 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16604 * gimple-ssa-store-merging.c: Likewise in block comment at start
16606 * tree-ssa-strlen.c: Include target.h.
16607 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16608 of SLOW_UNALIGNED_ACCESS.
16609 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16611 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16613 PR rtl-optimization/82185
16614 * expmed.c (emit_store_flag_int): Only test tem if it has been
16617 2017-09-12 Richard Biener <rguenther@suse.de>
16619 PR middle-end/82149
16620 * match.pd ((FTYPE) N CMP CST): Fix typo.
16622 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16624 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16626 (mips_near_type_p): Add 'short_call' attribute as a synonym
16628 * doc/extend.texi (short_call): Document new function attribute.
16630 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16633 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16634 assertion check that in the condition.
16635 (get_atomic_generic_size): Likewise. Before testing if parameter
16636 has pointer type, if it has array type, call for C++
16637 default_conversion to perform array-to-pointer conversion.
16639 2017-09-12 Richard Biener <rguenther@suse.de>
16641 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16642 for operations we cannot scalarize.
16644 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16646 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16647 vectors heap vectors. Clean up comments.
16648 Make visited_bbs a reference.
16649 (profitable_jump_thread_path): Make GC
16650 vectors heap vectors. Clean up comments.
16652 (convert_and_register_jump_thread_path): Make GC vectors heap
16654 (check_subpath_and_update_thread_path): Same. Clean up comments.
16655 Make visited_bbs a reference.
16656 (handle_phi): Abstract common code to to
16657 register_jump_thread_path_if_profitable.
16658 Rename VAR_BB to DEF_BB.
16660 Make GC vectors heap vectors.
16661 Make visited_bbs a reference.
16662 (handle_assignment): Same.
16663 (register_jump_thread_path_if_profitable): New.
16664 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16666 Make GC vectors heap vectors. Clean up comments.
16667 Make visited_bbs a reference.
16668 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16669 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16672 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16675 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16676 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16678 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16680 Revert r251800 and r251799.
16682 2017-09-11 Martin Jambor <mjambor@suse.cz>
16685 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16686 arguments in advance.
16687 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16688 use it to find predecessor edges.
16689 (naive_outof_ssa): Collect vector of predecessors.
16691 2017-09-08 Jason Merrill <jason@redhat.com>
16693 PR c++/70029 - ICE with ref-qualifier and -flto
16694 * langhooks.h (struct lang_hooks_for_types): Add
16695 copy_lang_qualifiers.
16696 * attribs.c (build_type_attribute_qual_variant): Use it.
16697 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16699 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16700 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16702 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16705 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16706 (*mulsi3_sp64): New instruction.
16707 (mulsi3): New expander.
16709 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16711 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16713 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16715 * sancov.c: Include memmodel.h.
16717 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16720 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16723 2017-09-07 Carl Love <cel@us.ibm.com>
16725 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16726 the sldi instruction.
16728 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16730 * sancov.c: Include tm_p.h.
16732 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16735 * output.h (switch_to_other_text_partition): New declaration.
16736 * varasm.c (switch_to_other_text_partition): New function.
16737 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16738 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16739 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16740 to the other text partition before emitting LCL label and switch back
16741 after emitting the word after it.
16743 2017-09-07 Richard Biener <rguenther@suse.de>
16745 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16746 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16747 critical edges here, after loop init.
16748 (pass_data_pre): Remove PROP_no_crit_edges flags.
16749 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16750 for valueization of call args to avoid leaking VN_TOP.
16751 (visit_use): Assert we do not visit default defs.
16752 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16753 Use error_mark_node to more easily detect leaking VN_TOP.
16754 All default-defs are varying, not VN_TOP. Mark them visited.
16755 (run_scc_vn): Make code match comment.
16757 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16759 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16760 OPTION_MASK_FLOAT128_KEYWORD.
16761 (POWERPC_MASKS): Likewise.
16762 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16763 support for the -mfloat128-type option, and make -mfloat128
16764 default on PowerPC Linux systems. Define or undefine
16765 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16766 Define __float128 to be __ieee128 if IEEE 128-bit support is
16767 enabled, or undefine it.
16768 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16769 Delete defining __FLOAT128_TYPE__.
16770 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16771 -mfloat128-type option and make -mfloat128 default on PowerPC
16773 (TARGET_FLOAT128_TYPE): Likewise.
16774 (-mfloat128-type): Likewise.
16775 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16776 Delete the -mfloat128-type option and make -mfloat128 default on
16777 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16778 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16779 128-bit floating point is enabled. Change tests from using
16780 -mfloat128-type to -mfloat128.
16781 (rs6000_mangle_type): Use the correct mangling for the __float128
16782 type even if normal long double is restricted to 64-bits.
16783 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16785 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16786 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16787 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16789 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16790 documentation for -mfloat128.
16792 2017-09-06 Olivier Hainque <hainque@adacore.com>
16794 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16796 2017-09-06 Wish Wu <wishwu007@gmail.com>
16797 Jakub Jelinek <jakub@redhat.com>
16799 * asan.c (initialize_sanitizer_builtins): Add
16800 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16801 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16802 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16803 BT_FN_VOID_UINT64_PTR variables.
16804 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16805 (BT_FN_VOID_UINT16_UINT16): Likewise.
16806 (BT_FN_VOID_UINT32_UINT32): Likewise.
16807 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16808 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16809 (BT_FN_VOID_UINT64_PTR): Likewise.
16810 * common.opt (flag_sanitize_coverage): New variable.
16811 (fsanitize-coverage=trace-pc): Remove.
16812 (fsanitize-coverage=): Add.
16813 * flag-types.h (enum sanitize_coverage_code): New enum.
16814 * fold-const.c (fold_range_test): Disable non-short-circuit
16815 optimization if flag_sanitize_coverage.
16816 (fold_truth_andor): Likewise.
16817 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16818 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16819 (coverage_sanitizer_opts): New array.
16820 (get_closest_sanitizer_option): Add OPTS argument, handle also
16821 OPT_fsanitize_coverage_.
16822 (parse_sanitizer_options): Adjusted to also handle
16823 OPT_fsanitize_coverage_.
16824 (common_handle_option): Add OPT_fsanitize_coverage_.
16825 * sancov.c (instrument_comparison, instrument_switch): New function.
16826 (sancov_pass): Add trace-cmp support.
16827 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16828 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16829 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16830 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16831 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16832 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16833 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16834 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16835 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16836 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16838 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16840 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16841 error. Only quit immediately if parsing is complete.
16842 (BEGIN): Initialize fatal_err and parse_done.
16843 (begin fpu, end fpu): Check number of arguments.
16844 (begin arch, end arch): Likewise.
16845 (begin cpu, end cpu): Likewise.
16846 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16847 (optalias): Likewise.
16849 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16851 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16852 * config/arm/arm-isa.h: Delete. Move definitions to ...
16853 * arm-cpus.in: ... here. Use new feature and fgroup values.
16854 * config/arm/arm.c (arm_option_override): Use lower case for feature
16856 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16857 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16858 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16860 (print_isa_bits_for): New function.
16861 (gen_isa): New function.
16862 (gen_comm_data): Use print_isa_bits_for.
16863 (define feature): New keyword.
16864 (define fgroup): New keyword.
16865 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16866 (arm-isa.h): Add rule to generate file.
16867 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16868 case for feature bit names.
16870 2017-09-06 Richard Biener <rguenther@suse.de>
16872 * tree-ssa-pre.c (NECESSARY): Remove.
16873 (create_expression_by_pieces): Do not touch pass-local flags.
16874 (insert_into_preds_of_block): Likewise.
16875 (do_pre_regular_insertion): Likewise.
16876 (eliminate_insert): Likewise.
16877 (eliminate_dom_walker::before_dom_children): Likewise.
16878 (fini_eliminate): Do not look at inserted_exprs.
16879 (mark_operand_necessary): Remove.
16880 (remove_dead_inserted_code): Replace with simple work-list
16881 algorithm based on inserted_exprs and SSA uses.
16882 (pass_pre::execute): Re-order fini_eliminate and
16883 remove_dead_inserted_code.
16885 2017-09-06 Olivier Hainque <hainque@adacore.com>
16887 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16888 for VxWorks 7. Adjust surrounding comments.
16890 2017-09-06 Richard Biener <rguenther@suse.de>
16892 * gimple-ssa-strength-reduction.c
16893 (find_candidates_dom_walker::before_dom_children): Also allow
16896 2017-09-06 Richard Biener <rguenther@suse.de>
16898 PR tree-optimization/82108
16899 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16900 for gap in the non-permutation SLP case.
16902 2017-09-06 Martin Jambor <mjambor@suse.cz>
16904 PR tree-optimization/82078
16905 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16906 add_access_to_work_queue...
16907 (build_accesses_from_assign): ...here.
16908 (propagate_all_subaccesses): Make sure racc is the group
16909 representative, if there is one.
16911 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16913 PR middle-end/82095
16914 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16917 2017-09-06 Richard Biener <rguenther@suse.de>
16919 * gimple-ssa-strength-reduction.c
16920 (find_candidates_dom_walker::before_doom_children): Use a
16921 type and not a mode check.
16923 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16926 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16928 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16930 2017-09-05 Jeff Law <law@redhat.com>
16932 PR tree-optimization/64910
16933 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16934 cases where we have 3 or more operands.
16936 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16938 PR middle-end/81768
16939 * omp-low.c (lower_omp_for): Recompute tree invariant if
16940 gimple_omp_for_initial/final is ADDR_EXPR.
16942 PR middle-end/81768
16943 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16944 into gimple val before gimplification fo the COND_EXPR.
16946 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16948 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16949 REGION_COPY argument.
16950 (thread_through_all_blocks): Remove unused argument to
16951 duplicate_thread_path.
16953 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16954 Alan Hayward <alan.hayward@arm.com>
16955 David Sherwood <david.sherwood@arm.com>
16957 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16958 Take a scalar_mode rather than a machine_mode.
16959 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16960 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16961 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16962 (aarch64_gen_adjusted_ldpstp): Likewise.
16963 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16965 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16966 Alan Hayward <alan.hayward@arm.com>
16967 David Sherwood <david.sherwood@arm.com>
16969 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16970 Take a scalar_int_mode instead of a machine_mode.
16971 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16972 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16973 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16974 (aarch64_simd_attr_length_rglist): Delete.
16975 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16976 a scalar_int_mode instead of a machine_mode.
16977 (aarch64_add_offset): Likewise.
16978 (aarch64_internal_mov_immediate): Likewise
16979 (aarch64_add_constant_internal): Likewise.
16980 (aarch64_add_constant): Likewise.
16981 (aarch64_movw_imm): Likewise.
16982 (aarch64_rtx_arith_op_extract_p): Likewise.
16983 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16984 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16985 Remove assert that the mode isn't a vector.
16986 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16987 (aarch64_expand_mov_immediate): Update calls after above changes.
16988 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16989 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16990 (aarch64_move_imm): Likewise.
16991 (aarch64_can_const_movi_rtx_p): Likewise.
16992 (aarch64_strip_extend): Likewise.
16993 (aarch64_extr_rtx_p): Likewise.
16994 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16995 a CONST_INT when the mode parameter is VOIDmode.
16996 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16998 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17000 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
17001 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17002 (bitwise_type_for_mode): Update accordingly.
17004 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17006 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17007 * stor-layout.c (mode_for_size_tree): Likewise.
17008 (mode_for_array): Update accordingly.
17009 (layout_decl): Likewise.
17010 (compute_record_mode): Likewise. Only set the mode once.
17012 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17014 * target.def (get_mask_mode): Change return type to opt_mode.
17016 * doc/tm.texi: Regenerate.
17017 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17018 * targhooks.c (default_get_mask_mode): Likewise.
17019 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17020 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17021 targetm.get_mask_mode.
17022 * tree.c (build_truth_vector_type): Likewise.
17024 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17026 * machmode.h (mode_for_vector): Return an opt_mode.
17027 * stor-layout.c (mode_for_vector): Likewise.
17028 (mode_for_int_vector): Update accordingly.
17029 (layout_type): Likewise.
17030 * config/i386/i386.c (emit_memmov): Likewise.
17031 (ix86_expand_set_or_movmem): Likewise.
17032 (ix86_expand_vector_init): Likewise.
17033 (ix86_get_mask_mode): Likewise.
17034 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17036 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17037 * expmed.c (extract_bit_field_1): Likewise.
17038 * expr.c (expand_expr_real_2): Likewise.
17039 * optabs-query.c (can_vec_perm_p): Likewise.
17040 (can_vec_mask_load_store_p): Likewise.
17041 * optabs.c (expand_vec_perm): Likewise.
17042 * targhooks.c (default_get_mask_mode): Likewise.
17043 * tree-vect-stmts.c (vectorizable_store): Likewise.
17044 (vectorizable_load): Likewise.
17045 (get_vectype_for_scalar_type_and_size): Likewise.
17047 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17049 * machmode.h (mode_for_int_vector): New function.
17050 * stor-layout.c (mode_for_int_vector): Likewise.
17051 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17052 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17053 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17054 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17055 (s390_expand_vcond): Likewise.
17057 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17059 * machmode.h (opt_machine_mode): New type.
17060 (opt_mode<T>): Allow construction from anything that can be
17062 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17063 (mode_for_size): Return an opt_machine_mode.
17064 * stor-layout.c (mode_for_size): Likewise.
17065 (mode_for_size_tree): Update call accordingly.
17066 (bitwise_mode_for_mode): Likewise.
17067 (make_fract_type): Likewise.
17068 (make_accum_type): Likewise.
17069 * caller-save.c (replace_reg_with_saved_mem): Update call
17071 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17072 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17073 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17074 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17075 * expmed.c (extract_bit_field_1): Likewise.
17076 * reload.c (get_secondary_mem): Likewise.
17077 * varasm.c (assemble_integer): Likewise.
17078 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17081 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17083 * machmode.h (decimal_float_mode_for_size): New function.
17084 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17085 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17086 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17087 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17088 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17089 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17091 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17093 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17094 (get_builtin_sync_mode): Likewise.
17095 (expand_ifn_atomic_compare_exchange): Likewise.
17096 (expand_builtin_atomic_clear): Likewise.
17097 (expand_builtin_atomic_test_and_set): Likewise.
17098 (fold_builtin_atomic_always_lock_free): Likewise.
17099 * calls.c (compute_argument_addresses): Likewise.
17100 (emit_library_call_value_1): Likewise.
17101 (store_one_arg): Likewise.
17102 * combine.c (combine_instructions): Likewise.
17103 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17104 * config/arm/arm.c (arm_function_value): Likewise.
17105 (aapcs_allocate_return_reg): Likewise.
17106 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17107 * config/i386/i386.c (construct_container): Likewise.
17108 (ix86_gimplify_va_arg): Likewise.
17109 (ix86_expand_sse_cmp): Likewise.
17110 (emit_memmov): Likewise.
17111 (emit_memset): Likewise.
17112 (expand_small_movmem_or_setmem): Likewise.
17113 (ix86_expand_pextr): Likewise.
17114 (ix86_expand_pinsr): Likewise.
17115 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17116 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17118 * config/mips/mips.c (mips_function_value_1) Likewise.
17119 (mips_block_move_straight): Likewise.
17120 (mips_expand_ins_as_unaligned_store): Likewise.
17121 * config/powerpcspe/powerpcspe.c
17122 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17123 (rs6000_darwin64_record_arg_flush): Likewise.
17124 * config/rs6000/rs6000.c
17125 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17126 (rs6000_darwin64_record_arg_flush): Likewise.
17127 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17128 (sparc_function_value_1): Likewise.
17129 * config/spu/spu.c (adjust_operand): Likewise.
17130 (spu_emit_branch_or_set): Likewise.
17131 (arith_immediate_p): Likewise.
17132 * emit-rtl.c (gen_lowpart_common): Likewise.
17133 * expr.c (expand_expr_real_1): Likewise.
17134 * function.c (assign_parm_setup_block): Likewise.
17135 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17136 * reload1.c (alter_reg): Likewise.
17137 * stor-layout.c (mode_for_vector): Likewise.
17138 (layout_type): Likewise.
17140 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17142 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17143 (spu_convert_move): Likewise.
17144 * lower-subreg.c (resolve_simple_move): Likewise.
17146 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17149 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17150 define_insn to a define_expand.
17151 (altivec_vsum2sws_direct): New define_insn.
17152 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17154 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17156 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17159 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17164 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17165 realign stack if stack alignment needed is less than incoming
17168 2017-09-05 Marek Polacek <polacek@redhat.com>
17171 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17174 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17176 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17178 2017-09-05 Richard Biener <rguenther@suse.de>
17180 PR tree-optimization/82084
17181 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17183 2017-09-05 Richard Biener <rguenther@suse.de>
17185 PR tree-optimization/82102
17186 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17188 2017-09-05 Martin Liska <mliska@suse.cz>
17190 PR tree-optimization/82032
17191 * tree-cfg.c (generate_range_test): New function.
17192 * tree-cfg.h (generate_range_test): Declared here.
17193 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17194 (cleanup_control_expr_graph): Use it.
17195 * tree-switch-conversion.c (try_switch_expansion): Remove
17197 (emit_case_nodes): Use generate_range_test.
17199 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17202 * config/i386/i386.md (*<btsc><mode>_mask): Add
17203 TARGET_USE_BT to insn constraint.
17204 (*btr<mode>_mask): Ditto.
17206 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17208 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17209 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17211 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17214 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17215 TARGET_NEON and TARGET_IWMMXT.
17216 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17217 TARGET_NEON and TARGET_IWMMXT.
17218 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17220 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17222 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17223 (ix86_rewrite_tls_address): Ditto.
17224 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17225 (ix86_rewrite_tls_address_1): Ditto.
17226 (ix86_rewrite_tls_address): Ditto.
17227 * config/i386/predicates.md (tls_address_pattern): New predicate.
17228 * config/i386/i386.md (TLS address splitter): New splitter.
17230 2017-09-04 Richard Biener <rguenther@suse.de>
17232 PR tree-optimization/82084
17233 * fold-const.h (can_native_encode_string_p): Declare.
17234 * fold-const.c (can_native_encode_string_p): Factor out from ...
17235 (native_encode_string): ... here.
17236 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17237 vectorizing stores from constants we later cannot handle.
17239 2017-09-04 Marek Polacek <polacek@redhat.com>
17242 * doc/invoke.texi: Update -Wtautological-compare documentation.
17244 2017-09-04 Jeff Law <law@redhat.com>
17246 PR tree-optimization/64910
17247 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17248 swap the first and last operand if the last is a constant.
17250 2017-09-04 Marek Polacek <polacek@redhat.com>
17253 * convert.c (do_narrow): When sanitizing signed integer overflows,
17254 bail out for signed types.
17255 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17257 2017-09-04 Richard Biener <rguenther@suse.de>
17259 PR tree-optimization/82060
17260 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17261 Move devirtualization after stmt folding and before EH/AB/noreturn
17262 cleanup to get the stmt refs canonicalized. Use a bool instead
17263 of gimple_modified_p since that doesn't work for NOPs. Schedule
17264 NOPs generated by folding for removal.
17266 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17267 Alan Hayward <alan.hayward@arm.com>
17268 David Sherwood <david.sherwood@arm.com>
17270 * coretypes.h (pad_direction): New enum.
17271 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17272 (FUNCTION_ARG_PADDING): Likewise.
17273 * target.def (function_arg_padding): New hook.
17274 * targhooks.h (default_function_arg_padding): Declare.
17275 * targhooks.c (default_function_arg_padding): New function.
17276 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17277 (TARGET_FUNCTION_ARG_PADDING): ...this.
17278 * doc/tm.texi: Regenerate.
17279 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17280 instead of direction.
17281 (compute_argument_addresses): Likewise.
17282 (load_register_parameters): Likewise.
17283 (emit_library_call_value_1): Likewise.
17284 (store_one_arg): Use targetm.calls.function_arg_padding instead
17285 of FUNCTION_ARG_PADDING.
17286 (must_pass_in_stack_var_size_or_pad): Likewise.
17287 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17288 (emit_group_store): Likewise.
17289 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17290 instead of FUNCTION_ARG_PADDING.
17291 (emit_push_insn): Likewise, and propagate enum change throughout
17293 * function.h (direction): Delete.
17294 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17296 * function.c (assign_parm_find_stack_rtl): Likewise.
17297 (assign_parm_setup_block_p): Likewise.
17298 (assign_parm_setup_block): Likewise.
17299 (gimplify_parameters): Likewise.
17300 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17301 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17303 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17304 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17305 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17306 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17307 (aarch64_function_arg_padding): ...this new function.
17308 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17309 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17310 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17311 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17312 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17313 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17314 (arm_pad_arg_upward): Replace with...
17315 (arm_function_arg_padding): ...this new function.
17316 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17318 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17319 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17320 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17321 (ia64_hpux_function_arg_padding): Replace with...
17322 (ia64_function_arg_padding): ...this new function. Use pad_direction
17323 instead of direction. Check for TARGET_HPUX.
17324 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17325 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17326 (iq2000_function_arg_padding): New function.
17327 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17328 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17329 (mips_function_arg_padding): ...this new function.
17330 (mips_pad_reg_upward): Update accordingly.
17331 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17332 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17333 targetm.calls.function_arg_padding.
17334 (FUNCTION_ARG_PADDING): Delete.
17335 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17336 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17337 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17338 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17339 (nios2_block_reg_padding): Return pad_direction instead of direction.
17340 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17341 instead of direction.
17342 (nios2_function_arg_padding): Likewise. Make static.
17343 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17344 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17345 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17346 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17347 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17348 (pa_function_arg_padding): Make static. Return pad_direction instead
17350 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17351 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17352 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17353 instead of direction. Use targetm.calls.function_arg_padding.
17354 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17355 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17356 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17357 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17358 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17360 (function_arg_padding): Rename to...
17361 (rs6000_function_arg_padding): ...this. Make static. Return
17362 pad_direction instead of direction.
17363 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17364 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17365 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17366 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17367 instead of direction. Use targetm.calls.function_arg_padding.
17368 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17369 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17370 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17371 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17372 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17373 (function_arg_padding): Rename to...
17374 (rs6000_function_arg_padding): ...this. Make static. Return
17375 pad_direction instead of direction.
17376 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17377 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17378 * config/s390/s390.c (s390_function_arg_padding): New function.
17379 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17380 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17381 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17382 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17383 (function_arg_padding): Rename to...
17384 (sparc_function_arg_padding): ...this. Make static. Return
17385 pad_direction instead of direction.
17386 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17387 * config/spu/spu.c (spu_function_arg_padding): New function.
17388 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17389 * system.h (FUNCTION_ARG_PADDING): Poison.
17391 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17392 Alan Hayward <alan.hayward@arm.com>
17393 David Sherwood <david.sherwood@arm.com>
17395 * target.def (modes_tieable_p): New hook.
17396 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17397 (TARGET_MODES_TIEABLE_P): ...this.
17398 * doc/tm.texi.in: Regenerate.
17399 * hooks.h (hook_bool_mode_mode_true): Declare.
17400 * hooks.c (hook_bool_mode_mode_true): New function.
17401 * combine.c (subst): Use targetm.modes_tieable_p instead of
17403 * dse.c (find_shift_sequence): Likewise.
17404 * expmed.c (extract_low_bits): Likewise.
17405 * lower-subreg.c: Include target.h.
17406 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17408 * rtlanal.c (rtx_cost): Likewise.
17409 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17410 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17411 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17412 (TARGET_MODES_TIEABLE_P): Redefine.
17413 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17414 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17415 (TARGET_MODES_TIEABLE_P): Redefine.
17416 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17417 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17418 (arc_modes_tieable_p): New function.
17419 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17420 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17421 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17422 (arm_modes_tieable_p): Make static.
17423 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17424 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17425 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17426 (TARGET_MODES_TIEABLE_P): Redefine.
17427 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17428 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17429 (TARGET_MODES_TIEABLE_P): Redefine.
17430 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17431 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17432 (cr16_modes_tieable_p): New function.
17433 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17434 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17435 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17436 (TRULY_NOOP_TRUNCATION): Update comment.
17437 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17438 (TRULY_NOOP_TRUNCATION): Update comment.
17439 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17440 (frv_modes_tieable_p): New function.
17441 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17442 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17443 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17444 (TARGET_MODES_TIEABLE_P): Redefine.
17445 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17446 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17447 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17448 (TARGET_MODES_TIEABLE_P): Redefine.
17449 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17450 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17451 (ia64_modes_tieable_p): New function.
17452 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17453 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17454 (iq2000_modes_tieable_p): New function.
17455 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17456 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17457 (lm32_modes_tieable_p): New function.
17458 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17459 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17460 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17461 (TARGET_MODES_TIEABLE_P): Redefine.
17462 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17463 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17464 (m32r_modes_tieable_p): New function.
17465 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17466 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17467 (m68k_modes_tieable_p): New function.
17468 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17469 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17470 (mcore_modes_tieable_p): New function.
17471 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17472 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17474 (TARGET_MODES_TIEABLE_P): Redefine.
17475 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17476 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17477 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17478 (TARGET_MODES_TIEABLE_P): Redefine.
17479 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17480 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17481 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17482 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17483 (mn10300_modes_tieable_p): ...this and make static.
17484 (TARGET_MODES_TIEABLE_P): Redefine.
17485 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17486 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17487 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17488 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17489 (msp430_modes_tieable_p): Make static.
17490 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17491 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17492 (TARGET_MODES_TIEABLE_P): Redefine.
17493 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17494 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17495 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17496 (TARGET_MODES_TIEABLE_P): Redefine.
17497 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17498 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17499 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17500 (TARGET_MODES_TIEABLE_P): Redefine.
17501 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17502 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17503 (pdp11_modes_tieable_p): New function.
17504 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17505 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17506 (rs6000_modes_tieable_p): New function.
17507 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17508 * config/powerpcspe/powerpcspe.md: Update comment.
17509 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17510 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17511 (TARGET_MODES_TIEABLE_P): Redefine.
17512 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17513 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17514 (rl78_modes_tieable_p): New function.
17515 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17516 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17517 (rs6000_modes_tieable_p): New function.
17518 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17519 * config/rs6000/rs6000.md: Update comment.
17520 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17521 * config/rx/rx.c (rx_modes_tieable_p): New function.
17522 (TARGET_MODES_TIEABLE_P): Redefine.
17523 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17524 * config/s390/s390.c (s390_modes_tieable_p): New function.
17525 (TARGET_MODES_TIEABLE_P): Redefine.
17526 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17527 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17528 (sh_modes_tieable_p): New function.
17529 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17530 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17531 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17532 (sparc_modes_tieable_p): Make static.
17533 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17534 * config/spu/spu.c (spu_modes_tieable_p): New function.
17535 (TARGET_MODES_TIEABLE_P): Redefine.
17536 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17537 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17538 (TARGET_MODES_TIEABLE_P): Redefine.
17539 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17540 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17541 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17542 * config/v850/v850.c (v850_modes_tieable_p): New function.
17543 (TARGET_MODES_TIEABLE_P): Redefine.
17544 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17545 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17546 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17547 (visium_modes_tieable_p): New function.
17548 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17549 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17550 (xtensa_modes_tieable_p): New function.
17551 * system.h (MODES_TIEABLE_P): Poison.
17553 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17554 Alan Hayward <alan.hayward@arm.com>
17555 David Sherwood <david.sherwood@arm.com>
17557 * target.def (hard_regno_mode_ok): New hook.
17558 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17559 (TARGET_HARD_REGNO_MODE_OK): ...this.
17560 * doc/tm.texi.in: Regenerate.
17561 * hooks.h (hook_bool_uint_mode_true): Declare.
17562 * hooks.c (hook_bool_uint_mode_true): New function.
17563 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17564 HARD_REGNO_MODE_OK.
17565 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17566 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17567 instead of HARD_REGNO_MODE_OK.
17568 * caller-save.c: Include target.h.
17569 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17570 HARD_REGNO_MODE_OK.
17571 * combine.c (can_combine_p): Likewise.
17572 (combinable_i3pat): Likewise.
17573 (can_change_dest_mode): Likewise.
17574 * expr.c (init_expr_target): Likewise.
17575 (convert_move): Likewise.
17576 (convert_modes): Likewise.
17577 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17578 (setup_prohibited_mode_move_regs): Likewise.
17579 * ira.h (target_ira): Likewise.
17580 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17581 * lra-constraints.c (process_alt_operands): Likewise.
17582 (split_reg): Likewise.
17583 * recog.c (peep2_find_free_register): Likewise.
17584 * ree.c (combine_reaching_defs): Likewise.
17585 * regcprop.c (maybe_mode_change): Likewise.
17586 * reginfo.c (init_reg_sets_1): Likewise.
17587 (choose_hard_reg_mode): Likewise.
17588 (simplifiable_subregs): Likewise.
17589 * regrename.c (check_new_reg_p): Likewise.
17590 * reload.c (find_valid_class): Likewise.
17591 (find_valid_class_1): Likewise.
17592 (reload_inner_reg_of_subreg): Likewise.
17593 (push_reload): Likewise.
17594 (combine_reloads): Likewise.
17595 (find_dummy_reload): Likewise.
17596 (find_reloads): Likewise.
17597 * reload1.c (find_reg): Likewise.
17598 (set_reload_reg): Likewise.
17599 (allocate_reload_reg): Likewise.
17600 (choose_reload_regs): Likewise.
17601 (reload_adjust_reg_for_temp): Likewise.
17602 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17603 (simplify_subreg_regno): Likewise.
17604 * sel-sched.c (init_regs_for_mode): Likewise.
17605 * varasm.c (make_decl_rtl): Likewise.
17606 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17607 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17608 HARD_REGNO_MODE_OK.
17609 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17610 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17611 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17612 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17613 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17614 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17615 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17616 (arc_mode_class): Delete.
17617 (HARD_REGNO_MODE_OK): Delete.
17618 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17619 (arc_hard_regno_mode_ok): Rename old array to...
17620 (arc_hard_regno_mode_ok_modes): ...this.
17621 (arc_conditional_register_usage): Update accordingly.
17622 (arc_mode_class): Make static.
17623 (arc_hard_regno_mode_ok): New function.
17624 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17625 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17626 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17627 (arm_hard_regno_mode_ok): Make static.
17628 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17629 HARD_REGNO_MODE_OK.
17630 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17631 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17632 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17634 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17635 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17636 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17637 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17638 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17639 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17640 * config/bfin/predicates.md (valid_reg_operand): Use
17641 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17642 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17643 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17644 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17645 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17646 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17647 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17648 (cr16_hard_regno_mode_ok): Make static and return a bool.
17649 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17650 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17651 (cris_hard_regno_mode_ok): New function.
17652 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17653 (epiphany_mode_class): Delete.
17654 (HARD_REGNO_MODE_OK): Delete.
17655 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17656 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17657 (hard_regno_mode_ok): Rename to...
17658 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17659 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17660 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17661 HARD_REGNO_MODE_OK.
17662 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17663 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17664 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17665 (frv_hard_regno_mode_ok): Make static and return a bool.
17666 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17667 HARD_REGNO_MODE_OK.
17668 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17669 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17670 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17671 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17673 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17674 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17675 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17676 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17678 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17679 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17680 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17681 (ia64_hard_regno_mode_ok): New function.
17682 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17683 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17684 (iq2000_hard_regno_mode_ok): New function.
17685 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17686 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17687 (lm32_hard_regno_mode_ok): New function.
17688 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17689 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17690 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17691 instead of HARD_REGNO_MODE_OK.
17692 (m32c_hard_regno_ok): Rename to...
17693 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17694 (m32c_cannot_change_mode_class): Update accordingly.
17695 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17696 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17697 (m32r_mode_class): Delete.
17698 (HARD_REGNO_MODE_OK): Delete.
17699 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17700 (m32r_hard_regno_mode_ok): Rename to...
17701 (m32r_hard_regno_modes): ...this.
17702 (m32r_mode_class): Make static.
17703 (m32r_hard_regno_mode_ok): New function.
17704 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17705 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17706 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17707 (m68k_hard_regno_mode_ok): Make static.
17708 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17709 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17710 (mcore_hard_regno_mode_ok): New function.
17711 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17712 (HARD_REGNO_MODE_OK): Delete.
17713 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17715 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17716 (microblaze_hard_regno_mode_ok): New function.
17717 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17718 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17719 (mips_hard_regno_mode_ok): Delete.
17720 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17721 (mips_hard_regno_mode_ok_p): ...this and make static.
17722 (mips_hard_regno_mode_ok_p): Rename to...
17723 (mips_hard_regno_mode_ok_uncached): ...this.
17724 (mips_hard_regno_mode_ok): New function.
17725 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17726 of HARD_REGNO_MODE_OK.
17727 (mips_option_override): Update after above name changes.
17728 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17729 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17730 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17731 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17732 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17733 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17734 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17735 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17736 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17737 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17738 (msp430_hard_regno_mode_ok): Make static and return a bool.
17739 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17740 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17741 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17743 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17744 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17745 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17746 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17747 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17748 (PA_HARD_REGNO_MODE_OK): ...this
17749 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17750 (PA_HARD_REGNO_MODE_OK): ...this.
17751 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17752 (pa_hard_regno_mode_ok): New function.
17753 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17754 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17755 (pdp11_hard_regno_mode_ok): New function.
17756 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17757 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17759 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17761 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17762 (rs6000_hard_regno_mode_ok): Rename to...
17763 (rs6000_hard_regno_mode_ok_uncached): ...this.
17764 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17765 (rs6000_hard_regno_mode_ok): New function.
17766 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17767 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17768 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17769 (riscv_hard_regno_mode_ok): ...this and make static.
17770 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17771 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17772 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17773 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17774 (rl78_hard_regno_mode_ok): Make static and return bool.
17775 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17776 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17778 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17779 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17780 (rs6000_hard_regno_mode_ok): Rename to...
17781 (rs6000_hard_regno_mode_ok_uncached): ...this.
17782 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17783 (rs6000_hard_regno_mode_ok): New function.
17784 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17785 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17786 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17787 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17788 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17789 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17790 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17791 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17792 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17793 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17794 (sh_hard_regno_mode_ok): Make static.
17795 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17796 instead of HARD_REGNO_MODE_OK.
17797 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17798 (sparc_mode_class): Delete.
17799 (HARD_REGNO_MODE_OK): Delete.
17800 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17801 (hard_regno_mode_classes): Make static.
17802 (sparc_mode_class): Likewise.
17803 (sparc_hard_regno_mode_ok): New function.
17804 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17805 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17806 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17808 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17809 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17810 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17811 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17812 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17813 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17814 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17815 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17816 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17817 (visium_hard_regno_mode_ok): New function.
17818 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17819 instead of HARD_REGNO_MODE_OK.
17820 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17821 (HARD_REGNO_MODE_OK): Delete.
17822 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17823 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17824 (xtensa_option_override): Update accordingly.
17825 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17826 (xtensa_hard_regno_mode_ok): New function.
17827 * system.h (HARD_REGNO_MODE_OK): Poison.
17829 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17830 Alan Hayward <alan.hayward@arm.com>
17831 David Sherwood <david.sherwood@arm.com>
17833 * target.def (hard_regno_call_part_clobbered): New hook.
17834 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17835 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17836 * doc/tm.texi: Regenerate.
17837 * hooks.h (hook_bool_uint_mode_false): Declare.
17838 * hooks.c (hook_bool_uint_mode_false): New function.
17839 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17840 * cselib.c (cselib_process_insn): Use
17841 targetm.hard_regno_call_part_clobbered instead of
17842 HARD_REGNO_CALL_PART_CLOBBERED.
17843 * ira-conflicts.c (ira_build_conflicts): Likewise.
17844 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17845 * lra-constraints.c (need_for_call_save_p): Likewise.
17846 * lra-lives.c: Include target.h.
17847 (check_pseudos_live_through_calls): Use
17848 targetm.hard_regno_call_part_clobbered instead of
17849 HARD_REGNO_CALL_PART_CLOBBERED.
17850 * regcprop.c: Include target.h.
17851 (copyprop_hardreg_forward_1): Use
17852 targetm.hard_regno_call_part_clobbered instead of
17853 HARD_REGNO_CALL_PART_CLOBBERED.
17854 * reginfo.c (choose_hard_reg_mode): Likewise.
17855 * regrename.c (check_new_reg_p): Likewise.
17856 * reload.c (find_equiv_reg): Likewise.
17857 * reload1.c (emit_reload_insns): Likewise.
17858 * sched-deps.c (deps_analyze_insn): Likewise.
17859 * sel-sched.c (init_regs_for_mode): Likewise.
17860 (mark_unavailable_hard_regs): Likewise.
17861 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17862 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17863 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17865 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17866 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17867 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17869 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17871 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17872 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17873 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17875 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17876 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17877 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17879 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17880 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17882 * config/powerpcspe/powerpcspe.c
17883 (rs6000_hard_regno_call_part_clobbered): New function.
17884 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17885 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17886 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17888 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17889 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17890 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17892 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17893 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17894 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17896 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17897 Alan Hayward <alan.hayward@arm.com>
17898 David Sherwood <david.sherwood@arm.com>
17900 * rtl.h (subreg_memory_offset): Declare.
17901 * emit-rtl.c (subreg_memory_offset): New function.
17902 * expmed.c (store_bit_field_1): Use it.
17903 * expr.c (undefined_operand_subword_p): Likewise.
17904 * simplify-rtx.c (simplify_subreg): Likewise.
17906 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17908 PR rtl-optimization/57448
17911 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17912 using atomic_load pattern.
17913 (expand_atomic_store): Likewise.
17915 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17918 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17919 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17920 handling. Use replace_call_with_value with NULL instead of
17921 gsi_replace, unlink_stmt_vdef and release_defs.
17923 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17926 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17928 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17931 * rtl.h (emit_library_call_value_1): Declare.
17932 (emit_library_call): Replace declaration with a series of overloads.
17933 Remove the parameter count argument.
17934 (emit_library_call_value): Likewise.
17935 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17936 with an "rtx_mode_t *".
17937 (emit_library_call_value): Delete.
17938 (emit_library_call): Likewise.
17939 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17940 (asan_emit_allocas_unpoison): Likewise.
17941 * builtins.c (expand_builtin_powi): Likewise.
17942 (expand_asan_emit_allocas_unpoison): Likewise.
17943 * cfgexpand.c (expand_main_function): Likewise.
17944 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17945 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17946 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17947 * config/arm/arm.c (arm_trampoline_init): Likewise.
17948 (arm_call_tls_get_addr): Likewise.
17949 (arm_expand_divmod_libfunc): Likewise.
17950 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17951 (smulsi3_highpart): Likewise.
17952 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17953 (c6x_expand_compare): Likewise.
17954 (c6x_expand_movmem): Likewise.
17955 * config/frv/frv.c (frv_trampoline_init): Likewise.
17956 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17957 (ix86_expand_divmod_libfunc): Likewise.
17958 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17959 (ia64_expand_compare): Likewise.
17960 (ia64_profile_hook): Likewise.
17961 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17962 (nonlocal_goto): Likewise.
17963 (restore_stack_nonlocal): Likewise.
17964 * config/m32r/m32r.c (block_move_call): Likewise.
17965 (m32r_trampoline_init): Likewise.
17966 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17967 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17968 (m68k_call_m68k_read_tp): Likewise.
17969 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17970 (microblaze_expand_divide): Likewise.
17971 * config/mips/mips.h (mips_args): Likewise.
17972 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17973 (MIPS_ICACHE_SYNC): Likewise.
17974 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17975 (nios2_trampoline_init): Likewise.
17976 * config/pa/pa.c (hppa_tls_call): Likewise.
17977 (pa_trampoline_init): Likewise.
17978 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17979 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17980 (expand_strn_compare): Likewise.
17981 (rs6000_generate_compare): Likewise.
17982 (rs6000_expand_float128_convert): Likewise.
17983 (output_profile_hook): Likewise.
17984 (rs6000_trampoline_init): Likewise.
17985 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17986 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17987 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17988 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17989 (rs6000_generate_compare): Likewise.
17990 (rs6000_expand_float128_convert): Likewise.
17991 (output_profile_hook): Likewise.
17992 (rs6000_trampoline_init): Likewise.
17993 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17994 * config/sh/sh.c (sh_trampoline_init): Likewise.
17995 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17996 (sparc_emit_float_lib_cmp): Likewise.
17997 (sparc32_initialize_trampoline): Likewise.
17998 (sparc64_initialize_trampoline): Likewise.
17999 (sparc_profile_hook): Likewise.
18000 * config/spu/spu.c (ea_load_store): Likewise.
18001 * config/spu/spu.md (floatunssidf2): Likewise.
18002 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
18003 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
18004 * config/visium/visium.c (expand_block_move_4): Likewise.
18005 (expand_block_move_2): Likewise.
18006 (expand_block_move_1): Likewise.
18007 (expand_block_set_4): Likewise.
18008 (expand_block_set_2): Likewise.
18009 (expand_block_set_1): Likewise.
18010 (visium_trampoline_init): Likewise.
18011 (visium_profile_hook): Likewise.
18012 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18013 (xtensa_setup_frame_addresses): Likewise.
18014 (xtensa_trampoline_init): Likewise.
18015 * except.c (sjlj_emit_function_enter): Likewise.
18016 (sjlj_emit_function_exit): Likewise.
18017 * explow.c (allocate_dynamic_stack_space): Likewise.
18018 (probe_stack_range): Likewise.
18019 * expr.c (convert_mode_scalar): Likewise.
18020 * optabs.c (expand_binop): Likewise.
18021 (expand_twoval_binop_libfunc): Likewise.
18022 (expand_unop): Likewise.
18023 (prepare_cmp_insn): Likewise.
18024 (prepare_float_lib_cmp): Likewise.
18025 (expand_float): Likewise.
18026 (expand_fix): Likewise.
18027 (expand_fixed_convert): Likewise.
18028 (maybe_emit_sync_lock_test_and_set): Likewise.
18029 (expand_atomic_compare_and_swap): Likewise.
18030 (expand_mem_thread_fence): Likewise.
18031 (expand_atomic_fetch_op): Likewise.
18033 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18035 * doc/generic.texi (OpenACC): Adjust URL.
18036 * doc/invoke.texi (C Dialect Options): Ditto.
18038 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18040 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18041 predicate for operand 1. Add (m,<S>) constraint.
18042 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18043 Prevent memory operand 1 with register operand 2.
18045 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18047 PR rtl-optimization/82024
18048 * combine.c (try_combine): If the combination result is a PARALLEL,
18049 and we only need to retain the SET in there that would be placed
18050 at I2, check that we can place that at I3 instead, before doing so.
18052 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18055 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18057 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18058 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18061 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18062 Jeff Law <law@redhat.com>
18064 * varasm.c (bss_initializer_p): Do not put constants into .bss
18065 (categorize_decl_for_section): Handle bss_initializer_p returning
18066 false when DECL_INITIAL is NULL.
18068 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18071 * config/s390/s390.c (s390_can_inline_p): New function.
18073 2017-09-01 Jeff Law <law@redhat.com>
18075 PR tree-optimization/82052
18076 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18077 Always initialize the returned slot after a hash table miss
18078 when INSERT is true.
18080 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18082 * config/s390/s390.md (mem_signal_fence): Remove.
18083 * doc/md.texi (mem_signal_fence): Remove.
18084 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18086 * target-insns.def (mem_signal_fence): Remove.
18088 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18091 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18094 * asan.c (create_odr_indicator): Strip name encoding from assembler
18095 name before appending it after __odr_asan_.
18097 2017-09-01 Martin Liska <mliska@suse.cz>
18099 PR tree-optimization/82059
18100 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18101 frequency only when an edge is redirected.
18103 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18105 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18106 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18107 (arc_conditional_register_usage): Remove ARC600 lp_count
18109 (arc_file_start): Emit Tag_ARC_CPU_variation.
18110 (arc_can_use_doloop_p): New conditions to use ZOLs.
18111 (hwloop_fail): New function.
18112 (hwloop_optimize): Likewise.
18113 (hwloop_pattern_reg): Likewise.
18114 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18115 (arc_reorg_loops): New function, calls reorg_loops.
18116 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18117 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18119 (arc_loop_hazard): Remove function, functionality moved into
18121 (arc_hazard): Remove arc_loop_hazard call.
18122 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18123 into hwloop_optimize.
18124 (arc_label_align): Remove ZOL handling.
18125 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18126 * config/arc/arc.md (doloop_begin): Remove pattern.
18127 (doloop_begin_i): Likewise.
18128 (doloop_end_i): Likewise.
18129 (doloop_fallback): Likewise.
18130 (doloop_fallback_m): Likewise.
18131 (doloop_end): Reimplement expand.
18132 (arc_lp): New pattern for LP instruction.
18133 (loop_end): New pattern.
18134 (loop_fail): Likewise.
18135 (decrement_and_branch_until_zero): Likewise.
18136 * config/arc/arc.opt (mlpc-width): New option.
18137 * doc/invoke.texi (mlpc-width): Document option.
18139 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18141 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18142 (arc_ccfsm_advance): Fix checking for delay slots.
18143 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18145 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18147 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18149 (movhi_insn): Update store instruction constraint which are saving
18150 6-bit short immediates.
18151 (movsi_insn): Consider also short scaled load operations.
18152 (zero_extendhisi2_i): Use Usd constraint instead of T.
18153 (extendhisi2_i): Add q constraint.
18154 (arc_clzsi2): Add type and length attributes.
18155 (arc_ctzsi2): Likewise.
18156 * config/arc/constraints.md (Usc): Update constraint, the
18157 assembler can parse two relocations for a single instruction.
18159 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18161 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18162 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18164 2017-08-31 Olivier Hainque <hainque@adacore.com>
18166 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18167 match as powerpc-wrs-vxworks*.
18169 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18171 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18172 register constraint for by-element operand.
18173 (aarch64_mls_elt_merge<mode>): Likewise.
18175 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18177 * config/arc/arc.c (arc_can_follow_jump): Check for short
18180 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18182 * config.gcc: Use g.opt for arc.
18183 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18184 functionality moved to ...
18185 (legitimate_scaled_address_p): New function, ...here.
18186 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18187 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18188 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18190 (arc_override_options): Handle G option.
18191 (arc_output_pic_addr_const): Correct function definition.
18192 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18193 (arc_decl_anon_ns_mem_p): Delete.
18194 (arc_in_small_data_p): Overhaul this function to take into
18195 consideration the value given via G option.
18196 (arc_rewrite_small_data_1): Renamed and corrected old
18197 arc_rewrite_small_data function.
18198 (arc_rewrite_small_data): New function.
18199 (small_data_pattern): Don't use pic_offset_table_rtx.
18200 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18201 * config/arc/simdext.md (movmisalignv2hi): Use
18202 prepare_move_operands function.
18204 (movmisalign*): Likewise.
18205 * doc/invoke.texi (ARC options): Document -G option.
18207 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18209 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18211 * config/arc/arc.c (arc_print_operand): Output scalled address for
18212 sdata whenever is possible.
18213 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18214 load/stores are available.
18215 (compact_sda_memory_operand): Check for the alignment required by
18216 code density instructions.
18217 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18219 * config/arc/constraints.md (Usd): Update constraint.
18220 (Us0): New constraint.
18221 (Usc): Update constraint.
18223 2017-08-31 Richard Biener <rguenther@suse.de>
18225 PR middle-end/82054
18226 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18227 function only once.
18229 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18231 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18232 Resize type_signature.
18234 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18235 Alan Hayward <alan.hayward@arm.com>
18236 David Sherwood <david.sherwood@arm.com>
18238 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18239 subregs whose inner modes can be stored in GPRs.
18240 (aarch64_classify_index): Likewise.
18242 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18243 Alan Hayward <alan.hayward@arm.com>
18244 David Sherwood <david.sherwood@arm.com>
18246 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18247 (V_INT_EQUIV): ...this.
18248 (v_cmp_result): Rename to...
18249 (v_int_equiv): ...this.
18250 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18251 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18252 (copysign<mode>3): Likewise.
18253 (aarch64_simd_bsl<mode>_internal): Likewise.
18254 (aarch64_simd_bsl<mode>): Likewise.
18255 (vec_cmp<mode><mode>): Likewise.
18256 (vcond<mode><mode>): Likewise.
18257 (vcond<v_cmp_mixed><mode>): Likewise.
18258 (vcondu<mode><v_cmp_mixed>): Likewise.
18259 (aarch64_cm<optab><mode>): Likewise.
18260 (aarch64_cmtst<mode>): Likewise.
18261 (aarch64_fac<optab><mode>): Likewise.
18262 (vec_perm_const<mode>): Likewise.
18263 (vcond_mask_<mode><v_cmp_result>): Rename to...
18264 (vcond_mask_<mode><v_int_equiv>): ...this.
18265 (vec_cmp<mode><v_cmp_result>): Rename to...
18266 (vec_cmp<mode><v_int_equiv>): ...this.
18268 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18269 Alan Hayward <alan.hayward@arm.com>
18270 David Sherwood <david.sherwood@arm.com>
18272 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18274 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18275 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18276 (UNSPEC_LD4_DREG): New unspecs.
18277 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18278 (aarch64_ld2<mode>_dreg_be): Replace with...
18279 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18281 (aarch64_ld3<mode>_dreg_le)
18282 (aarch64_ld3<mode>_dreg_be): Replace with...
18283 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18285 (aarch64_ld4<mode>_dreg_le)
18286 (aarch64_ld4<mode>_dreg_be): Replace with...
18287 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18290 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18292 PR tree-optimization/81987
18293 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18294 insert an initializer in a location not dominated by the stride
18297 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18299 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18300 on the entire header of the finally block in the fallthru case.
18302 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18304 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18306 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18308 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18309 rs6000_emit_move_from_cr and call renamed function.
18310 (rs6000_emit_prologue): Call renamed functions.
18311 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18312 movesi_from_cr, remove volatile CRs.
18314 2017-08-30 Jon Beniston <jon@beniston.com>
18315 Richard Biener <rguenther@suse.de>
18317 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18318 of VECTOR_MODE_P check.
18319 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18320 element vector types.
18322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18324 * df.h (df_read_modify_subreg_p): Remove in favor of...
18325 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18326 const_rtx instead of an rtx.
18327 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18328 * df-problems.c (df_word_lr_mark_ref): Likewise.
18329 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18330 (mark_pseudo_reg_dead): Likewise.
18331 (mark_ref_dead): Likewise.
18332 * reginfo.c (init_subregs_of_mode): Likewise.
18333 * sched-deps.c (sched_analyze_1): Likewise.
18334 * df-scan.c (df_def_record_1): Likewise.
18335 (df_uses_record): Likewise.
18336 (df_read_modify_subreg_p): Remove in favor of...
18337 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18338 const_rtx instead of an rtx.
18340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18341 Alan Hayward <alan.hayward@arm.com>
18342 David Sherwood <david.sherwood@arm.com>
18344 * rtl.h (partial_subreg_p): New function.
18345 * caller-save.c (save_call_clobbered_regs): Use it.
18346 * calls.c (expand_call): Likewise.
18347 * combine.c (combinable_i3pat): Likewise.
18348 (simplify_set): Likewise.
18349 (make_extraction): Likewise.
18350 (make_compound_operation_int): Likewise.
18351 (gen_lowpart_or_truncate): Likewise.
18352 (force_to_mode): Likewise.
18353 (make_field_assignment): Likewise.
18354 (reg_truncated_to_mode): Likewise.
18355 (record_truncated_value): Likewise.
18356 (move_deaths): Likewise.
18357 * cse.c (record_jump_cond): Likewise.
18358 (cse_insn): Likewise.
18359 * cselib.c (cselib_lookup_1): Likewise.
18360 * expmed.c (extract_bit_field_using_extv): Likewise.
18361 * function.c (assign_parm_setup_reg): Likewise.
18362 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18363 * ira-build.c (create_insn_allocnos): Likewise.
18364 * lra-coalesce.c (merge_pseudos): Likewise.
18365 * lra-constraints.c (match_reload): Likewise.
18366 (simplify_operand_subreg): Likewise.
18367 (curr_insn_transform): Likewise.
18368 * lra-lives.c (process_bb_lives): Likewise.
18369 * lra.c (new_insn_reg): Likewise.
18370 (lra_substitute_pseudo): Likewise.
18371 * regcprop.c (mode_change_ok): Likewise.
18372 (maybe_mode_change): Likewise.
18373 (copyprop_hardreg_forward_1): Likewise.
18374 * reload.c (push_reload): Likewise.
18375 (find_reloads): Likewise.
18376 (find_reloads_subreg_address): Likewise.
18377 * reload1.c (alter_reg): Likewise.
18378 (eliminate_regs_1): Likewise.
18379 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18381 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18383 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18384 back to if statements, including unpack.
18386 2017-08-30 Martin Liska <mliska@suse.cz>
18388 PR inline-asm/82001
18389 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18391 (func_checker::compare_asm_inputs_outputs): ... this function.
18392 (func_checker::compare_gimple_asm): Use the function to compare
18393 also ASM constrains.
18394 * ipa-icf-gimple.h: Rename the function.
18396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18397 Alan Hayward <alan.hayward@arm.com>
18398 David Sherwood <david.sherwood@arm.com>
18400 * coretypes.h (complex_mode): New type.
18401 * gdbhooks.py (build_pretty_printer): Handle it.
18402 * machmode.h (complex_mode): New class.
18403 (complex_mode::includes_p): New function.
18404 (is_complex_int_mode): Likewise.
18405 (is_complex_float_mode): Likewise.
18406 * genmodes.c (get_mode_class): Handle complex mode classes.
18407 * function.c (expand_function_end): Use is_complex_int_mode.
18409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18410 Alan Hayward <alan.hayward@arm.com>
18411 David Sherwood <david.sherwood@arm.com>
18413 * coretypes.h (scalar_mode_pod): New typedef.
18414 * gdbhooks.py (build_pretty_printer): Handle it.
18415 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18416 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18417 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18418 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18419 as_a <scalar_mode>.
18421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18422 Alan Hayward <alan.hayward@arm.com>
18423 David Sherwood <david.sherwood@arm.com>
18425 * machmode.h (mode_for_vector): Take a scalar_mode instead
18427 * stor-layout.c (mode_for_vector): Likewise.
18428 * explow.c (promote_mode): Use as_a <scalar_mode>.
18429 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18432 Alan Hayward <alan.hayward@arm.com>
18433 David Sherwood <david.sherwood@arm.com>
18435 * target.def (preferred_simd_mode): Take a scalar_mode
18436 instead of a machine_mode.
18437 * targhooks.h (default_preferred_simd_mode): Likewise.
18438 * targhooks.c (default_preferred_simd_mode): Likewise.
18439 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18440 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18441 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18442 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18443 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18444 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18445 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18446 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18448 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18449 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18450 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18451 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18452 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18453 * doc/tm.texi: Regenerate.
18454 * optabs-query.c (can_vec_mask_load_store_p): Return false for
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 * target.def (scalar_mode_supported_p): Take a scalar_mode
18462 instead of a machine_mode.
18463 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18464 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18465 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18466 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18467 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18468 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18469 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18470 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18471 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18472 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18473 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18474 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18475 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18476 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18478 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18479 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18480 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18481 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18482 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18484 * doc/tm.texi: Regenerate.
18486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18487 Alan Hayward <alan.hayward@arm.com>
18488 David Sherwood <david.sherwood@arm.com>
18490 * coretypes.h (opt_scalar_mode): New typedef.
18491 * gdbhooks.py (build_pretty_printers): Handle it.
18492 * machmode.h (mode_iterator::get_2xwider): Add overload for
18494 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18496 * expr.c (convert_mode_scalar): Likewise.
18497 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18498 * optabs.c (expand_float): Likewise.
18499 (expand_fix): Likewise.
18500 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18503 Alan Hayward <alan.hayward@arm.com>
18504 David Sherwood <david.sherwood@arm.com>
18506 * optabs.c (expand_float): Explicitly check for scalars before
18507 using a branching expansion.
18508 (expand_fix): Likewise.
18510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18511 Alan Hayward <alan.hayward@arm.com>
18512 David Sherwood <david.sherwood@arm.com>
18514 * expr.c (convert_mode): Split scalar handling out into...
18515 (convert_mode_scalar): ...this new function. Treat the modes
18518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18519 Alan Hayward <alan.hayward@arm.com>
18520 David Sherwood <david.sherwood@arm.com>
18522 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18524 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18527 Alan Hayward <alan.hayward@arm.com>
18528 David Sherwood <david.sherwood@arm.com>
18530 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18531 rather than a machine_mode.
18532 (fixed_from_string): Likewise.
18533 (fixed_convert): Likewise.
18534 (fixed_convert_from_int): Likewise.
18535 (fixed_convert_from_real): Likewise.
18536 (real_convert_from_fixed): Likewise.
18537 * fixed-value.c (fixed_from_double_int): Likewise.
18538 (fixed_from_string): Likewise.
18539 (fixed_convert): Likewise.
18540 (fixed_convert_from_int): Likewise.
18541 (fixed_convert_from_real): Likewise.
18542 (real_convert_from_fixed): Likewise.
18543 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18546 Alan Hayward <alan.hayward@arm.com>
18547 David Sherwood <david.sherwood@arm.com>
18549 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18550 of separate mode class checks. Do not allow vector modes here.
18551 (immed_wide_int_const): Use as_a <scalar_mode>.
18552 * explow.c (trunc_int_for_mode): Likewise.
18553 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18554 (wi::shwi): Likewise.
18555 (wi::min_value): Likewise.
18556 (wi::max_value): Likewise.
18557 * dwarf2out.c (loc_descriptor): Likewise.
18558 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18559 for CONST_WIDE_INT.
18561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18562 Alan Hayward <alan.hayward@arm.com>
18563 David Sherwood <david.sherwood@arm.com>
18565 * tree.h (SCALAR_TYPE_MODE): New macro.
18566 * expr.c (expand_expr_addr_expr_1): Use it.
18567 (expand_expr_real_2): Likewise.
18568 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18569 (fold_convert_const_fixed_from_int): Likewise.
18570 (fold_convert_const_fixed_from_real): Likewise.
18571 (native_encode_fixed): Likewise
18572 (native_encode_complex): Likewise
18573 (native_encode_vector): Likewise.
18574 (native_interpret_fixed): Likewise.
18575 (native_interpret_real): Likewise.
18576 (native_interpret_complex): Likewise.
18577 (native_interpret_vector): Likewise.
18578 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18579 (simd_clone_adjust_argument_types): Likewise.
18580 (simd_clone_init_simd_arrays): Likewise.
18581 (simd_clone_adjust): Likewise.
18582 * stor-layout.c (layout_type): Likewise.
18583 * tree.c (build_minus_one_cst): Likewise.
18584 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18585 * tree-inline.c (estimate_move_cost): Likewise.
18586 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18587 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18588 (vectorizable_reduction): Likewise.
18589 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18590 (vect_recog_mixed_size_cond_pattern): Likewise.
18591 (check_bool_pattern): Likewise.
18592 (adjust_bool_pattern): Likewise.
18593 (search_type_for_mask_1): Likewise.
18594 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18595 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18596 (vectorizable_load): Likewise.
18597 (vectorizable_store): Likewise.
18598 * ubsan.c (ubsan_encode_value): Likewise.
18599 * varasm.c (output_constant): Likewise.
18601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18602 Alan Hayward <alan.hayward@arm.com>
18603 David Sherwood <david.sherwood@arm.com>
18605 * coretypes.h (scalar_mode): New class.
18606 * machmode.h (scalar_mode): Likewise.
18607 (scalar_mode::includes_p): New function.
18608 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18609 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18610 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18611 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18612 * expmed.c (store_bit_field_1): Likewise.
18613 (extract_bit_field_1): Likewise.
18614 * expr.c (write_complex_part): Likewise.
18615 (read_complex_part): Likewise.
18616 (emit_move_complex_push): Likewise.
18617 (expand_expr_real_2): Likewise.
18618 * function.c (assign_parm_setup_reg): Likewise.
18619 (assign_parms_unsplit_complex): Likewise.
18620 * optabs.c (expand_binop): Likewise.
18621 * rtlanal.c (subreg_get_info): Likewise.
18622 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18623 * varasm.c (output_constant_pool_2): Likewise.
18625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18626 Alan Hayward <alan.hayward@arm.com>
18627 David Sherwood <david.sherwood@arm.com>
18629 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18631 (expmed_mult_highpart_optab): Likewise.
18632 (expmed_mult_highpart): Likewise.
18634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18635 Alan Hayward <alan.hayward@arm.com>
18636 David Sherwood <david.sherwood@arm.com>
18638 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18639 instead of a machine_mode.
18640 (builtin_memset_read_str): Likewise.
18641 * builtins.c (c_readstr): Likewise.
18642 (builtin_memcpy_read_str): Likewise.
18643 (builtin_strncpy_read_str): Likewise.
18644 (builtin_memset_read_str): Likewise.
18645 (builtin_memset_gen_str): Likewise.
18646 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18647 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18648 instead of a machine_mode.
18649 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18651 (make_extraction): Likewise.
18652 (try_widen_shift_mode): Take and return scalar_int_modes instead
18654 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18655 a scalar_int_mode instead of a machine_mode.
18656 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18657 (avr_addr_space_pointer_mode): Likewise.
18658 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18659 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18660 (msp430_unwind_word_mode): Likewise.
18661 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18662 (spu_addr_space_pointer_mode): Likewise.
18663 (spu_addr_space_address_mode): Likewise.
18664 (spu_libgcc_cmp_return_mode): Likewise.
18665 (spu_libgcc_shift_count_mode): Likewise.
18666 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18667 (rl78_addr_space_pointer_mode): Likewise.
18668 (fl78_unwind_word_mode): Likewise.
18669 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18671 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18672 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18673 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18674 (mips_valid_pointer_mode): Likewise.
18675 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18676 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18677 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18679 (ft32_addr_space_address_mode): Likewise.
18680 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18681 scalar_int_mode instead of a machine_mode.
18682 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18684 (m32c_addr_space_address_mode): Likewise.
18685 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18686 (rs6000_eh_return_filter_mode): Likewise.
18687 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18688 (rs6000_eh_return_filter_mode): Likewise.
18689 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18690 (s390_libgcc_shift_count_mode): Likewise.
18691 (s390_unwind_word_mode): Likewise.
18692 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18694 * target.def (mode_rep_extended): Likewise.
18695 (valid_pointer_mode): Likewise.
18696 (addr_space.valid_pointer_mode): Likewise.
18697 (eh_return_filter_mode): Return a scalar_int_mode rather than
18699 (libgcc_cmp_return_mode): Likewise.
18700 (libgcc_shift_count_mode): Likewise.
18701 (unwind_word_mode): Likewise.
18702 (addr_space.pointer_mode): Likewise.
18703 (addr_space.address_mode): Likewise.
18704 * doc/tm.texi: Regenerate.
18705 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18707 (do_jump): Use scalar_int_mode for local variables.
18708 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18709 rather than a machine_mode.
18710 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18711 (scompare_loc_descriptor_wide): Likewise.
18712 (scompare_loc_descriptor_narrow): Likewise.
18713 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18715 * except.c (sjlj_emit_dispatch_table): Likewise.
18716 (expand_builtin_eh_copy_values): Likewise.
18717 * explow.c (convert_memory_address_addr_space_1): Likewise.
18718 Take a scalar_int_mode rather than a machine_mode.
18719 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18720 than a machine_mode.
18721 (memory_address_addr_space): Use scalar_int_mode for local variables.
18722 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18723 rather than a machine_mode.
18724 * expmed.c (mask_rtx): Likewise.
18725 (init_expmed_one_conv): Likewise.
18726 (expand_mult_highpart_adjust): Likewise.
18727 (extract_high_half): Likewise.
18728 (expmed_mult_highpart_optab): Likewise.
18729 (expmed_mult_highpart): Likewise.
18730 (expand_smod_pow2): Likewise.
18731 (expand_sdiv_pow2): Likewise.
18732 (emit_store_flag_int): Likewise.
18733 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18735 (extract_low_bits): Likewise.
18736 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18738 * expr.c (pieces_addr::adjust): Likewise.
18739 (can_store_by_pieces): Likewise.
18740 (store_by_pieces): Likewise.
18741 (clear_by_pieces_1): Likewise.
18742 (expand_expr_addr_expr_1): Likewise.
18743 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18744 (expand_expr_real_1): Likewise.
18745 (try_casesi): Likewise.
18746 * final.c (shorten_branches): Likewise.
18747 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18748 type of "mode" to machine_mode.
18749 * internal-fn.c (expand_arith_overflow_result_store): Take a
18750 scalar_int_mode rather than a machine_mode.
18751 (expand_mul_overflow): Use scalar_int_mode for local variables.
18752 * loop-doloop.c (doloop_modify): Likewise.
18753 (doloop_optimize): Likewise.
18754 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18755 than a machine_mode.
18756 (expand_doubleword_shift_condmove): Likewise.
18757 (expand_doubleword_shift): Likewise.
18758 (expand_doubleword_clz): Likewise.
18759 (expand_doubleword_popcount): Likewise.
18760 (expand_doubleword_parity): Likewise.
18761 (expand_absneg_bit): Use scalar_int_mode for local variables.
18762 (prepare_float_lib_cmp): Likewise.
18763 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18764 rather than a machine_mode.
18765 (convert_memory_address_addr_space): Likewise.
18766 (get_mode_bounds): Likewise.
18767 (get_address_mode): Return a scalar_int_mode rather than a
18769 * rtlanal.c (get_address_mode): Likewise.
18770 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18771 than a machine_mode.
18772 * targhooks.c (default_mode_rep_extended): Likewise.
18773 (default_valid_pointer_mode): Likewise.
18774 (default_addr_space_valid_pointer_mode): Likewise.
18775 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18776 than a machine_mode.
18777 (default_libgcc_cmp_return_mode): Likewise.
18778 (default_libgcc_shift_count_mode): Likewise.
18779 (default_unwind_word_mode): Likewise.
18780 (default_addr_space_pointer_mode): Likewise.
18781 (default_addr_space_address_mode): Likewise.
18782 * targhooks.h (default_eh_return_filter_mode): Likewise.
18783 (default_libgcc_cmp_return_mode): Likewise.
18784 (default_libgcc_shift_count_mode): Likewise.
18785 (default_unwind_word_mode): Likewise.
18786 (default_addr_space_pointer_mode): Likewise.
18787 (default_addr_space_address_mode): Likewise.
18788 (default_mode_rep_extended): Take a scalar_int_mode rather than
18790 (default_valid_pointer_mode): Likewise.
18791 (default_addr_space_valid_pointer_mode): Likewise.
18792 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18794 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18795 rather than a machine_mode.
18796 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18797 for local variables.
18798 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18799 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18800 than a machine_mode.
18802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18803 Alan Hayward <alan.hayward@arm.com>
18804 David Sherwood <david.sherwood@arm.com>
18806 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18807 the mode argument to scalar_int_mode.
18808 (do_jump_by_parts_zero_rtx): Likewise.
18809 (do_jump_by_parts_equality_rtx): Likewise.
18810 (do_jump_by_parts_greater): Take a mode argument.
18811 (do_jump_by_parts_equality): Likewise.
18812 (do_jump_1): Update calls accordingly.
18814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18815 Alan Hayward <alan.hayward@arm.com>
18816 David Sherwood <david.sherwood@arm.com>
18818 * is-a.h (safe_dyn_cast): New function.
18819 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18820 (jump_table_for_label): Likewise.
18821 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18822 instead of an rtx_insn *.
18823 (shorten_branches): Use dyn_cast instead of LABEL_P and
18824 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18825 rtx_jump_table_data::get_data_mode.
18826 (final_scan_insn): Likewise.
18828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18829 Alan Hayward <alan.hayward@arm.com>
18830 David Sherwood <david.sherwood@arm.com>
18832 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18833 trying to combine a full-register integer set with a subreg
18836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18837 Alan Hayward <alan.hayward@arm.com>
18838 David Sherwood <david.sherwood@arm.com>
18840 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18841 that is always either address_mode or pointer_mode.
18843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18844 Alan Hayward <alan.hayward@arm.com>
18845 David Sherwood <david.sherwood@arm.com>
18847 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18848 when the two are known to be equal.
18850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18851 Alan Hayward <alan.hayward@arm.com>
18852 David Sherwood <david.sherwood@arm.com>
18854 * simplify-rtx.c (simplify_const_unary_operation): Use
18855 is_a <scalar_int_mode> instead of checking for a nonzero
18856 precision. Forcibly convert op_mode to a scalar_int_mode
18857 in that case. More clearly differentiate the operand and
18858 result modes and use the former when deciding what the value
18859 of a count-bits operation should be. Use is_int_mode instead
18860 of checking for a MODE_INT. Remove redundant check for whether
18861 this mode has a zero precision.
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 * optabs.c (widen_leading): Change the type of the mode argument
18868 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18869 (widen_bswap): Likewise.
18870 (expand_parity): Likewise.
18871 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18872 (expand_ffs): Likewise.
18873 (epand_unop): Check for scalar integer modes before calling the
18876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18877 Alan Hayward <alan.hayward@arm.com>
18878 David Sherwood <david.sherwood@arm.com>
18880 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18882 (expand_expr_real_1): Update call accordingly.
18884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18885 Alan Hayward <alan.hayward@arm.com>
18886 David Sherwood <david.sherwood@arm.com>
18888 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18889 value_mode arguments. Use scalar_int_mode internally.
18890 (store_bit_field_1): Rename the new integer mode from imode
18891 to op0_mode and use it instead of GET_MODE (op0). Update calls
18892 to store_split_bit_field, store_bit_field_using_insv and
18893 store_fixed_bit_field.
18894 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18895 Use scalar_int_mode internally. Use a bit count rather than a mode
18896 when calculating the largest bit size for get_best_mode.
18897 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18898 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18899 Remove assertion that OP0 has a scalar integer mode.
18900 (store_split_bit_field): Add op0_mode and value_mode arguments.
18901 Update calls to extract_fixed_bit_field.
18902 (extract_bit_field_using_extv): Add an op0_mode argument.
18903 Use scalar_int_mode internally.
18904 (extract_bit_field_1): Rename the new integer mode from imode to
18905 op0_mode and use it instead of GET_MODE (op0). Update calls to
18906 extract_split_bit_field, extract_bit_field_using_extv and
18907 extract_fixed_bit_field.
18908 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18909 to extract_split_bit_field and extract_fixed_bit_field_1.
18910 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18911 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18912 on the target mode.
18913 (extract_split_bit_field): Add an op0_mode argument. Update call
18914 to extract_fixed_bit_field.
18916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18917 Alan Hayward <alan.hayward@arm.com>
18918 David Sherwood <david.sherwood@arm.com>
18920 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18921 * explow.c (hard_function_value): Likewise.
18922 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18923 convert_to_mode call outside the loop.
18924 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18925 for the mode iterator. Require the mode specified by max_pieces
18927 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18929 (copy_blkmode_to_reg): Likewise.
18930 (set_storage_via_setmem): Likewise.
18931 * optabs.c (prepare_cmp_insn): Likewise.
18932 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18933 * stor-layout.c (finish_bitfield_representative): Likewise.
18935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18936 Alan Hayward <alan.hayward@arm.com>
18937 David Sherwood <david.sherwood@arm.com>
18939 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18940 * expr.c (convert_move): Use them.
18941 (convert_modes): Likewise.
18942 (store_expr_with_bounds): Likewise.
18944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18945 Alan Hayward <alan.hayward@arm.com>
18946 David Sherwood <david.sherwood@arm.com>
18948 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18949 parameter for the mode of "x". Remove the "known_x", "known_mode"
18950 and "known_ret" arguments. Change the type of the mode argument
18951 to scalar_int_mode.
18952 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18953 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18954 (reg_num_sign_bit_copies_for_combine): Likewise.
18955 * rtlanal.c (nonzero_bits1): Likewise.
18956 (num_sign_bit_copies1): Likewise.
18957 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18958 (reg_num_sign_bit_copies_general): Likewise.
18959 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18960 (reg_nonzero_bits_general): Likewise.
18962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18963 Alan Hayward <alan.hayward@arm.com>
18964 David Sherwood <david.sherwood@arm.com>
18966 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18967 than in subroutines. Return 1 for non-integer modes.
18968 (cached_num_sign_bit_copies): Change the type of the mode parameter
18969 to scalar_int_mode.
18970 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18971 classes. Handle CONST_INT_P first and then check whether X also
18972 has a scalar integer mode. Check the same thing for inner registers
18973 of a SUBREG and for values that are being extended or truncated.
18975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18976 Alan Hayward <alan.hayward@arm.com>
18977 David Sherwood <david.sherwood@arm.com>
18979 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18980 in subroutines. Return the mode mask for non-integer modes.
18981 (cached_nonzero_bits): Change the type of the mode parameter
18982 to scalar_int_mode.
18983 (nonzero_bits1): Likewise. Remove early exit for other mode
18984 classes. Handle CONST_INT_P first and then check whether X
18985 also has a scalar integer mode.
18987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18988 Alan Hayward <alan.hayward@arm.com>
18989 David Sherwood <david.sherwood@arm.com>
18991 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18992 Return a scalar_int_mode and assert that the size is greater than
18994 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18995 (op_by_pieces_d::op_by_pieces_d): Likewise.
18996 (op_by_pieces_d::run): Likewise.
18997 (can_store_by_pieces): Likewise.
18999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19000 Alan Hayward <alan.hayward@arm.com>
19001 David Sherwood <david.sherwood@arm.com>
19003 * combine.c (extract_left_shift): Add a mode argument and update
19005 (make_compound_operation_int): Change the type of the mode parameter
19006 to scalar_int_mode and update the call to extract_left_shift.
19008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19009 Alan Hayward <alan.hayward@arm.com>
19010 David Sherwood <david.sherwood@arm.com>
19012 * combine.c (simplify_and_const_int): Change the type of the mode
19013 parameter to scalar_int_mode.
19014 (simplify_and_const_int_1): Likewise. Update recursive call.
19016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19017 Alan Hayward <alan.hayward@arm.com>
19018 David Sherwood <david.sherwood@arm.com>
19020 * combine.c (simplify_compare_const): Check that the mode is a
19021 scalar_int_mode (rather than VOIDmode) before testing its
19023 (simplify_comparison): Move COMPARISON_P handling out of the
19024 loop and restrict the latter part of the loop to scalar_int_modes.
19025 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19026 and when considering SUBREG_REGs. Use is_int_mode instead of
19027 checking GET_MODE_CLASS against MODE_INT.
19029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19030 Alan Hayward <alan.hayward@arm.com>
19031 David Sherwood <david.sherwood@arm.com>
19033 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19034 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19035 shift_unit_mode and for modes involved in scalar shifts.
19037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19038 Alan Hayward <alan.hayward@arm.com>
19039 David Sherwood <david.sherwood@arm.com>
19041 * combine.c (force_int_to_mode): New function, split out from...
19042 (force_to_mode): ...here. Keep xmode up-to-date and use it
19043 instead of GET_MODE (x).
19045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19046 Alan Hayward <alan.hayward@arm.com>
19047 David Sherwood <david.sherwood@arm.com>
19049 * optabs-query.h (extraction_insn::struct_mode): Change type to
19050 opt_scalar_int_mode and update comment.
19051 (extraction_insn::field_mode): Change type to scalar_int_mode.
19052 (extraction_insn::pos_mode): Likewise.
19053 * combine.c (make_extraction): Update accordingly.
19054 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19055 (get_optab_extraction_insn): Likewise.
19056 * recog.c (simplify_while_replacing): Likewise.
19057 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19058 parameter to opt_scalar_int_mode.
19060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19061 Alan Hayward <alan.hayward@arm.com>
19062 David Sherwood <david.sherwood@arm.com>
19064 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19065 to a scalar_int_mode instead of a machine_mode.
19066 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19067 (get_best_mode): Return a boolean and use a pointer argument to store
19068 the selected mode. Replace the limit mode parameter with a bit limit.
19069 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19070 for the values returned by bit_field_mode_iterator::next_mode.
19071 (store_bit_field): Update call to get_best_mode.
19072 (store_fixed_bit_field): Likewise.
19073 (extract_fixed_bit_field): Likewise.
19074 * expr.c (optimize_bitfield_assignment_op): Likewise.
19075 * fold-const.c (optimize_bit_field_compare): Likewise.
19076 (fold_truth_andor_1): Likewise.
19077 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19078 Update for new type of m_mode.
19079 (get_best_mode): As above.
19081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19082 Alan Hayward <alan.hayward@arm.com>
19083 David Sherwood <david.sherwood@arm.com>
19085 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19086 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19087 (store_bit_field): Check is_a <scalar_int_mode> before calling
19088 strict_volatile_bitfield_p.
19089 (extract_bit_field): Likewise.
19091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19092 Alan Hayward <alan.hayward@arm.com>
19093 David Sherwood <david.sherwood@arm.com>
19095 * target.def (cstore_mode): Return a scalar_int_mode.
19096 * doc/tm.texi: Regenerate.
19097 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19098 * targhooks.h (default_cstore_mode): Likewise.
19099 * targhooks.c (default_cstore_mode): Likewise, using a forced
19101 * expmed.c (emit_cstore): Expect the target of the cstore to be
19104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19105 Alan Hayward <alan.hayward@arm.com>
19106 David Sherwood <david.sherwood@arm.com>
19108 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19110 (niter_desc): Likewise mode.
19111 (iv_analyze): Add a mode parameter.
19113 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19114 and change its type to scalar_int_mode.
19115 * loop-iv.c: Update commentary at head of file.
19116 (iv_constant): Pass the mode paraeter before the rtx it describes
19117 and change its type to scalar_int_mode. Remove VOIDmode handling.
19118 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19119 (iv_extend): Likewise.
19120 (shorten_into_mode): Likewise.
19121 (iv_add): Use scalar_int_mode.
19122 (iv_mult): Likewise.
19123 (iv_shift): Likewise.
19124 (canonicalize_iv_subregs): Likewise.
19125 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19126 it describes and change its mode to scalar_int_mode. Also change
19127 the type of the returned inner_mode to scalar_int_mode.
19128 (get_biv_step): Likewise, turning outer_mode from a pointer
19129 into a direct parameter. Update call to get_biv_step_1.
19130 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19131 iv_constant and get_biv_step.
19132 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19133 and change its type to scalar_int_mode. Don't initialise iv->mode
19134 to VOIDmode and remove later checks for its still being VOIDmode.
19135 Update calls to iv_analyze_op and iv_analyze_expr. Check
19136 is_a <scalar_int_mode> when changing the mode under consideration.
19137 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19138 Update call to iv_analyze_expr.
19139 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19140 inner register is not also a scalar_int_mode. Update call to
19142 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19143 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19144 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19145 separate mode class checks. Update calls to iv_analyze. Remove
19146 fix-up of VOIDmodes after iv_analyze_biv.
19147 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19148 don't have a scalar_int_mode. Update call to biv_p.
19150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19151 Alan Hayward <alan.hayward@arm.com>
19152 David Sherwood <david.sherwood@arm.com>
19154 * cfgexpand.c (convert_debug_memory_address): Use
19155 as_a <scalar_int_mode>.
19156 * combine.c (expand_compound_operation): Likewise.
19157 (make_extraction): Likewise.
19158 (change_zero_ext): Likewise.
19159 (simplify_comparison): Likewise.
19160 * cse.c (cse_insn): Likewise.
19161 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19162 (mem_loc_descriptor): Likewise.
19163 (loc_descriptor): Likewise.
19164 * expmed.c (init_expmed_one_mode): Likewise.
19165 (synth_mult): Likewise.
19166 (emit_store_flag_1): Likewise.
19167 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19168 of a comparison with size.
19169 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19170 (reduce_to_bit_field_precision): Likewise.
19171 * function.c (expand_function_end): Likewise.
19172 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19173 * loop-doloop.c (doloop_modify): Likewise.
19174 * optabs.c (expand_binop): Likewise.
19175 (expand_unop): Likewise.
19176 (expand_copysign_absneg): Likewise.
19177 (prepare_cmp_insn): Likewise.
19178 (maybe_legitimize_operand): Likewise.
19179 * recog.c (const_scalar_int_operand): Likewise.
19180 * rtlanal.c (get_address_mode): Likewise.
19181 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19182 (simplify_cond_clz_ctz): Likewise.
19183 * tree-nested.c (get_nl_goto_field): Likewise.
19184 * tree.c (build_vector_type_for_mode): Likewise.
19185 * var-tracking.c (use_narrower_mode): Likewise.
19187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19188 Alan Hayward <alan.hayward@arm.com>
19189 David Sherwood <david.sherwood@arm.com>
19191 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19192 * builtins.c (expand_builtin_signbit): Use it.
19193 * cfgexpand.c (expand_debug_expr): Likewise.
19194 * dojump.c (do_jump): Likewise.
19195 (do_compare_and_jump): Likewise.
19196 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19197 * expmed.c (make_tree): Likewise.
19198 * expr.c (expand_expr_real_2): Likewise.
19199 (expand_expr_real_1): Likewise.
19200 (try_casesi): Likewise.
19201 * fold-const-call.c (fold_const_call_ss): Likewise.
19202 * fold-const.c (unextend): Likewise.
19203 (extract_muldiv_1): Likewise.
19204 (fold_single_bit_test): Likewise.
19205 (native_encode_int): Likewise.
19206 (native_encode_string): Likewise.
19207 (native_interpret_int): Likewise.
19208 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19209 * internal-fn.c (expand_addsub_overflow): Likewise.
19210 (expand_neg_overflow): Likewise.
19211 (expand_mul_overflow): Likewise.
19212 (expand_arith_overflow): Likewise.
19213 * match.pd: Likewise.
19214 * stor-layout.c (layout_type): Likewise.
19215 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19216 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19217 * tree-ssanames.c (get_range_info): Likewise.
19218 * tree-switch-conversion.c (array_value_type) Likewise.
19219 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19220 (vect_recog_divmod_pattern): Likewise.
19221 (vect_recog_mixed_size_cond_pattern): Likewise.
19222 * tree-vrp.c (extract_range_basic): Likewise.
19223 (simplify_float_conversion_using_ranges): Likewise.
19224 * tree.c (int_fits_type_p): Likewise.
19225 * ubsan.c (instrument_bool_enum_load): Likewise.
19226 * varasm.c (mergeable_string_section): Likewise.
19227 (narrowing_initializer_constant_valid_p): Likewise.
19228 (output_constant): Likewise.
19230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19231 Alan Hayward <alan.hayward@arm.com>
19232 David Sherwood <david.sherwood@arm.com>
19234 * machmode.h (NARROWEST_INT_MODE): New macro.
19235 * expr.c (alignment_for_piecewise_move): Use it instead of
19236 GET_CLASS_NARROWEST_MODE (MODE_INT).
19237 (push_block): Likewise.
19238 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19240 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19243 Alan Hayward <alan.hayward@arm.com>
19244 David Sherwood <david.sherwood@arm.com>
19246 * postreload.c (move2add_valid_value_p): Change the type of the
19247 mode parameter to scalar_int_mode.
19248 (move2add_use_add2_insn): Add a mode parameter and use it instead
19250 (move2add_use_add3_insn): Likewise.
19251 (reload_cse_move2add): Update accordingly.
19253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19254 Alan Hayward <alan.hayward@arm.com>
19255 David Sherwood <david.sherwood@arm.com>
19257 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19259 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19260 * optabs.c (expand_unop): Likewise.
19262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19263 Alan Hayward <alan.hayward@arm.com>
19264 David Sherwood <david.sherwood@arm.com>
19266 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19267 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19268 (popcount_loc_descriptor): Likewise.
19269 (bswap_loc_descriptor): Likewise.
19270 (rotate_loc_descriptor): Likewise.
19271 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19272 calling the functions above.
19274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19275 Alan Hayward <alan.hayward@arm.com>
19276 David Sherwood <david.sherwood@arm.com>
19278 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19280 (try_combine): Likewise.
19281 (simplify_if_then_else): Likewise.
19282 * cse.c (cse_insn): Likewise.
19283 * dwarf2out.c (mem_loc_descriptor): Likewise.
19284 * emit-rtl.c (gen_lowpart_common): Likewise.
19285 * simplify-rtx.c (simplify_truncation): Likewise.
19286 (simplify_binary_operation_1): Likewise.
19287 (simplify_const_relational_operation): Likewise.
19288 (simplify_ternary_operation): Likewise.
19289 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19292 Alan Hayward <alan.hayward@arm.com>
19293 David Sherwood <david.sherwood@arm.com>
19295 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19296 * reload.c (push_reload): Likewise.
19297 (find_reloads): Likewise.
19299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19300 Alan Hayward <alan.hayward@arm.com>
19301 David Sherwood <david.sherwood@arm.com>
19303 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19304 (make_compound_operation_int): Likewise.
19305 (change_zero_ext): Likewise.
19306 * expr.c (convert_move): Likewise.
19307 (convert_modes): Likewise.
19308 * fwprop.c (forward_propagate_subreg): Likewise.
19309 * loop-iv.c (get_biv_step_1): Likewise.
19310 * optabs.c (widen_operand): Likewise.
19311 * postreload.c (move2add_valid_value_p): Likewise.
19312 * recog.c (simplify_while_replacing): Likewise.
19313 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19314 (simplify_binary_operation_1): Likewise. Remove redundant
19315 mode equality check.
19317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19318 Alan Hayward <alan.hayward@arm.com>
19319 David Sherwood <david.sherwood@arm.com>
19321 * combine.c (combine_simplify_rtx): Add checks for
19322 is_a <scalar_int_mode>.
19323 (simplify_if_then_else): Likewise.
19324 (make_field_assignment): Likewise.
19325 (simplify_comparison): Likewise.
19326 * ifcvt.c (noce_try_bitop): Likewise.
19327 * loop-invariant.c (canonicalize_address_mult): Likewise.
19328 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19331 Alan Hayward <alan.hayward@arm.com>
19332 David Sherwood <david.sherwood@arm.com>
19334 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19335 is_a <scalar_int_mode> instead of != BLKmode.
19337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19338 Alan Hayward <alan.hayward@arm.com>
19339 David Sherwood <david.sherwood@arm.com>
19341 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19342 instead of != VOIDmode.
19343 * combine.c (if_then_else_cond): Likewise.
19344 (change_zero_ext): Likewise.
19345 * dwarf2out.c (mem_loc_descriptor): Likewise.
19346 (loc_descriptor): Likewise.
19347 * rtlanal.c (canonicalize_condition): Likewise.
19348 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19351 Alan Hayward <alan.hayward@arm.com>
19352 David Sherwood <david.sherwood@arm.com>
19354 * simplify-rtx.c (simplify_binary_operation_1): Use
19355 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19358 Alan Hayward <alan.hayward@arm.com>
19359 David Sherwood <david.sherwood@arm.com>
19361 * wide-int.h (int_traits<unsigned char>) New class.
19362 (int_traits<unsigned short>) Likewise.
19363 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19364 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19366 * combine.c (set_nonzero_bits_and_sign_copies): Use
19367 is_a <scalar_int_mode>.
19368 (find_split_point): Likewise.
19369 (combine_simplify_rtx): Likewise.
19370 (simplify_logical): Likewise.
19371 (expand_compound_operation): Likewise.
19372 (expand_field_assignment): Likewise.
19373 (make_compound_operation): Likewise.
19374 (extended_count): Likewise.
19375 (change_zero_ext): Likewise.
19376 (simplify_comparison): Likewise.
19377 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19378 (ucompare_loc_descriptor): Likewise.
19379 (minmax_loc_descriptor): Likewise.
19380 (mem_loc_descriptor): Likewise.
19381 (loc_descriptor): Likewise.
19382 * expmed.c (init_expmed_one_mode): Likewise.
19383 * lra-constraints.c (lra_constraint_offset): Likewise.
19384 * optabs.c (prepare_libcall_arg): Likewise.
19385 * postreload.c (move2add_note_store): Likewise.
19386 * reload.c (operands_match_p): Likewise.
19387 * rtl.h (load_extend_op): Likewise.
19388 * rtlhooks.c (gen_lowpart_general): Likewise.
19389 * simplify-rtx.c (simplify_truncation): Likewise.
19390 (simplify_unary_operation_1): Likewise.
19391 (simplify_binary_operation_1): Likewise.
19392 (simplify_const_binary_operation): Likewise.
19393 (simplify_const_relational_operation): Likewise.
19394 (simplify_subreg): Likewise.
19395 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19396 * var-tracking.c (adjust_mems): Likewise.
19397 (prepare_call_arguments): Likewise.
19399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19400 Alan Hayward <alan.hayward@arm.com>
19401 David Sherwood <david.sherwood@arm.com>
19403 * machmode.h (is_int_mode): New fuction.
19404 * combine.c (find_split_point): Use it.
19405 (combine_simplify_rtx): Likewise.
19406 (simplify_if_then_else): Likewise.
19407 (simplify_set): Likewise.
19408 (simplify_shift_const_1): Likewise.
19409 (simplify_comparison): Likewise.
19410 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19411 * cse.c (notreg_cost): Likewise.
19412 (cse_insn): Likewise.
19413 * cselib.c (cselib_lookup_1): Likewise.
19414 * dojump.c (do_jump_1): Likewise.
19415 (do_compare_rtx_and_jump): Likewise.
19416 * dse.c (get_call_args): Likewise.
19417 * dwarf2out.c (rtl_for_decl_init): Likewise.
19418 (native_encode_initializer): Likewise.
19419 * expmed.c (emit_store_flag_1): Likewise.
19420 (emit_store_flag): Likewise.
19421 * expr.c (convert_modes): Likewise.
19422 (store_field): Likewise.
19423 (expand_expr_real_1): Likewise.
19424 * fold-const.c (fold_read_from_constant_string): Likewise.
19425 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19426 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19427 * optabs.c (expand_binop): Likewise.
19428 (expand_unop): Likewise.
19429 (expand_abs_nojump): Likewise.
19430 (expand_one_cmpl_abs_nojump): Likewise.
19431 * simplify-rtx.c (mode_signbit_p): Likewise.
19432 (val_signbit_p): Likewise.
19433 (val_signbit_known_set_p): Likewise.
19434 (val_signbit_known_clear_p): Likewise.
19435 (simplify_relational_operation_1): Likewise.
19436 * tree.c (vector_type_mode): Likewise.
19438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19439 Alan Hayward <alan.hayward@arm.com>
19440 David Sherwood <david.sherwood@arm.com>
19442 * machmode.h (smallest_mode_for_size): Fix formatting.
19443 (smallest_int_mode_for_size): New function.
19444 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19445 instead of smallest_mode_for_size.
19446 * combine.c (make_extraction): Likewise.
19447 * config/arc/arc.c (arc_expand_movmem): Likewise.
19448 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19449 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19450 * config/s390/s390.c (s390_expand_insv): Likewise.
19451 * config/sparc/sparc.c (assign_int_registers): Likewise.
19452 * config/spu/spu.c (spu_function_value): Likewise.
19453 (spu_function_arg): Likewise.
19454 * coverage.c (get_gcov_type): Likewise.
19455 (get_gcov_unsigned_t): Likewise.
19456 * dse.c (find_shift_sequence): Likewise.
19457 * expmed.c (store_bit_field_1): Likewise.
19458 * expr.c (convert_move): Likewise.
19459 (store_field): Likewise.
19460 * internal-fn.c (expand_arith_overflow): Likewise.
19461 * optabs-query.c (get_best_extraction_insn): Likewise.
19462 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19463 * stor-layout.c (layout_type): Likewise.
19464 (initialize_sizetypes): Likewise.
19465 * targhooks.c (default_get_mask_mode): Likewise.
19466 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19469 Alan Hayward <alan.hayward@arm.com>
19470 David Sherwood <david.sherwood@arm.com>
19472 * machmode.h (opt_mode::else_blk): New function.
19473 (int_mode_for_mode): Declare.
19474 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19475 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19477 * cfgexpand.c (expand_debug_expr): Likewise.
19478 * combine.c (gen_lowpart_or_truncate): Likewise.
19479 (gen_lowpart_for_combine): Likewise.
19480 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19481 * config/avr/avr.c (avr_to_int_mode): Likewise.
19482 (avr_out_plus_1): Likewise.
19483 (avr_out_plus): Likewise.
19484 (avr_out_round): Likewise.
19485 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19486 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19487 (s390_expand_vcond): Likewise.
19488 * config/spu/spu.c (spu_split_immediate): Likewise.
19489 (spu_expand_mov): Likewise.
19490 * dse.c (get_stored_val): Likewise.
19491 * expmed.c (store_bit_field_1): Likewise.
19492 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19494 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19495 (extract_low_bits): Likewise.
19496 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19497 handling rather than repeating the check.
19498 (emit_group_store): Likewise.
19499 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19500 * optabs.c (expand_absneg_bit): Likewise.
19501 (expand_copysign_absneg): Likewise.
19502 (expand_copysign_bit): Likewise.
19503 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19504 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19505 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19506 * var-tracking.c (prepare_call_arguments): Likewise.
19507 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19508 int_mode_for_mode instead of mode_for_size.
19509 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19512 Alan Hayward <alan.hayward@arm.com>
19513 David Sherwood <david.sherwood@arm.com>
19515 * machmode.h (int_mode_for_size): New function.
19516 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19517 instead of mode_for_size.
19518 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19520 * combine.c (expand_field_assignment): Use int_mode_for_size
19521 instead of mode_for_size.
19522 (make_extraction): Likewise.
19523 (simplify_shift_const_1): Likewise.
19524 (simplify_comparison): Likewise.
19525 * dojump.c (do_jump): Likewise.
19526 * dwarf2out.c (mem_loc_descriptor): Likewise.
19527 * emit-rtl.c (init_derived_machine_modes): Likewise.
19528 * expmed.c (flip_storage_order): Likewise.
19529 (convert_extracted_bit_field): Likewise.
19530 * expr.c (copy_blkmode_from_reg): Likewise.
19531 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19532 * internal-fn.c (expand_mul_overflow): Likewise.
19533 * lower-subreg.c (simple_move): Likewise.
19534 * optabs-libfuncs.c (init_optabs): Likewise.
19535 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19536 * tree.c (vector_type_mode): Likewise.
19537 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19538 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19539 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19540 * tree-vect-stmts.c (vectorizable_load): Likewise.
19541 (vectorizable_store): Likewise.
19543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19544 Alan Hayward <alan.hayward@arm.com>
19545 David Sherwood <david.sherwood@arm.com>
19547 * coretypes.h (pod_mode): New type.
19548 (scalar_int_mode_pod): New typedef.
19549 * machmode.h (pod_mode): New class.
19550 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19551 * genmodes.c (emit_mode_int_n): Update accordingly.
19552 * lower-subreg.h (target_lower_subreg): Change type to
19553 scalar_int_mode_pod.
19554 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19555 scalar_int_mode_pod.
19557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19558 Alan Hayward <alan.hayward@arm.com>
19559 David Sherwood <david.sherwood@arm.com>
19561 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19562 machine_mode to scalar_int_mode.
19563 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19564 (rs6000_option_override_internal): Remove cast to int.
19565 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19566 machine_mode to scalar_int_mode.
19567 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19568 (rs6000_option_override_internal): Remove cast to int.
19569 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19570 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19572 * config/s390/s390.c (s390_expand_builtin): Likewise.
19573 * coretypes.h (scalar_int_mode): New type.
19574 (opt_scalar_int_mode): New typedef.
19575 * machmode.h (scalar_int_mode): New class.
19576 (scalar_int_mode::includes_p): New function.
19577 (byte_mode): Change type to scalar_int_mode.
19578 (word_mode): Likewise.
19579 (ptr_mode): Likewise.
19580 * emit-rtl.c (byte_mode): Likewise.
19581 (word_mode): Likewise.
19582 (ptr_mode): Likewise.
19583 (init_derived_machine_modes): Update accordingly.
19584 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19585 and MODE_PARTIAL_INT.
19586 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19587 opt_scalar_int_mode.
19589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19590 Alan Hayward <alan.hayward@arm.com>
19591 David Sherwood <david.sherwood@arm.com>
19593 * target.def (libgcc_floating_mode_supported_p): Take a
19595 * doc/tm.texi: Regenerate.
19596 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19598 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19599 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19603 Alan Hayward <alan.hayward@arm.com>
19604 David Sherwood <david.sherwood@arm.com>
19606 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19607 * doc/tm.texi: Regenerate.
19608 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19609 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19610 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19611 * targhooks.h (default_floatn_mode): Likewise.
19612 * targhooks.c (default_floatn_mode): Likewise.
19613 * tree.c (build_common_tree_nodes): Update accordingly.
19615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19616 Alan Hayward <alan.hayward@arm.com>
19617 David Sherwood <david.sherwood@arm.com>
19619 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19620 (mode_iterator::iterate_p): Likewise.
19621 (mode_iterator::get_wider): Likewise.
19622 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19625 Alan Hayward <alan.hayward@arm.com>
19626 David Sherwood <david.sherwood@arm.com>
19628 * coretypes.h (opt_scalar_float_mode): New typedef.
19629 * machmode.h (float_mode_for_size): New function.
19630 * emit-rtl.c (double_mode): Delete.
19631 (init_emit_once): Use float_mode_for_size.
19632 * stor-layout.c (layout_type): Likewise.
19633 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19636 Alan Hayward <alan.hayward@arm.com>
19637 David Sherwood <david.sherwood@arm.com>
19639 * output.h (assemble_real): Take a scalar_float_mode.
19640 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19641 * config/arm/arm.md (consttable_4): Likewise.
19642 (consttable_8): Likewise.
19643 (consttable_16): Likewise.
19644 * config/mips/mips.md (consttable_float): Likewise.
19645 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19646 * varasm.c (assemble_real): Take a scalar_float_mode.
19647 (output_constant_pool_2): Update accordingly.
19648 (output_constant): Likewise.
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 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19655 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19656 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19657 (native_encode_real): Likewise.
19658 (native_interpret_real): Likewise.
19659 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19660 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19663 Alan Hayward <alan.hayward@arm.com>
19664 David Sherwood <david.sherwood@arm.com>
19666 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19667 <scalar_float_mode>. Simplify.
19668 (gen_extend_conv_libfunc): Likewise.
19670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19671 Alan Hayward <alan.hayward@arm.com>
19672 David Sherwood <david.sherwood@arm.com>
19674 * coretypes.h (scalar_float_mode): New type.
19675 * machmode.h (mode_traits::from_int): Use machine_mode if
19676 USE_ENUM_MODES is defined.
19677 (is_a): New function.
19679 (dyn_cast): Likewise.
19680 (scalar_float_mode): New class.
19681 (scalar_float_mode::includes_p): New function.
19682 (is_float_mode): Likewise.
19683 * gdbhooks.py (MachineModePrinter): New class.
19684 (build_pretty_printer): Use it for scalar_float_mode.
19685 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19686 (format_helper::format_helper): Turn into a template.
19687 * genmodes.c (get_mode_class): New function.
19688 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19689 or machine_mode if none.
19690 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19691 as_a <scalar_float_mode>.
19692 * dwarf2out.c (mem_loc_descriptor): Likewise.
19693 (insert_float): Likewise.
19694 (add_const_value_attribute): Likewise.
19695 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19696 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19697 (expand_unop): Update accordingly.
19698 (expand_abs_nojump): Likewise.
19699 (expand_copysign_absneg): Take a scalar_float_mode.
19700 (expand_copysign_bit): Likewise.
19701 (expand_copysign): Update accordingly.
19703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19704 Alan Hayward <alan.hayward@arm.com>
19705 David Sherwood <david.sherwood@arm.com>
19707 * coretypes.h (opt_mode): New class.
19708 * machmode.h (opt_mode): Likewise.
19709 (opt_mode::else_void): New function.
19710 (opt_mode::require): Likewise.
19711 (opt_mode::exists): Likewise.
19712 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19713 (GET_MODE_2XWIDER_MODE): Likewise.
19714 (mode_iterator::get_wider): Update accordingly.
19715 (mode_iterator::get_2xwider): Likewise.
19716 (mode_iterator::get_known_wider): Likewise, turning into a template.
19717 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19718 forcing a wider mode to exist.
19719 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19720 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19721 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19722 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19723 * lower-subreg.c (init_lower_subreg): Likewise.
19724 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19725 on the final iteration.
19726 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19727 a wider mode exists before asking for a move pattern.
19728 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19729 forcing a wider mode to exist.
19730 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19731 returning false if no such mode exists.
19732 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19733 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19734 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19735 Avoid checking for a MODE_INT if we already know the mode is not a
19737 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19738 forcing a wider mode to exist.
19739 (expmed_mult_highpart_optab): Likewise.
19740 (expmed_mult_highpart): Likewise.
19741 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19743 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19744 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19745 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19746 * internal-fn.c (expand_mul_overflow): Update use of
19747 GET_MODE_2XWIDER_MODE.
19748 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19749 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19750 GET_MODE_WIDER_MODE.
19751 (convert_plusminus_to_widen): Likewise.
19752 * tree-switch-conversion.c (array_value_type): Likewise.
19753 * var-tracking.c (emit_note_insn_var_location): Likewise.
19754 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19755 Return false inside rather than outside the loop if no wider mode
19757 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19758 and GET_MODE_2XWIDER_MODE
19759 (can_compare_p): Use else_void.
19760 * gdbhooks.py (OptMachineModePrinter): New class.
19761 (build_pretty_printer): Use it for opt_mode.
19763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19764 Alan Hayward <alan.hayward@arm.com>
19765 David Sherwood <david.sherwood@arm.com>
19767 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19768 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19771 Alan Hayward <alan.hayward@arm.com>
19772 David Sherwood <david.sherwood@arm.com>
19774 * machmode.h (mode_traits): New structure.
19775 (get_narrowest_mode): New function.
19776 (mode_iterator::start): Likewise.
19777 (mode_iterator::iterate_p): Likewise.
19778 (mode_iterator::get_wider): Likewise.
19779 (mode_iterator::get_known_wider): Likewise.
19780 (mode_iterator::get_2xwider): Likewise.
19781 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19782 (FOR_EACH_MODE): Likewise.
19783 (FOR_EACH_MODE_FROM): Likewise.
19784 (FOR_EACH_MODE_UNTIL): Likewise.
19785 (FOR_EACH_WIDER_MODE): Likewise.
19786 (FOR_EACH_2XWIDER_MODE): Likewise.
19787 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19788 * combine.c (simplify_comparison): Likewise
19789 * config/i386/i386.c (type_natural_mode): Likewise.
19790 * cse.c (cse_insn): Likewise.
19791 * dse.c (find_shift_sequence): Likewise.
19792 * emit-rtl.c (init_derived_machine_modes): Likewise.
19793 (init_emit_once): Likewise.
19794 * explow.c (hard_function_value): Likewise.
19795 * expmed.c (extract_fixed_bit_field_1): Likewise.
19796 (extract_bit_field_1): Likewise.
19797 (expand_divmod): Likewise.
19798 (emit_store_flag_1): Likewise.
19799 * expr.c (init_expr_target): Likewise.
19800 (convert_move): Likewise.
19801 (alignment_for_piecewise_move): Likewise.
19802 (widest_int_mode_for_size): Likewise.
19803 (emit_block_move_via_movmem): Likewise.
19804 (copy_blkmode_to_reg): Likewise.
19805 (set_storage_via_setmem): Likewise.
19806 (compress_float_constant): Likewise.
19807 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19808 * optabs-query.c (get_best_extraction_insn): Likewise.
19809 * optabs.c (expand_binop): Likewise.
19810 (expand_twoval_unop): Likewise.
19811 (expand_twoval_binop): Likewise.
19812 (widen_leading): Likewise.
19813 (widen_bswap): Likewise.
19814 (expand_parity): Likewise.
19815 (expand_unop): Likewise.
19816 (prepare_cmp_insn): Likewise.
19817 (prepare_float_lib_cmp): Likewise.
19818 (expand_float): Likewise.
19819 (expand_fix): Likewise.
19820 (expand_sfix_optab): Likewise.
19821 * postreload.c (move2add_use_add2_insn): Likewise.
19822 * reg-stack.c (reg_to_stack): Likewise.
19823 * reginfo.c (choose_hard_reg_mode): Likewise.
19824 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19825 * stor-layout.c (mode_for_size): Likewise.
19826 (smallest_mode_for_size): Likewise.
19827 (mode_for_vector): Likewise.
19828 (finish_bitfield_representative): Likewise.
19829 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19830 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19831 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19832 * var-tracking.c (prepare_call_arguments): Likewise.
19834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19835 Alan Hayward <alan.hayward@arm.com>
19836 David Sherwood <david.sherwood@arm.com>
19838 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19839 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19840 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19841 * machmode.h (mode_size): Move earlier in file.
19842 (mode_precision): Likewise.
19843 (mode_inner): Likewise.
19844 (mode_nunits): Likewise.
19845 (mode_unit_size): Likewise.
19846 (unit_unit_precision): Likewise.
19847 (mode_wider): Likewise.
19848 (mode_2xwider): Likewise.
19849 (machine_mode): New class.
19850 (mode_to_bytes): New function.
19851 (mode_to_bits): Likewise.
19852 (mode_to_precision): Likewise.
19853 (mode_to_inner): Likewise.
19854 (mode_to_unit_size): Likewise.
19855 (mode_to_unit_precision): Likewise.
19856 (mode_to_nunits): Likewise.
19857 (GET_MODE_SIZE): Use mode_to_bytes.
19858 (GET_MODE_BITSIZE): Use mode_to_bits.
19859 (GET_MODE_PRECISION): Use mode_to_precision.
19860 (GET_MODE_INNER): Use mode_to_inner.
19861 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19862 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19863 (GET_MODE_NUNITS): Use mode_to_nunits.
19864 * system.h (ALWAYS_INLINE): New macro.
19865 * config/powerpcspe/powerpcspe-c.c
19866 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19867 int for arg1_mode and arg2_mode.
19869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19870 Alan Hayward <alan.hayward@arm.com>
19871 David Sherwood <david.sherwood@arm.com>
19873 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19874 Prefix mode names with E_ in case statements.
19875 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19876 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19877 (aarch64_split_simd_move): Likewise.
19878 (aarch64_gen_storewb_pair): Likewise.
19879 (aarch64_gen_loadwb_pair): Likewise.
19880 (aarch64_gen_store_pair): Likewise.
19881 (aarch64_gen_load_pair): Likewise.
19882 (aarch64_get_condition_code_1): Likewise.
19883 (aarch64_constant_pool_reload_icode): Likewise.
19884 (get_rsqrte_type): Likewise.
19885 (get_rsqrts_type): Likewise.
19886 (get_recpe_type): Likewise.
19887 (get_recps_type): Likewise.
19888 (aarch64_gimplify_va_arg_expr): Likewise.
19889 (aarch64_simd_container_mode): Likewise.
19890 (aarch64_emit_load_exclusive): Likewise.
19891 (aarch64_emit_store_exclusive): Likewise.
19892 (aarch64_expand_compare_and_swap): Likewise.
19893 (aarch64_gen_atomic_cas): Likewise.
19894 (aarch64_emit_bic): Likewise.
19895 (aarch64_emit_atomic_swap): Likewise.
19896 (aarch64_emit_atomic_load_op): Likewise.
19897 (aarch64_evpc_trn): Likewise.
19898 (aarch64_evpc_uzp): Likewise.
19899 (aarch64_evpc_zip): Likewise.
19900 (aarch64_evpc_ext): Likewise.
19901 (aarch64_evpc_rev): Likewise.
19902 (aarch64_evpc_dup): Likewise.
19903 (aarch64_gen_ccmp_first): Likewise.
19904 (aarch64_gen_ccmp_next): Likewise.
19905 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19906 (alpha_emit_xfloating_libcall): Likewise.
19907 (emit_insxl): Likewise.
19908 (alpha_arg_type): Likewise.
19909 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19910 (arc_preferred_simd_mode): Likewise.
19911 (arc_secondary_reload): Likewise.
19912 (get_arc_condition_code): Likewise.
19913 (arc_print_operand): Likewise.
19914 (arc_legitimate_constant_p): Likewise.
19915 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19916 * config/arc/arc.md (casesi_load): Likewise.
19917 (casesi_compact_jump): Likewise.
19918 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19919 (cc_use_register): Likewise.
19920 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19921 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19922 (arm_init_iwmmxt_builtins): Likewise.
19923 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19924 (neon_expand_vector_init): Likewise.
19925 (arm_attr_length_move_neon): Likewise.
19926 (maybe_get_arm_condition_code): Likewise.
19927 (arm_emit_vector_const): Likewise.
19928 (arm_preferred_simd_mode): Likewise.
19929 (arm_output_iwmmxt_tinsr): Likewise.
19930 (thumb1_output_casesi): Likewise.
19931 (thumb2_output_casesi): Likewise.
19932 (arm_emit_load_exclusive): Likewise.
19933 (arm_emit_store_exclusive): Likewise.
19934 (arm_expand_compare_and_swap): Likewise.
19935 (arm_evpc_neon_vuzp): Likewise.
19936 (arm_evpc_neon_vzip): Likewise.
19937 (arm_evpc_neon_vrev): Likewise.
19938 (arm_evpc_neon_vtrn): Likewise.
19939 (arm_evpc_neon_vext): Likewise.
19940 (arm_validize_comparison): Likewise.
19941 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19942 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19943 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19944 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19945 (c6x_preferred_simd_mode): Likewise.
19946 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19947 (epiphany_rtx_costs): Likewise.
19948 * config/epiphany/predicates.md (proper_comparison_operator):
19950 * config/frv/frv.c (condexec_memory_operand): Likewise.
19951 (frv_emit_move): Likewise.
19952 (output_move_single): Likewise.
19953 (output_condmove_single): Likewise.
19954 (frv_hard_regno_mode_ok): Likewise.
19955 (frv_matching_accg_mode): Likewise.
19956 * config/h8300/h8300.c (split_adds_subs): Likewise.
19957 (h8300_rtx_costs): Likewise.
19958 (h8300_print_operand): Likewise.
19959 (compute_mov_length): Likewise.
19960 (output_logical_op): Likewise.
19961 (compute_logical_op_length): Likewise.
19962 (compute_logical_op_cc): Likewise.
19963 (h8300_shift_needs_scratch_p): Likewise.
19964 (output_a_shift): Likewise.
19965 (compute_a_shift_length): Likewise.
19966 (compute_a_shift_cc): Likewise.
19967 (expand_a_rotate): Likewise.
19968 (output_a_rotate): Likewise.
19969 * config/i386/i386.c (classify_argument): Likewise.
19970 (function_arg_advance_32): Likewise.
19971 (function_arg_32): Likewise.
19972 (function_arg_64): Likewise.
19973 (function_value_64): Likewise.
19974 (ix86_gimplify_va_arg): Likewise.
19975 (ix86_legitimate_constant_p): Likewise.
19976 (put_condition_code): Likewise.
19977 (split_double_mode): Likewise.
19978 (ix86_avx256_split_vector_move_misalign): Likewise.
19979 (ix86_expand_vector_logical_operator): Likewise.
19980 (ix86_split_idivmod): Likewise.
19981 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19982 (ix86_build_const_vector): Likewise.
19983 (ix86_build_signbit_mask): Likewise.
19984 (ix86_match_ccmode): Likewise.
19985 (ix86_cc_modes_compatible): Likewise.
19986 (ix86_expand_branch): Likewise.
19987 (ix86_expand_sse_cmp): Likewise.
19988 (ix86_expand_sse_movcc): Likewise.
19989 (ix86_expand_int_sse_cmp): Likewise.
19990 (ix86_expand_vec_perm_vpermi2): Likewise.
19991 (ix86_expand_vec_perm): Likewise.
19992 (ix86_expand_sse_unpack): Likewise.
19993 (ix86_expand_int_addcc): Likewise.
19994 (ix86_split_to_parts): Likewise.
19995 (ix86_vectorize_builtin_gather): Likewise.
19996 (ix86_vectorize_builtin_scatter): Likewise.
19997 (avx_vpermilp_parallel): Likewise.
19998 (inline_memory_move_cost): Likewise.
19999 (ix86_tieable_integer_mode_p): Likewise.
20000 (x86_maybe_negate_const_int): Likewise.
20001 (ix86_expand_vector_init_duplicate): Likewise.
20002 (ix86_expand_vector_init_one_nonzero): Likewise.
20003 (ix86_expand_vector_init_one_var): Likewise.
20004 (ix86_expand_vector_init_concat): Likewise.
20005 (ix86_expand_vector_init_interleave): Likewise.
20006 (ix86_expand_vector_init_general): Likewise.
20007 (ix86_expand_vector_set): Likewise.
20008 (ix86_expand_vector_extract): Likewise.
20009 (emit_reduc_half): Likewise.
20010 (ix86_emit_i387_round): Likewise.
20011 (ix86_mangle_type): Likewise.
20012 (ix86_expand_round_sse4): Likewise.
20013 (expand_vec_perm_blend): Likewise.
20014 (canonicalize_vector_int_perm): Likewise.
20015 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20016 (expand_vec_perm_1): Likewise.
20017 (expand_vec_perm_interleave3): Likewise.
20018 (expand_vec_perm_even_odd_pack): Likewise.
20019 (expand_vec_perm_even_odd_1): Likewise.
20020 (expand_vec_perm_broadcast_1): Likewise.
20021 (ix86_vectorize_vec_perm_const_ok): Likewise.
20022 (ix86_expand_vecop_qihi): Likewise.
20023 (ix86_expand_mul_widen_hilo): Likewise.
20024 (ix86_expand_sse2_abs): Likewise.
20025 (ix86_expand_pextr): Likewise.
20026 (ix86_expand_pinsr): Likewise.
20027 (ix86_preferred_simd_mode): Likewise.
20028 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20029 * config/i386/sse.md (*andnot<mode>3): Likewise.
20030 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20031 (*<code><mode>3): Likewise.
20032 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20033 (ia64_expand_atomic_op): Likewise.
20034 (ia64_arg_type): Likewise.
20035 (ia64_mode_to_int): Likewise.
20036 (ia64_scalar_mode_supported_p): Likewise.
20037 (ia64_vector_mode_supported_p): Likewise.
20038 (expand_vec_perm_broadcast): Likewise.
20039 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20040 (iq2000_function_arg_advance): Likewise.
20041 (iq2000_function_arg): Likewise.
20042 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20043 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20044 (m68k_libcall_value): Likewise.
20045 (m68k_function_value): Likewise.
20046 (sched_attr_op_type): Likewise.
20047 * config/mcore/mcore.c (mcore_output_move): Likewise.
20048 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20050 (microblaze_function_arg): Likewise.
20051 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20052 (mips_print_operand): Likewise.
20053 (mips_mode_ok_for_mov_fmt_p): Likewise.
20054 (mips_vector_mode_supported_p): Likewise.
20055 (mips_preferred_simd_mode): Likewise.
20056 (mips_expand_vpc_loongson_even_odd): Likewise.
20057 (mips_expand_vec_unpack): Likewise.
20058 (mips_expand_vi_broadcast): Likewise.
20059 (mips_expand_vector_init): Likewise.
20060 (mips_expand_vec_reduc): Likewise.
20061 (mips_expand_msa_cmp): Likewise.
20062 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20063 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20064 (cc_flags_for_mode): Likewise.
20065 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20066 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20067 (nds32_output_casesi_pc_relative): Likewise.
20068 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20069 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20070 (nvptx_gen_unpack): Likewise.
20071 (nvptx_gen_pack): Likewise.
20072 (nvptx_gen_shuffle): Likewise.
20073 (nvptx_gen_wcast): Likewise.
20074 (nvptx_preferred_simd_mode): Likewise.
20075 * config/pa/pa.c (pa_secondary_reload): Likewise.
20076 * config/pa/predicates.md (base14_operand): Likewise.
20077 * config/powerpcspe/powerpcspe-c.c
20078 (altivec_resolve_overloaded_builtin): Likewise.
20079 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20081 (rs6000_preferred_simd_mode): Likewise.
20082 (output_vec_const_move): Likewise.
20083 (rs6000_expand_vector_extract): Likewise.
20084 (rs6000_split_vec_extract_var): Likewise.
20085 (reg_offset_addressing_ok_p): Likewise.
20086 (rs6000_legitimate_offset_address_p): Likewise.
20087 (rs6000_legitimize_address): Likewise.
20088 (rs6000_emit_set_const): Likewise.
20089 (rs6000_const_vec): Likewise.
20090 (rs6000_emit_move): Likewise.
20091 (spe_build_register_parallel): Likewise.
20092 (rs6000_darwin64_record_arg_recurse): Likewise.
20093 (swap_selector_for_mode): Likewise.
20094 (spe_init_builtins): Likewise.
20095 (paired_init_builtins): Likewise.
20096 (altivec_init_builtins): Likewise.
20097 (do_load_for_compare): Likewise.
20098 (rs6000_generate_compare): Likewise.
20099 (rs6000_expand_float128_convert): Likewise.
20100 (emit_load_locked): Likewise.
20101 (emit_store_conditional): Likewise.
20102 (rs6000_output_function_epilogue): Likewise.
20103 (rs6000_handle_altivec_attribute): Likewise.
20104 (rs6000_function_value): Likewise.
20105 (emit_fusion_gpr_load): Likewise.
20106 (emit_fusion_p9_load): Likewise.
20107 (emit_fusion_p9_store): Likewise.
20108 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20109 (fusion_gpr_mem_load): Likewise.
20110 (fusion_addis_mem_combo_load): Likewise.
20111 (fusion_addis_mem_combo_store): Likewise.
20112 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20113 (fusion_gpr_mem_load): Likewise.
20114 (fusion_addis_mem_combo_load): Likewise.
20115 (fusion_addis_mem_combo_store): Likewise.
20116 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20118 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20119 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20120 (rs6000_preferred_simd_mode): Likewise.
20121 (output_vec_const_move): Likewise.
20122 (rs6000_expand_vector_extract): Likewise.
20123 (rs6000_split_vec_extract_var): Likewise.
20124 (reg_offset_addressing_ok_p): Likewise.
20125 (rs6000_legitimate_offset_address_p): Likewise.
20126 (rs6000_legitimize_address): Likewise.
20127 (rs6000_emit_set_const): Likewise.
20128 (rs6000_const_vec): Likewise.
20129 (rs6000_emit_move): Likewise.
20130 (rs6000_darwin64_record_arg_recurse): Likewise.
20131 (swap_selector_for_mode): Likewise.
20132 (paired_init_builtins): Likewise.
20133 (altivec_init_builtins): Likewise.
20134 (rs6000_expand_float128_convert): Likewise.
20135 (emit_load_locked): Likewise.
20136 (emit_store_conditional): Likewise.
20137 (rs6000_output_function_epilogue): Likewise.
20138 (rs6000_handle_altivec_attribute): Likewise.
20139 (rs6000_function_value): Likewise.
20140 (emit_fusion_gpr_load): Likewise.
20141 (emit_fusion_p9_load): Likewise.
20142 (emit_fusion_p9_store): Likewise.
20143 * config/rx/rx.c (rx_gen_move_template): Likewise.
20144 (flags_from_mode): Likewise.
20145 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20146 (s390_slb_comparison): Likewise.
20147 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20148 (s390_vector_mode_supported_p): Likewise.
20149 (s390_cc_modes_compatible): Likewise.
20150 (s390_match_ccmode_set): Likewise.
20151 (s390_canonicalize_comparison): Likewise.
20152 (s390_emit_compare_and_swap): Likewise.
20153 (s390_branch_condition_mask): Likewise.
20154 (s390_rtx_costs): Likewise.
20155 (s390_secondary_reload): Likewise.
20156 (__SECONDARY_RELOAD_CASE): Likewise.
20157 (s390_expand_cs): Likewise.
20158 (s390_preferred_simd_mode): Likewise.
20159 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20160 * config/sh/sh.c (sh_print_operand): Likewise.
20161 (dump_table): Likewise.
20162 (sh_secondary_reload): Likewise.
20163 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20164 * config/sh/sh.md (casesi_worker_1): Likewise.
20165 (casesi_worker_2): Likewise.
20166 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20167 (fcc_comparison_operator): Likewise.
20168 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20169 (emit_soft_tfmode_cvt): Likewise.
20170 (sparc_preferred_simd_mode): Likewise.
20171 (output_cbranch): Likewise.
20172 (sparc_print_operand): Likewise.
20173 (sparc_expand_vec_perm_bmask): Likewise.
20174 (vector_init_bshuffle): Likewise.
20175 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20176 (spu_vector_mode_supported_p): Likewise.
20177 (spu_expand_insv): Likewise.
20178 (spu_emit_branch_or_set): Likewise.
20179 (spu_handle_vector_attribute): Likewise.
20180 (spu_builtin_splats): Likewise.
20181 (spu_builtin_extract): Likewise.
20182 (spu_builtin_promote): Likewise.
20183 (spu_expand_sign_extend): Likewise.
20184 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20185 (tilegx_simd_int): Likewise.
20186 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20187 (tilepro_simd_int): Likewise.
20188 * config/v850/v850.c (const_double_split): Likewise.
20189 (v850_print_operand): Likewise.
20190 (ep_memory_offset): Likewise.
20191 * config/vax/vax.c (vax_rtx_costs): Likewise.
20192 (vax_output_int_move): Likewise.
20193 (vax_output_int_add): Likewise.
20194 (vax_output_int_subtract): Likewise.
20195 * config/visium/predicates.md (visium_branch_operator): Likewise.
20196 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20197 (visium_print_operand_address): Likewise.
20198 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20199 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20200 (xtensa_expand_conditional_branch): Likewise.
20201 (xtensa_copy_incoming_a7): Likewise.
20202 (xtensa_output_literal): Likewise.
20203 * dfp.c (decimal_real_maxval): Likewise.
20204 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20207 Alan Hayward <alan.hayward@arm.com>
20208 David Sherwood <david.sherwood@arm.com>
20210 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20211 (mode_nunits_inline): Likewise.
20212 (mode_inner_inline): Likewise.
20213 (mode_unit_size_inline): Likewise.
20214 (mode_unit_precision_inline): Likewise.
20215 (emit_insn_modes_h): Likewise. Also emit a #define of the
20217 (emit_mode_wider): Add an E_ prefix to mode names.
20218 (emit_mode_complex): Likewise.
20219 (emit_mode_inner): Likewise.
20220 (emit_mode_adjustments): Likewise.
20221 (emit_mode_int_n): Likewise.
20222 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20223 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20224 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20225 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20226 (CRC32_BUILTIN, ENTRY): Likewise.
20227 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20228 (aarch64_pop_regs): Likewise.
20229 (aarch64_process_components): Likewise.
20230 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20231 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20232 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20233 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20234 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20235 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20237 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20238 (pcmpistr): Likewise.
20239 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20240 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20241 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20243 * config/rl78/rl78.c (mduc_regs): Likewise.
20244 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20245 (htm_expand_builtin): Likewise.
20246 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20247 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20248 * config/xtensa/xtensa.c (print_operand): Likewise.
20249 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20250 (NUM_MODE_VECTOR_INT): Likewise.
20251 * genoutput.c (null_operand): Likewise.
20252 (output_operand_data): Likewise.
20253 * genrecog.c (print_parameter_value): Likewise.
20254 * lra.c (debug_operand_data): Likewise.
20256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20258 * dwarf2out.c (scompare_loc_descriptor_wide)
20259 (scompare_loc_descriptor_narrow): New functions, split out from...
20260 (scompare_loc_descriptor): ...here.
20261 * expmed.c (emit_store_flag_int): New function, split out from...
20262 (emit_store_flag): ...here.
20264 2017-08-30 Richard Biener <rguenther@suse.de>
20266 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20267 (dwarf2out_early_finish): Move setting of AT_pubnames from
20268 early debug output to early finish.
20270 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20272 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20273 and -mdata-region to the assembler.
20275 2017-08-30 Richard Biener <rguenther@suse.de>
20277 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20279 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20280 (dwarf2out_early_global_decl): Only generate a DIE for the
20281 abstract origin if it doesn't already exist or is a declaration DIE.
20282 (resolve_addr): Do not add the linkage name twice when
20283 generating a stub DIE for the DW_TAG_GNU_call_site target.
20285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20287 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20288 Use machine_mode rather than int for arg1_mode.
20290 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20293 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20294 that the second argument of the built-in functions to unpack
20295 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20296 switch statement instead a lot of if statements.
20297 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20298 Allow 64-bit values to be in Altivec registers as well as
20299 traditional floating point registers.
20300 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20302 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20304 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20305 MAX_REGS_PER_ADDRESS == 1.
20307 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20309 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20310 * config/i386/i386.c (ix86_option_override_internal): Simplify
20311 setting of opts->x_flag_entry.
20313 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20314 Jakub Jelinek <jakub@redhat.com>
20315 Richard Biener <rguenther@suse.de>
20317 PR tree-optimization/81503
20318 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20319 folded constant fits in the target type; reorder tests for clarity.
20321 2017-08-29 Martin Liska <mliska@suse.cz>
20323 * passes.def: Include pass_lower_switch.
20324 * stmt.c (dump_case_nodes): Remove and move to
20325 tree-switch-conversion.
20326 (case_values_threshold): Likewise.
20327 (expand_switch_as_decision_tree_p): Likewise.
20328 (emit_case_decision_tree): Likewise.
20329 (expand_case): Likewise.
20330 (balance_case_nodes): Likewise.
20331 (node_has_low_bound): Likewise.
20332 (node_has_high_bound): Likewise.
20333 (node_is_bounded): Likewise.
20334 (emit_case_nodes): Likewise.
20335 (struct simple_case_node): New struct.
20336 (add_case_node): Remove.
20337 (emit_case_dispatch_table): Use vector instead of case_list.
20338 (reset_out_edges_aux): Remove.
20339 (compute_cases_per_edge): Likewise.
20340 (expand_case): Build list of simple_case_node.
20341 (expand_sjlj_dispatch_table): Use it.
20342 * tree-switch-conversion.c (struct case_node): Moved from
20343 stmt.c and adjusted.
20344 (emit_case_nodes): Likewise.
20345 (node_has_low_bound): Likewise.
20346 (node_has_high_bound): Likewise.
20347 (node_is_bounded): Likewise.
20348 (case_values_threshold): Likewise.
20349 (reset_out_edges_aux): Likewise.
20350 (compute_cases_per_edge): Likewise.
20351 (add_case_node): Likewise.
20352 (dump_case_nodes): Likewise.
20353 (balance_case_nodes): Likewise.
20354 (expand_switch_as_decision_tree_p): Likewise.
20355 (emit_jump): Likewise.
20356 (emit_case_decision_tree): Likewise.
20357 (try_switch_expansion): Likewise.
20358 (do_jump_if_equal): Likewise.
20359 (emit_cmp_and_jump_insns): Likewise.
20360 (fix_phi_operands_for_edge): New function.
20361 (record_phi_operand_mapping): Likewise.
20362 (class pass_lower_switch): New pass.
20363 (pass_lower_switch::execute): New function.
20364 (make_pass_lower_switch): Likewise.
20365 (conditional_probability):
20366 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20367 * tree-pass.h: Add make_pass_lower_switch.
20369 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20372 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20375 2017-08-29 Richard Biener <rguenther@suse.de>
20377 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20378 we do not add a DW_AT_inline attribute twice.
20379 (gen_subprogram_die): Remove code setting DW_AT_inline on
20380 DECL_ABSTRACT_P nodes.
20382 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20384 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20385 calls to internal functions.
20386 (gimplify_modify_expr): Likewise.
20387 * tree-call-cdce.c (use_internal_fn): Likewise.
20388 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20389 (convert_to_divmod): Set the nothrow flag.
20390 * tree-if-conv.c (predicate_mem_writes): Likewise.
20391 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20392 (vectorizable_call): Likewise.
20393 (vectorizable_store): Likewise.
20394 (vectorizable_load): Likewise.
20395 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20396 (vect_recog_mask_conversion_pattern): Likewise.
20398 2017-08-29 Martin Liska <mliska@suse.cz>
20401 * gcc.c (driver_handle_option): Add new argument.
20402 * opts-common.c (handle_option): Pass
20403 target_option_override_hook.
20404 * opts-global.c (lang_handle_option): Add new option.
20405 (set_default_handlers): Add new argument.
20406 (decode_options): Likewise.
20407 * opts.c (target_handle_option): Likewise.
20408 (common_handle_option): Call target_option_override_hook.
20409 * opts.h (struct cl_option_handler_func): Add hook for
20410 target option override.
20411 (struct cl_option_handlers): Likewise.
20412 (set_default_handlers): Add new argument.
20413 (decode_options): Likewise.
20414 (common_handle_option): Likewise.
20415 (target_handle_option): Likewise.
20416 * toplev.c (toplev::main): Pass targetm.target_option.override
20419 2017-08-29 Richard Biener <rguenther@suse.de>
20420 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20422 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20423 life to the active subtree.
20425 2017-08-28 Jeff Law <law@redhat.com>
20427 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20428 derive_equivalences.
20429 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20430 Code moved into....
20431 (edge_info::derive_equivalences): New private member function
20433 * tree-ssa-dom.c (class edge_info): Changed from a struct
20434 to a class. Add ctor/dtor, methods and data members.
20435 (edge_info::edge_info): Renamed from allocate_edge_info.
20436 Initialize additional members.
20437 (edge_info::~edge_info): New.
20438 (free_dom_edge_info): Delete the edge info.
20439 (record_edge_info): Use new class & associated member functions.
20440 Tighten forms for testing for edge equivalences.
20441 (record_temporary_equivalences): Iterate over the simple
20442 equivalences rather than assuming there's only one per edge.
20443 (cprop_into_successor_phis): Iterate over the simple
20444 equivalences rather than assuming there's only one per edge.
20445 (optimize_stmt): Use operand_equal_p rather than pointer
20446 equality for mini-DSE code.
20448 2017-08-28 Nathan Sidwell <nathan@acm.org>
20450 * gcc.c (execute): Fold SIGPIPE handling into switch
20451 statement. Adjust internal error message.
20453 2017-08-28 Richard Biener <rguenther@suse.de>
20456 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20457 Do nothing for removed DIEs.
20459 2017-08-28 Richard Biener <rguenther@suse.de>
20461 PR tree-optimization/81977
20462 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20465 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20468 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20469 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20470 using targetm.gen_mem_thread_fence.
20472 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20475 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20476 predicate to register_operand. Reorder operands.
20477 (*btr<mode>): Ditto.
20478 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20479 (*btr<mode>_mask): Ditto.
20481 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20483 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20484 * config/rs6000/xmmintrin.h: New file.
20485 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20487 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20490 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20491 parameter and_insn and return it.
20492 (recombine_lvx_pattern): Insert a copy to ensure availability of
20493 the base register of the copied masking operation at the point of
20494 the instruction replacement.
20495 (recombine_stvx_pattern): Likewise.
20497 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20499 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20500 undocumented switches.
20501 (-mpower9-dform-vector): Likewise.
20502 (-mpower9-dform): Likewise.
20503 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20504 comments to delete references to -mpower9-dform* switches.
20505 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20506 Delete reference to -mpower9-dform* switches, test for
20507 -mpower9-vector instead.
20508 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20509 (OTHER_P9_VECTOR_MASKS): Likewise.
20510 (POWERPC_MASKS): Likewise.
20511 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20512 tests against -mpower9-dform* to -mpower9-vector. Delete code
20513 that checked for -mpower9-dform* consistancy with other options.
20514 Add test for -mpower9-misc to enable other power9 switches.
20515 (rs6000_init_hard_regno_mode_ok): Likewise.
20516 (rs6000_option_override_internal): Likewise.
20517 (rs6000_emit_prologue): Likewise.
20518 (rs6000_emit_epilogue): Likewise.
20519 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20520 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20521 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20523 (emit_fusion_p9_store): Likewise.
20524 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20525 resetting these macros if the assembler does not support ISA 3.0
20527 (TARGET_P9_DFORM_VECTOR): Likewise.
20528 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20529 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20531 2017-08-25 Alan Modra <amodra@gmail.com>
20533 PR rtl-optimization/81747
20534 * cse.c (cse_extended_basic_block): Don't attempt to record
20535 equivalences for degenerate conditional branches that branch
20536 to their fall-through.
20538 2017-08-24 Martin Sebor <msebor@redhat.com>
20540 PR middle-end/81908
20541 * gimple-fold.c (size_must_be_zero_p): New function.
20542 (gimple_fold_builtin_memory_op): Call it.
20544 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20546 * config/rs6000/mm_malloc.h: New file.
20548 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20550 PR tree-optimization/81913
20551 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20552 analysis when either IVs in condition can wrap.
20554 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20556 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20557 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20559 2017-08-24 Richard Biener <rguenther@suse.de>
20562 * targhooks.c (default_target_can_inline_p): Properly
20563 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20564 is present and always compare.
20565 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20566 infer -mfpmath=sse from TARGET_SSE_P.
20567 (ix86_can_inline_p): Properly use target_option_default_node when
20568 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20570 2017-08-24 Richard Biener <rguenther@suse.de>
20573 * dwarf2out.c (output_die): Handle flag_generate_offload like
20575 (output_comp_unit): Likewise.
20576 (gen_array_type_die): Likewise.
20577 (dwarf2out_early_finish): Likewise.
20578 (note_variable_value_in_expr): Likewise.
20579 (dwarf2out_finish): Likewise. Adjust assert.
20580 * cgraphunit.c (symbol_table::compile): Move setting of
20581 flag_generate_offload earlier ...
20582 (symbol_table::finalize_compilation_unit): ... here, before
20583 early debug finalization.
20585 2017-08-24 Richard Biener <rguenther@suse.de>
20587 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20588 and ipa-fnsummary.h.
20589 (ix86_can_inline_p): When ix86_fpmath flags do not match
20590 check whether the callee uses FP math at all.
20592 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20594 PR middle-end/81931
20595 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20596 instead of TYPE_PRECISION.
20598 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20599 Alan Hayward <alan.hayward@arm.com>
20600 David Sherwood <david.sherwood@arm.com>
20602 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20604 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20606 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20608 (ix86_frame::stack_realign_allocate): New field.
20609 (struct machine_frame_state): Modify comments.
20610 (machine_frame_state::sp_realigned_fp_end): New field.
20611 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20612 layout calculation.
20613 (sp_valid_at): Add assertion to assure no attempt to access invalid
20614 offset of a realigned stack.
20615 (fp_valid_at): Likewise.
20616 (choose_baseaddr): Modify comments.
20617 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20618 ix86_expand_prologue.
20619 (ix86_expand_prologue): Modify stack realignment and allocation.
20620 (ix86_expand_epilogue): Modify comments.
20621 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20622 avx2_runtime, avx512f, and avx512f_runtime.
20624 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20626 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20627 (mstackrealign): Do not init to -1.
20628 * config/i386/i386.c (ix86_option_override_internal):
20629 Check opts_set, not opts when setting default value of
20630 opts->x_ix86_force_align_arg_pointer.
20632 2017-08-23 Richard Biener <rguenther@suse.de>
20634 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20635 lang_hooks.decl_printable_name.
20636 * print-rtl-function.c (print_rtx_function): Likewise.
20637 * tree-pretty-print.c (dump_function_header): Likewise.
20639 2017-08-23 Richard Biener <rguenther@suse.de>
20642 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20643 -g0 at compile-time.
20645 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20647 PR middle-end/19706
20648 * doc/sourcebuild.texi (Other hardware attributes):
20651 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20653 PR middle-end/19706
20654 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20655 Fix single-use check.
20657 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20659 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20661 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20663 * doc/install.texi: Modify to add more details on running selected
20666 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20668 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20669 is combined with -mabi=ms.
20670 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20673 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20675 PR tree-optimization/81488
20676 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20677 and cached_basis fields.
20678 (MAX_SPREAD): New constant.
20679 (alloc_cand_and_find_basis): Initialize new fields.
20680 (clear_visited): New function.
20681 (create_phi_basis_1): Rename from create_phi_basis, set visited
20682 and cached_basis fields.
20683 (create_phi_basis): New wrapper function.
20684 (phi_add_costs_1): Rename from phi_add_costs, add spread
20685 parameter, set visited field, short-circuit when limits reached.
20686 (phi_add_costs): New wrapper function.
20687 (record_phi_increments_1): Rename from record_phi_increments, set
20689 (record_phi_increments): New wrapper function.
20690 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20691 (phi_incr_cost): New wrapper function.
20692 (all_phi_incrs_profitable_1): Rename from
20693 all_phi_incrs_profitable, set visited field.
20694 (all_phi_incrs_profitable): New wrapper function.
20696 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20697 Alan Hayward <alan.hayward@arm.com>
20698 David Sherwood <david.sherwood@arm.com>
20700 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20701 that takes the outer and inner modes.
20702 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20703 comparison as the canonical test for a paradoxical subreg.
20704 * combine.c (simplify_set): Use paradoxical_subreg_p.
20705 (make_extraction): Likewise.
20706 (force_to_mode): Likewise.
20707 (rtx_equal_for_field_assignment_p): Likewise.
20708 (gen_lowpart_for_combine): Likewise.
20709 (simplify_comparison): Likewise.
20710 * cse.c (equiv_constant): Likewise.
20711 * expmed.c (store_bit_field_1): Likewise.
20712 * final.c (alter_subreg): Likewise.
20713 * fwprop.c (propagate_rtx): Likewise.
20714 (forward_propagate_subreg): Likewise.
20715 * ira-conflicts.c (ira_build_conflicts): Likewise.
20716 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20717 * lra-constraints.c (curr_insn_transform): Likewise.
20718 (split_reg): Likewise.
20719 * lra-eliminations.c (move_plus_up): Likewise.
20720 (lra_eliminate_regs_1): Likewise.
20721 * recog.c (general_operand): Likewise.
20722 * ree.c (combine_reaching_defs): Likewise.
20723 * reload.c (push_reload): Likewise.
20724 (find_reloads): Likewise.
20725 * reload1.c (elimination_effects): Likewise.
20726 (compute_reload_subreg_offset): Likewise.
20727 (choose_reload_regs): Likewise.
20728 * rtlanal.c (subreg_lsb_1): Likewise.
20729 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20730 (simplify_subreg): Likewise.
20731 * var-tracking.c (track_loc_p): Likewise.
20732 * emit-rtl.c (byte_lowpart_offset): Likewise.
20733 (paradoxical_subreg_p): Delete out-of-line definition.
20735 2017-08-22 Jeff Law <law@redhat.com>
20737 PR tree-optimization/81741
20738 PR tree-optimization/71947
20739 * tree-ssa-dom.c: Include tree-inline.h.
20740 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20741 equivalences if one is more expensive to compute than the other.
20742 * tree-ssa-scopedtables.h (class const_or_copies): Make
20743 record_const_or_copy_raw method private.
20744 (class avail_exprs_stack): New method simplify_binary_operation.
20745 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20746 avail_exprs_stack::simplify_binary_operation as needed.
20747 (avail_exprs_stack::simplify_binary_operation): New function.
20749 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20751 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20752 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20753 (DOT_SYMBOLS): Likewise.
20754 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20755 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20756 (RS6000_ABI_NAME): Likewise.
20757 (TARGET_CMODEL): Likewise.
20758 (TOC_SECTION_ASM_OP): Likewise.
20759 (SET_CMODEL): New macro.
20760 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20762 2017-08-22 Richard Biener <rguenther@suse.de>
20764 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20765 to allow for free-lang-data replacements similar to verify_type_variant.
20767 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20768 Alan Hayward <alan.hayward@arm.com>
20769 David Sherwood <david.sherwood@arm.com>
20771 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20772 VOIDmode for the LABEL_REF.
20774 2017-08-22 Richard Biener <rguenther@suse.de>
20776 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20778 2017-08-22 Richard Biener <rguenther@suse.de>
20780 * common.opt (feliminate-dwarf2-dups): Ignore.
20781 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20782 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20783 same_die_p_wrap, compute_section_prefix,
20784 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20785 (comdat_symbol_id, comdat_symbol_number): Likewise.
20786 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20788 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20789 (output_die): Mark unreachable path unreachable.
20790 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20791 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20792 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20793 (dwarf2out_early_finish): Likewise.
20795 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20797 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20799 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20802 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20803 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20804 (avr_attribute_table) <io, io_low, address>: Initialize
20805 .decl_required with true.
20807 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20809 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20810 undocumented debugging options.
20811 (-mvsx-scalar-double): Likewise.
20812 (-mallow-df-permute): Likewise.
20813 (-mvectorize-builtins): Likewise.
20814 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20815 (rs6000_builtin_vectorized_function): Likewise.
20816 (rs6000_builtin_md_vectorized_function): Likewise.
20817 (rs6000_opt_vars): Likewise.
20819 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20822 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20823 (*btrq_imm): Rename from *btrq.
20824 (*btcq_imm): Rename from *btcq.
20825 (btsc): New code attribute.
20826 (*<btsc><mode>): New insn pattern.
20827 (*btr<mode>): Ditto.
20828 (*<btsc><mode>_mask): New insn_and_split pattern.
20829 (*btr<mode>_mask): Ditto.
20831 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20832 Alan Hayward <alan.hayward@arm.com>
20833 David Sherwood <david.sherwood@arm.com>
20835 * function.c (pad_below): Simplify padding calculation.
20837 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20838 Alan Hayward <alan.hayward@arm.com>
20839 David Sherwood <david.sherwood@arm.com>
20841 * target.def (function_prologue): Remove frame size argument.
20842 (function_epilogue): Likewise.
20843 * doc/tm.texi: Regenerate.
20844 * final.c (final_start_function): Update call to function_prologue.
20845 (final_end_function): Update call to function_epilogue.
20846 (default_function_pro_epilogue): Remove frame size argument.
20847 * output.h (default_function_pro_epilogue): Likewise.
20848 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20849 (arm_output_function_prologue): Likewise.
20850 * config/frv/frv.c (frv_function_prologue): Likewise.
20851 (frv_function_epilogue): Likewise.
20852 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20853 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20854 (ia64_output_function_epilogue): Likewise.
20855 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20856 (m32r_output_function_epilogue): Likewise.
20857 * config/microblaze/microblaze.c (microblaze_function_prologue)
20858 (microblaze_function_epilogue): Likewise.
20859 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20860 (mips_output_function_epilogue): Likewise.
20861 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20862 (mmix_target_asm_function_epilogue): Likewise.
20863 * config/msp430/msp430.c (msp430_start_function): Likewise.
20864 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20865 (nds32_asm_function_epilogue): Likewise.
20866 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20867 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20868 (pa_output_function_epilogue): Likewise.
20869 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20870 (rs6000_output_function_epilogue): Likewise.
20871 * config/rl78/rl78.c (rl78_start_function): Likewise.
20872 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20873 (rs6000_output_function_epilogue): Likewise.
20874 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20875 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20876 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20877 (sparc_asm_function_epilogue): Likewise.
20879 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20881 * tree.h (type_has_mode_precision_p): New function.
20882 * convert.c (convert_to_integer_1): Use it.
20883 * expr.c (expand_expr_real_2): Likewise.
20884 (expand_expr_real_1): Likewise.
20885 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20886 * match.pd: Likewise.
20887 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20888 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20889 * tree-tailcall.c (process_assignment): Likewise.
20890 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20891 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20892 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20893 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20894 (vectorizable_assignment): Likewise.
20895 (vectorizable_shift): Likewise.
20896 (vectorizable_operation): Likewise.
20897 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20899 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20901 * match.pd: Add pow (C, x) simplification.
20903 2017-08-21 Richard Biener <rguenther@suse.de>
20905 PR tree-optimization/81900
20906 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20907 for blocks with abnormal predecessors.
20908 (compute_antic): Do not set visited flag prematurely.
20910 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20913 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20915 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20917 * stor-layout.h (vector_type_mode): Move to...
20918 * tree.h (vector_type_mode): ...here.
20919 * stor-layout.c (vector_type_mode): Move to...
20920 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20922 2017-08-21 Richard Biener <rguenther@suse.de>
20924 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20925 register_external_die hooks.
20926 (debug_false_tree_charstarstar_uhwistar): Declare.
20927 (debug_nothing_tree_charstar_uhwi): Likewise.
20928 * debug.c (do_nothing_debug_hooks): Adjust.
20929 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20930 (debug_nothing_tree_charstar_uhwi): Likewise.
20931 * dbxout.c (dbx_debug_hooks): Adjust.
20932 (xcoff_debug_hooks): Likewise.
20933 * sdbout.c (sdb_debug_hooks): Likewise.
20934 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20935 * dwarf2out.c (macinfo_label_base): New global.
20936 (dwarf2out_register_external_die): New function for the
20937 register_external_die hook.
20938 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20939 (dwarf2_debug_hooks): Use them.
20940 (dwarf2_lineno_debug_hooks): Adjust.
20941 (struct die_struct): Add with_offset flag.
20942 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20943 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20944 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20945 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20946 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20947 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20948 defining section names for the early LTO debug variants.
20949 (reset_indirect_string): New helper.
20950 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20951 (print_dw_val): Add support for offsetted symbol references.
20952 (get_ultimate_context): Split out from is_cxx.
20953 (is_cxx): Use get_ultimate_context.
20954 (is_fortran): Add decl overload.
20955 (compute_comp_unit_symbol): Split out worker from
20956 compute_section_prefix.
20957 (compute_section_prefix): Call compute_comp_unit_symbol and
20958 set comdat_type_p here.
20959 (output_die): Skip DIE symbol output for the LTO added one.
20960 Handle DIE symbol references with offset.
20961 (output_comp_unit): Guard section name mangling properly.
20962 For LTO debug sections emit a symbol at the section beginning
20963 which we use to refer to its DIEs.
20964 (add_abstract_origin_attribute): For DIEs registered via
20965 dwarf2out_register_external_die directly refer to the early
20966 DIE rather than indirectly through the shadow one we created.
20967 Remove obsolete call to dwarf2out_abstract_function for
20968 non-function/block origins.
20969 (gen_array_type_die): When generating early LTO debug do
20970 not emit DW_AT_string_length.
20971 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20972 late when in LTO. As suggested place a gcc_unreachable for
20973 the DECL_ABSTRACT_P case.
20974 (gen_subprogram_die): Avoid another specification DIE
20975 for early built declarations/definitions for the late LTO case.
20976 (gen_variable_die): Add type references for late duplicated VLA dies
20978 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20979 we have the abstract instance already.
20980 (process_scope_var): Adjust decl DIE contexts in LTO which
20981 first puts them in limbo.
20982 (gen_decl_die): Do not generate type DIEs late apart from
20983 types for VLAs or for decls we do not yet have a DIE. Do not
20984 call dwarf2out_abstract_function late.
20985 (dwarf2out_early_global_decl): Make sure to create DIEs
20986 for abstract instances of a decl first.
20987 (dwarf2out_late_global_decl): Adjust comment.
20988 (output_macinfo_op): With multiple macro sections use
20989 macinfo_label_base to distinguish labels.
20990 (output_macinfo): Likewise. Update macinfo_label_base.
20991 Pass in the line info label.
20992 (note_variable_value_in_expr): When generating LTO resolve
20993 all variable values here by generating DIEs as needed.
20994 (init_sections_and_labels): Add early LTO debug flag parameter
20995 and generate different sections and names if set. Add generation
20996 counter for the labels so we can have multiple of them.
20997 (reset_dies): Helper to allow DIEs to be output multiple times.
20998 (dwarf2out_finish): When outputting DIEs to the fat part of an
20999 LTO object first reset DIEs.
21000 (dwarf2out_early_finish): Output early DIEs when generating LTO.
21001 (modified_type_die): Check for decl_ultimate_origin being self
21003 (gen_type_die_with_usage): Likewise.
21004 (gen_typedef_die): Allow decl_ultimate_origin being self.
21005 (set_decl_abstract_flags): Remove.
21006 (set_block_abstract_flags): Likewise.
21007 (dwarf2out_abstract_function): Treat the early generated DIEs
21008 as the abstract copy and only add DW_AT_inline and
21009 DW_AT_artificial here and call set_decl_origin_self.
21010 If the DIE has an abstract origin don't do anything.
21011 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21012 if we have none yet (Go fails to build one, PR78628).
21013 (variably_modified_type_p): Prevent endless recursion for Ada
21014 cyclic pointer types.
21015 * lto-streamer-in.c: Include debug.h.
21016 (dref_queue): New global.
21017 (lto_read_tree_1): Stream in DIE references.
21018 (lto_input_tree): Register DIE references.
21019 (input_function): Stream DECL_DEBUG_ARGS.
21020 * lto-streamer-out.c: Include debug.h.
21021 (lto_write_tree_1): Output DIE references.
21022 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21023 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21024 (output_function): Stream DECL_DEBUG_ARGS.
21025 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21026 Stream DECL_ABSTRACT_ORIGIN.
21027 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21028 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21029 DECL_CONTEXT for file-scope decls.
21030 * lto-streamer.h (struct dref_entry): Declare.
21031 (dref_queue): Likewise.
21032 * cfgexpand.c (pass_expand::execute): Do not call the
21033 outlining_inline_function hook here.
21034 * lto-wrapper.c (debug_obj): New global.
21035 (tool_cleanup): Unlink it if required.
21036 (debug_objcopy): New function.
21037 (run_gcc): Handle early debug sections in the IL files by
21038 extracting them to separate files, partially linkin them and
21039 feeding the result back as result to the linker.
21040 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21041 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21042 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21043 sections into a separate segment.
21044 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21046 (darwin_asm_dwarf_section): Likewise.
21047 (darwin_asm_output_dwarf_offset): Likewise.
21048 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
21050 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21051 Alan Hayward <alan.hayward@arm.com>
21052 David Sherwood <david.sherwood@arm.com>
21054 * read-md.h (md_reader::record_potential_iterator_use): Replace
21055 pointer argument with an rtx and an index.
21056 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21057 (apply_mode_iterator): Likewise.
21058 (apply_code_iterator): Likewise.
21059 (apply_int_iterator): Likewise.
21060 (apply_subst_iterator): Likewise.
21061 (record_iterator_use): Likewise.
21062 (record_attribute_use): Likewise.
21063 (md_reader::record_potential_iterator_use): Likewise. Update calls
21064 to record_iterator_use and apply_iterator.
21065 (iterator_use): Replace ptr with x and index.
21066 (attribute_use): Likewise.
21067 (apply_attribute_uses): Update calls to apply_iterator.
21068 (apply_iterators): Likewise. Update initialization of iterator_use.
21069 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21070 and record_potential_iterator_use.
21071 (rtx_reader::read_rtx_operand): Likewise.
21073 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21074 Alan Hayward <alan.hayward@arm.com>
21075 David Sherwood <david.sherwood@arm.com>
21077 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21080 2017-08-21 Richard Biener <rguenther@suse.de>
21082 PR middle-end/81884
21083 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21084 at struct end conservatively when comparing common bases.
21086 2017-08-21 Richard Biener <rguenther@suse.de>
21088 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21089 (mem_ref_in_stmt): Remove.
21090 (determine_max_movement): Use ref index to get at the reference.
21091 (invariantness_dom_walker::before_dom_children): Deal with
21092 lim data already initialized.
21093 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21095 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21097 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21098 (TARGET_ISA_ROUND): Ditto.
21099 (TARGET_ROUND): Ditto.
21100 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21101 * config/i386/i386.md: Ditto.
21102 * config/i386/sse.md: Ditto.
21103 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21104 with OPTION_MASK_ISA_SSE4_1.
21106 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21109 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21110 __builtin_ia32_lzcnt_u16.
21112 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21115 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21116 (rs6000_set_current_function): Rewrite function to use it.
21118 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21121 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21122 and TYPE_WARN_IF_NOT_ALIGN.
21123 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21124 (handle_warn_if_not_align): New.
21125 (place_union_field): Call handle_warn_if_not_align.
21126 (place_field): Call handle_warn_if_not_align.
21127 Copy TYPE_WARN_IF_NOT_ALIGN.
21128 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21129 (layout_type): Likewise.
21130 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21132 (tree_decl_common): Add warn_if_not_align.
21133 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21134 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21135 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21136 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21137 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21138 * doc/extend.texi: Document warn_if_not_aligned attribute.
21139 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21141 2017-08-17 Martin Liska <mliska@suse.cz>
21144 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21145 (get_data_dependence): Use it as pointer type.
21146 (distribute_loop): Likewise.
21148 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21150 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21151 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21152 (p8_vmrgow_<mode>_direct): New define_insn.
21153 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21154 handle endianness for vmrgew and vmrgow permute patterns.
21156 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21158 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21159 * config/rs6000/rs6000-cpus.def: Remove comment.
21160 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21161 (POWERPC_MASKS): Likewise.
21162 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21163 use of TARGET_VSX_TIMODE.
21164 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21165 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21166 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21167 (rs6000_option_override_internal): Remove dead code.
21168 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21169 (rs6000_legitimize_reload_address): Likewise.
21170 (rs6000_legitimate_address_p): Likewise.
21171 (rs6000_opt_masks): Delete "vsx-timode".
21172 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21173 from function comment.
21174 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21175 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21176 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21178 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21179 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21180 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21181 (VSX_TI): Likewise.
21183 (define_peephole2): Likewise.
21185 2017-08-17 Martin Sebor <msebor@redhat.com>
21188 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21189 past the end of an array.
21190 (test_pp_format): Add test cases.
21192 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21194 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21195 missing ECF_NOTHROW flags.
21197 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21200 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21201 operands residing in integer registers.
21202 (*vsx_le_perm_load_<mode>): Likewise.
21203 (*vsx_le_perm_store_<mode>): Likewise.
21204 (define_peephole2): Add peepholes to optimize the above.
21206 2017-08-17 Marek Polacek <polacek@redhat.com>
21208 PR middle-end/81814
21209 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21210 to mimic what shorten_compare did. Change the return type to bool.
21211 (fold_cond_expr_with_comparison): Update call to
21212 operand_equal_for_comparison_p.
21213 (fold_ternary_loc): Likewise.
21215 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21217 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21219 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21221 2017-08-17 Richard Biener <rguenther@suse.de>
21223 * tree-ssa-structalias.c (solve_graph): When propagating
21224 to successors update the graphs succ edges and avoid duplicate work.
21226 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21229 * config/i386/i386.c (ix86_option_override_internal): Save target
21230 specific options after ix86_stack_protector_guard_reg was changed.
21232 2017-08-17 Richard Biener <rguenther@suse.de>
21234 PR tree-optimization/81827
21235 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21236 (new_var_info): Initialize it conservatively.
21237 (get_call_vi): Mark register vars.
21238 (new_scalar_tmp_constraint_exp): Likewise.
21239 (handle_rhs_call): Likewise.
21240 (handle_const_call): Likewise.
21241 (create_function_info_for): Likewise.
21242 (solve_constraints): Sort varinfos to separate register from
21243 non-register vars to pack points-to solution bitmaps during
21246 2017-08-17 Marek Polacek <polacek@redhat.com>
21248 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21250 2017-08-17 Richard Biener <rguenther@suse.de>
21252 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21253 to true when overflow is undefined and we saturated the result.
21255 2017-08-17 Alan Modra <amodra@gmail.com>
21258 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21259 Don't use store multiple if only one reg needs saving.
21260 (interesting_frame_related_regno): New function.
21261 (rs6000_frame_related): Don't emit frame info for regs that
21263 (rs6000_emit_epilogue): Likewise.
21265 2017-08-16 Nathan Sidwell <nathan@acm.org>
21267 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21268 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21269 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21270 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21271 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21272 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21273 (verify_type): Adjust for TYPE_BINFO move.
21274 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21275 process TYPE_BINFO directly.
21276 (hash_tree): Likewise.
21277 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21279 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21282 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21284 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21286 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21289 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21290 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21291 Add ix86_binary_operator_ok to insn constraint.
21292 (*iordi_1_bts): Ditto.
21293 (*xordi_1_btc): Ditto.
21294 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21295 Update corresponding peephole2 pattern.
21299 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21301 PR tree-optimization/81832
21302 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21303 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21305 2017-08-16 Marek Polacek <polacek@redhat.com>
21308 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21309 regarding size_zero_node.
21311 2017-08-16 Martin Liska <mliska@suse.cz>
21314 * config.gcc: Respect previously set extra_objs in case
21317 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21319 PR tree-optimization/81835
21320 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21321 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21322 not depend on the phi.
21324 2017-08-16 Alan Modra <amodra@gmail.com>
21326 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21329 2017-08-16 Alan Modra <amodra@gmail.com>
21331 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21332 (save_reg_p): ..into this. Update all callers.
21333 (first_reg_to_save): Simplify.
21335 2017-08-16 Alan Modra <amodra@gmail.com>
21337 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21340 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21344 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21345 constant count if that count is less than 32.
21347 2017-08-15 Nathan Sidwell <nathan@acm.org>
21349 * gcc.c (execute): Emit friendlier message if inferior is killed
21350 by an external cause.
21352 2017-08-15 Richard Biener <rguenther@suse.de>
21354 PR tree-optimization/81790
21355 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21356 CONSTRUCTORs from simplifying and VN.
21358 2017-08-14 Martin Sebor <msebor@redhat.com>
21360 * builtin-attrs.def: Add comments.
21362 2017-08-14 Martin Sebor <msebor@redhat.com>
21365 * doc/extend.texi (attribute nonstring): Document new attribute.
21367 2017-08-14 Martin Sebor <msebor@redhat.com>
21370 * tree-diagnostic.c (default_tree_printer): Handle %G.
21371 * gimple-pretty-print.h (percent_G_format): Declare new function.
21372 * gimple-pretty-print.c (percent_G_format): Define.
21373 * tree-pretty-print.c (percent_K_format): Add argument.
21375 2017-08-14 Martin Sebor <msebor@redhat.com>
21377 PR translation/79998
21378 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21379 Remove a stray space.
21381 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21384 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21385 (*iordi_1_bts): Ditto.
21386 (*xordi_1_btc): Ditto.
21388 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21391 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21392 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21394 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21395 quoted strings, and make more translator-friendly.
21396 (darwin_rs6000_override_options): Likewise.
21397 (rs6000_option_override_internal): Likewise.
21398 (rs6000_return_in_memory): Fix overlong line.
21399 (init_cmulative_args): Use quoted strings, and make more
21400 translator-friendly.
21401 (rs6000_pass_by_reference): Fix overlong line.
21402 (def_builtin): Use quoted strings.
21403 (altivec_expand_predicate_builtin): Use quoted strings, and make
21404 more translator-friendly.
21405 (htm_expand_builtin): Use quoted strings.
21406 (cpu_expand_builtin): Use quoted strings, and make more
21407 translator-friendly.
21408 (altivec_expand_builtin): Likewise.
21409 (paired_expand_predicate_builtin): Likewise.
21410 (rs6000_invalid_builtin): Likewise.
21411 (builtin_function_type): Use quoted strings.
21412 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21413 more translator-friendly.
21414 (rs6000_trampoline_init): Likewise.
21415 (rs6000_handle_altivec_attribute): Likewise.
21416 (rs6000_inner_target_options): Use quoted strings.
21417 (rs6000_disable_incompatible_switches): Likewise.
21418 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21419 strings, and make more translator-friendly.
21420 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21422 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21424 PR tree-optimization/81799
21425 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21426 cond_expr to simple gimple operand.
21428 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21430 PR middle-end/46932
21431 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21433 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21437 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21438 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21439 TARGET_GASISR_PROLOGUES.
21440 * config/avr/avr.c (avr_option_override): Same.
21441 (avr_pass_pre_proep::execute): Same.
21443 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21446 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21447 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21449 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21451 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21452 define_insn_and_split. Split to a memory load from 0 in
21453 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21454 using PTR mode iterator.
21455 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21456 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21457 (*add_tp_<mode>): Redefine as define_insn_and_split.
21458 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21459 address space. Merge with *add_tp_x32 using PTR mode iterator.
21460 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21461 Split to an add with a memory load from 0 in
21462 DEFAULT_TLS_SEG_REG address space.
21464 2017-08-12 Andrew Pinski <apinski@cavium.com>
21466 * config/aarch64/aarch64-option-extensions.def (rdma):
21467 Fix feature string to what Linux prints out in /proc/cpuinfo.
21469 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21472 * dwarf2out.c (modified_type_die): For C typedef types that have
21473 an ultimate origin, process the ultimate origin instead of the
21475 (gen_typedef_die): Assert that input DECLs have no ultimate
21477 (gen_type_die_with_usage): For typedef variants that have an
21478 ultimate origin, just call gen_decl_die on the original DECL.
21479 (process_scope_var): Avoid creating DIEs for local typedefs and
21480 concrete static variables.
21482 2017-08-12 Alan Modra <amodra@gmail.com>
21486 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21487 match gnu-user.h startfile.
21488 (ENDFILE_LINUX_SPEC): Similarly.
21490 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21493 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21495 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21497 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21498 * config/aarch64/aarch64.md (mov<mode>): Change.
21499 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21500 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21501 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21503 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21505 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21506 for storage order barriers.
21508 2017-08-11 Martin Liska <mliska@suse.cz>
21511 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21512 variables of void type.
21514 2017-08-11 Martin Liska <mliska@suse.cz>
21516 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21517 TARGET_SUPPORTS_ALIASES.
21518 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21519 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21520 (optimize_weakref): Likewise.
21521 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21522 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21523 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21525 2017-08-11 Martin Liska <mliska@suse.cz>
21528 * config/i386/i386.c (make_resolver_func): Do complete
21529 refactoring of the function.
21531 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21534 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21535 * config/i386/i386.c (ix86_stack_protect_guard): Use
21536 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21537 * doc/invoke.texi (x86 Options): Document
21538 -mstack-protector-guard-symbol= option.
21540 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21542 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21543 * config/i386/i386.c (ix86_split_stack_guard): New function.
21544 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21545 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21546 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21547 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21548 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21549 (split_stack_space_check): Call ix86_split_stack_guard.
21551 2017-08-10 Martin Sebor <msebor@redhat.com>
21553 * print-tree.c (print_node): Print location using the established
21555 Replace spaces with colons.
21556 (debug_raw, debug): Ditto.
21558 2017-08-10 Martin Sebor <msebor@redhat.com>
21561 * pretty-print.c (pp_format): Correct the handling of %s precision.
21563 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21566 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21568 (ix86_finalize_stack_frame_flags): This. Also clear
21569 frame_pointer_needed if -fno-omit-frame-pointer is used without
21571 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21572 with ix86_finalize_stack_frame_flags.
21573 (ix86_expand_epilogue): Likewise.
21574 (ix86_expand_split_stack_prologue): Likewise.
21575 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21577 2017-08-10 Martin Liska <mliska@suse.cz>
21580 * c-attribs.c (handle_target_attribute):
21581 Report warning for an empty string argument of target attribute.
21583 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21586 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21588 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21589 or DECL_NONLOCAL labels.
21590 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21591 or DECL_NONLOCAL labels here.
21593 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21595 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21596 to indicate when early gimple folding has been disabled.
21597 (rs6000_gimple_fold_builtin): Add debug content.
21598 (rs6000_invalid_builtin): Fix whitespace.
21599 (rs6000_expand_builtin): Fix whitespace.
21600 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21602 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21605 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21606 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21608 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21610 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21611 (qdf24xx): Likewise.
21612 * config/aarch64/aarch64.md: Include falkor.md.
21613 * config/aarch64/falkor.md: New.
21615 2017-08-09 Marek Polacek <polacek@redhat.com>
21618 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21619 * diagnostic.c (emit_diagnostic): Add a comment.
21620 (emit_diagnostic_valist): New function.
21622 2017-08-09 Marek Polacek <polacek@redhat.com>
21625 * input.c (make_location): New overload.
21626 * input.h (make_location): Declare.
21628 2017-08-08 Alan Modra <amodra@gmail.com>
21629 H.J. Lu <hongjiu.lu@intel.com>
21632 * gcc.c (NO_PIE_SPEC): Delete.
21633 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21635 (LINK_PIE_SPEC): ..to here.
21636 (LINK_COMMAND_SPEC): Support -no-pie.
21637 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21638 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21640 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21641 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21642 (ENDFILE_CRTEND_SPEC): Similarly.
21644 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21647 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21648 (mstack-protector-guard-offset=): Ditto.
21649 * config/i386/i386.c (ix86_option_override): Handle
21650 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21652 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21653 ix86_stack_protect_guard_offset variables.
21654 (TARGET_STACK_PROTECT_GUARD): Always define.
21655 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21656 and -mstack-protector-guard-offset= options.
21658 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21660 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21661 boundary case for the last candidate.
21663 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21665 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21666 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21668 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21670 PR middle-end/19706
21671 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21672 * config/aarch64/aarch64-builtins.c
21673 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21674 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21675 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21677 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21678 Andrew Pinski <pinskia@gmail.com>
21680 PR middle-end/19706
21681 * internal-fn.def (XORSIGN): New.
21682 * optabs.def (xorsign_optab): New.
21683 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21684 (convert_expand_mult_copysign): New.
21685 (pass_optimize_widening_mul::execute): Call
21686 convert_expand_mult_copysign.
21688 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21690 PR tree-optimization/81354
21691 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21692 Insert on edges rather than explicitly creating landing pads.
21693 (analyze_candidates_and_replace): Commit edge inserts.
21695 2017-08-08 Richard Biener <rguenther@suse.de>
21697 PR middle-end/81719
21698 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21699 (expand_simple_operations): Also look through ADDR_EXPRs with
21700 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21702 2017-08-08 Richard Biener <rguenther@suse.de>
21704 PR tree-optimization/81723
21705 * tree-vect-slp.c (struct bst_traits): New hash traits.
21706 (bst_fail): New global.
21707 (vect_build_slp_tree_2): New worker, split out from ...
21708 (vect_build_slp_tree): ... this now wrapping it with using
21709 bst_fail set to cache SLP tree build fails. Properly handle
21711 (vect_analyze_slp_instance): Allocate and free bst_fail.
21713 2017-08-08 Martin Liska <mliska@suse.cz>
21716 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21717 LABEL_DECLs that can be from a different function.
21719 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21721 PR tree-optimization/81744
21722 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21723 loop's number of iterations.
21725 2017-08-08 Martin Liska <mliska@suse.cz>
21727 * asan.c: Include header files.
21728 * attribs.c (build_decl_attribute_variant): New function moved
21730 (build_type_attribute_qual_variant): Likewise.
21731 (cmp_attrib_identifiers): Likewise.
21732 (simple_cst_list_equal): Likewise.
21733 (omp_declare_simd_clauses_equal): Likewise.
21734 (attribute_value_equal): Likewise.
21735 (comp_type_attributes): Likewise.
21736 (build_type_attribute_variant): Likewise.
21737 (lookup_ident_attribute): Likewise.
21738 (remove_attribute): Likewise.
21739 (merge_attributes): Likewise.
21740 (merge_type_attributes): Likewise.
21741 (merge_decl_attributes): Likewise.
21742 (merge_dllimport_decl_attributes): Likewise.
21743 (handle_dll_attribute): Likewise.
21744 (attribute_list_equal): Likewise.
21745 (attribute_list_contained): Likewise.
21746 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21747 (lookup_attribute_by_prefix): Likewise.
21748 * bb-reorder.c: Include header files.
21749 * builtins.c: Likewise.
21750 * calls.c: Likewise.
21751 * cfgexpand.c: Likewise.
21752 * cgraph.c: Likewise.
21753 * cgraphunit.c: Likewise.
21754 * convert.c: Likewise.
21755 * dwarf2out.c: Likewise.
21756 * final.c: Likewise.
21757 * fold-const.c: Likewise.
21758 * function.c: Likewise.
21759 * gimple-expr.c: Likewise.
21760 * gimple-fold.c: Likewise.
21761 * gimple-pretty-print.c: Likewise.
21762 * gimple.c: Likewise.
21763 * gimplify.c: Likewise.
21764 * hsa-common.c: Likewise.
21765 * hsa-gen.c: Likewise.
21766 * internal-fn.c: Likewise.
21767 * ipa-chkp.c: Likewise.
21768 * ipa-cp.c: Likewise.
21769 * ipa-devirt.c: Likewise.
21770 * ipa-fnsummary.c: Likewise.
21771 * ipa-inline.c: Likewise.
21772 * ipa-visibility.c: Likewise.
21774 * lto-cgraph.c: Likewise.
21775 * omp-expand.c: Likewise.
21776 * omp-general.c: Likewise.
21777 * omp-low.c: Likewise.
21778 * omp-offload.c: Likewise.
21779 * omp-simd-clone.c: Likewise.
21780 * opts-global.c: Likewise.
21781 * passes.c: Likewise.
21782 * predict.c: Likewise.
21783 * sancov.c: Likewise.
21784 * sanopt.c: Likewise.
21785 * symtab.c: Likewise.
21786 * toplev.c: Likewise.
21787 * trans-mem.c: Likewise.
21788 * tree-chkp.c: Likewise.
21789 * tree-eh.c: Likewise.
21790 * tree-into-ssa.c: Likewise.
21791 * tree-object-size.c: Likewise.
21792 * tree-parloops.c: Likewise.
21793 * tree-profile.c: Likewise.
21794 * tree-ssa-ccp.c: Likewise.
21795 * tree-ssa-live.c: Likewise.
21796 * tree-ssa-loop.c: Likewise.
21797 * tree-ssa-sccvn.c: Likewise.
21798 * tree-ssa-structalias.c: Likewise.
21799 * tree-ssa.c: Likewise.
21800 * tree-streamer-in.c: Likewise.
21801 * tree-vectorizer.c: Likewise.
21802 * tree-vrp.c: Likewise.
21803 * tsan.c: Likewise.
21804 * ubsan.c: Likewise.
21805 * varasm.c: Likewise.
21806 * varpool.c: Likewise.
21807 * tree.c: Remove functions moved to attribs.[ch].
21808 * tree.h: Likewise.
21809 * config/aarch64/aarch64.c: Add attrs.h header file.
21810 * config/alpha/alpha.c: Likewise.
21811 * config/arc/arc.c: Likewise.
21812 * config/arm/arm.c: Likewise.
21813 * config/avr/avr.c: Likewise.
21814 * config/bfin/bfin.c: Likewise.
21815 * config/c6x/c6x.c: Likewise.
21816 * config/cr16/cr16.c: Likewise.
21817 * config/cris/cris.c: Likewise.
21818 * config/darwin.c: Likewise.
21819 * config/epiphany/epiphany.c: Likewise.
21820 * config/fr30/fr30.c: Likewise.
21821 * config/frv/frv.c: Likewise.
21822 * config/ft32/ft32.c: Likewise.
21823 * config/h8300/h8300.c: Likewise.
21824 * config/i386/winnt.c: Likewise.
21825 * config/ia64/ia64.c: Likewise.
21826 * config/iq2000/iq2000.c: Likewise.
21827 * config/lm32/lm32.c: Likewise.
21828 * config/m32c/m32c.c: Likewise.
21829 * config/m32r/m32r.c: Likewise.
21830 * config/m68k/m68k.c: Likewise.
21831 * config/mcore/mcore.c: Likewise.
21832 * config/microblaze/microblaze.c: Likewise.
21833 * config/mips/mips.c: Likewise.
21834 * config/mmix/mmix.c: Likewise.
21835 * config/mn10300/mn10300.c: Likewise.
21836 * config/moxie/moxie.c: Likewise.
21837 * config/msp430/msp430.c: Likewise.
21838 * config/nds32/nds32-isr.c: Likewise.
21839 * config/nds32/nds32.c: Likewise.
21840 * config/nios2/nios2.c: Likewise.
21841 * config/nvptx/nvptx.c: Likewise.
21842 * config/pa/pa.c: Likewise.
21843 * config/pdp11/pdp11.c: Likewise.
21844 * config/powerpcspe/powerpcspe.c: Likewise.
21845 * config/riscv/riscv.c: Likewise.
21846 * config/rl78/rl78.c: Likewise.
21847 * config/rx/rx.c: Likewise.
21848 * config/s390/s390.c: Likewise.
21849 * config/sh/sh.c: Likewise.
21850 * config/sol2.c: Likewise.
21851 * config/sparc/sparc.c: Likewise.
21852 * config/spu/spu.c: Likewise.
21853 * config/stormy16/stormy16.c: Likewise.
21854 * config/tilegx/tilegx.c: Likewise.
21855 * config/tilepro/tilepro.c: Likewise.
21856 * config/v850/v850.c: Likewise.
21857 * config/vax/vax.c: Likewise.
21858 * config/visium/visium.c: Likewise.
21859 * config/xtensa/xtensa.c: Likewise.
21861 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21864 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21865 constraints since the -mupper-regs-* switches have been
21867 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21868 into a vector from a double word element that was extracted from
21869 another vector, and eliminate extra XXPERMDI instructions.
21870 (vsx_concat_<mode>_2): Likewise.
21871 (vsx_concat_<mode>_3): Likewise.
21872 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21873 concat to allow optimizing inserts from previous extracts.
21875 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21877 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21878 memory reference to a SSP offset in TLS address space.
21879 (ix86_print_operand) <case '@'>: Remove.
21880 (ix86_print_operand_punct_valid_p): Remove '@' code.
21881 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21882 UNSPEC_SP_TLS_TEST.
21883 (stack_tls_protect_set_<mode>): Remove.
21884 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21885 (stack_tls_protect_test_<mode>): Remove.
21886 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21888 2017-08-07 Olivier Hainque <hainque@adacore.com>
21891 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21893 2017-08-07 Douglas Rupp <rupp@adacore.com>
21895 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21896 variable was referenced as multidir in command.
21898 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21901 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21904 2017-08-07 Martin Liska <mliska@suse.cz>
21906 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21907 * config/rl78/rl78.c: Add include of attribs.h.
21908 * config/sh/sh.c: Likewise.
21909 * config/v850/v850.c: Likewise.
21911 2017-08-07 Tom de Vries <tom@codesourcery.com>
21913 PR middle-end/78266
21914 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21916 2017-08-07 Martin Liska <mliska@suse.cz>
21918 * config/mips/mips.c: Include attribs.h.
21920 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21923 * doc/invoke.texi: Document change in behvaior for -Ofast for
21926 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21928 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21930 (aarch64_pop_regs): Likewise.
21931 (aarch64_gen_load_pair): Likewise.
21932 (aarch64_save_callee_saves): Likewise.
21933 (aarch64_restore_callee_saves): Likewise.
21935 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21937 * config/i386/i386.c: Revert the last change.
21939 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21942 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21944 (ix86_finalize_stack_frame_flags): This. Also clear
21945 frame_pointer_needed if -fno-omit-frame-pointer is used without
21947 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21948 with ix86_finalize_stack_frame_flags.
21949 (ix86_expand_epilogue): Likewise.
21950 (ix86_expand_split_stack_prologue): Likewise.
21952 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21955 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21956 to P_AES for Westmere.
21958 2017-08-07 Jonathan Yong <10walls@gmail.com>
21960 * config/i386/mingw.opt (fset-stack-executable): Removed.
21961 * config/i386/cygming.opt (fset-stack-executable): Moved
21963 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21965 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21967 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21969 2017-08-07 Marek Polacek <polacek@redhat.com>
21971 PR middle-end/81737
21972 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21974 2017-08-07 Martin Liska <mliska@suse.cz>
21976 * attribs.h (canonicalize_attr_name): New function.
21977 (cmp_attribs): Move from c-format.c and adjusted.
21978 (is_attribute_p): Moved from tree.h.
21979 * tree-inline.c: Add new includes.
21980 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21981 (private_is_attribute_p): Remove.
21982 (private_lookup_attribute): Likewise.
21983 (private_lookup_attribute_by_prefix): Simplify.
21984 (remove_attribute): Use is_attribute_p.
21985 * tree.h: Remove removed declarations.
21987 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21989 PR middle-end/81698
21990 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21991 instead of computing it in the function. Formatting fix.
21992 (expand_case): Don't rely on default_edge being the first edge,
21993 clear it if removing it, pass default_edge to
21994 emit_case_dispatch_table.
21995 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21998 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
22000 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
22001 insn in the function, emit NOP after the insn.
22003 2017-08-06 Tom de Vries <tom@codesourcery.com>
22005 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22008 2017-08-06 Tom de Vries <tom@codesourcery.com>
22010 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22013 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22015 PR tree-optimization/57371
22016 * match.pd: New pattern.
22018 2017-08-04 Marek Polacek <polacek@redhat.com>
22020 PR middle-end/81695
22021 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22022 perform the computation in offset_int.
22024 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22026 PR tree-optimization/81136
22027 * tree-vectorizer.h: Include tree-hash-traits.h.
22028 (vec_base_alignments): New typedef.
22029 (vec_info): Add a base_alignments field.
22030 (vect_record_base_alignments): Declare.
22031 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22033 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22034 (create_data_ref): Add an is_conditional_in_stmt argument.
22035 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22036 the is_conditional_in_stmt field.
22037 (data_ref_loc): Add an is_conditional_in_stmt field.
22038 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22039 (find_data_references_in_stmt): Update call to create_data_ref.
22040 (graphite_find_data_references_in_stmt): Likewise.
22041 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22042 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22043 (vect_record_base_alignment): New function.
22044 (vect_record_base_alignments): Likewise.
22045 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22046 for nested statements even if we fail to compute a misalignment.
22047 Use pooled base alignments for unconditional references.
22048 (vect_find_same_alignment_drs): Compare base addresses instead
22050 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22051 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22053 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22055 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22056 Add an explicit name for the enum. Use auto_vec for slp_instances
22057 and grouped_stores.
22058 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22060 (_bb_vec_info): Add a constructor and destructor.
22061 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22062 (destroy_loop_vec_info): Delete.
22063 (vect_destroy_datarefs): Likewise.
22064 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22065 (vec_info::vec_info): New function.
22066 (vec_info::~vec_info): Likewise.
22067 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22068 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22069 destroy_loop_vec_info.
22070 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22071 (_loop_vec_info::_loop_vec_info): ...this.
22072 (destroy_loop_vec_info): Replace with...
22073 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22074 the stmt_vec_infos. Leave handling of vec_info information to its
22075 destructor. Remove explicit vector releases.
22076 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22077 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22078 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22079 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22080 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22081 (destroy_bb_vec_info): Replace with...
22082 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22083 information to its destructor.
22084 (vect_slp_analyze_bb_1): Use new and delete instead of
22085 new_bb_vec_info and destroy_bb_vec_info.
22086 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22089 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22091 * tree-data-ref.h (subscript): Add access_fn field.
22092 (data_dependence_relation): Add could_be_independent_p.
22093 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22094 (same_access_functions): Move to tree-data-ref.c.
22095 * tree-data-ref.c (ref_contains_union_access_p): New function.
22096 (access_fn_component_p): Likewise.
22097 (access_fn_components_comparable_p): Likewise.
22098 (dr_analyze_indices): Add a reference to access_fn_component_p.
22099 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22101 (constant_access_functions): Likewise.
22102 (add_other_self_distances): Likewise.
22103 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22104 (initialize_data_dependence_relation): Use XCNEW and remove
22105 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22106 of access functions that have the same type. Allow the
22107 subsequence to end with different bases in some circumstances.
22108 Record the chosen access functions in SUB_ACCESS_FN.
22109 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22110 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22111 (subscript_dependence_tester_1): Likewise dra and drb.
22112 (build_classic_dist_vector): Update calls accordingly.
22113 (subscript_dependence_tester): Likewise.
22114 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22115 DDR_COULD_BE_INDEPENDENT_P.
22116 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22117 comp_alias_ddrs instead of may_alias_ddrs.
22118 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22120 (vect_analyze_data_ref_dependence): Use it if
22121 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22122 distance vectors if that fails.
22123 (dependence_distance_ge_vf): New function.
22124 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22125 LOOP_VINFO_MAY_ALIAS_DDRS.
22127 2017-08-04 Richard Biener <rguenther@suse.de>
22129 PR middle-end/81705
22130 * fold-const.c (fold_binary_loc): Properly restrict
22131 minus_var0 && minus_var1 case when associating undefined overflow
22134 2017-08-04 Tom de Vries <tom@codesourcery.com>
22136 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22138 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22140 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22141 Don't start diagnostic messages with a capital letter.
22142 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22144 (rs6000_invalid_builtin): Likewise.
22145 (rs6000_trampoline_init): Likewise.
22147 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22150 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22151 after setting changeable df flags.
22153 2017-08-03 Richard Biener <rguenther@suse.de>
22155 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22156 up if the use is in USE - X.
22158 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22160 * toplev.c (dumpfile.h): New include.
22161 (internal_error_reentered): New static function. Use it...
22162 (internal_error_function): ...here to handle reentered internal_error.
22164 2017-08-03 Richard Biener <rguenther@suse.de>
22166 PR middle-end/81148
22167 * fold-const.c (split_tree): Add minus_var and minus_con
22168 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22169 here but always use minus_*.
22170 (associate_trees): Assert we never associate with MINUS_EXPR
22171 and NULL first operand. Do not recurse for PLUS_EXPR operands
22172 when associating as MINUS_EXPR either.
22173 (fold_binary_loc): Track minus_var and minus_con.
22175 2017-08-03 Tom de Vries <tom@codesourcery.com>
22178 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22179 ACCEL_COMPILER, apply finish_options on
22180 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22182 2017-08-03 Tom de Vries <tom@codesourcery.com>
22185 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22186 function_entry_patch_area_size > 0.
22188 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22191 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22192 instead of 10??LU, perform unit multiplication in wide_int,
22193 don't change alloc_object_size_limit if the limit is larger
22196 PR tree-optimization/81655
22197 PR tree-optimization/81588
22198 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22199 the case when ranges[i].low and high are 1 for unsigned type with
22202 PR middle-end/81052
22203 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22204 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22206 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22208 * tree-vrp.h: Add include guard.
22210 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22213 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22214 (ud2): New insn pattern.
22215 * config/i386/i386.c (ix86_expand_epilogue):
22216 For naked functions, generate ud2 instead of trap insn.
22218 2017-08-02 Marek Polacek <polacek@redhat.com>
22221 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22223 2017-08-02 Tom de Vries <tom@codesourcery.com>
22224 Cesar Philippidis <cesar@codesourcery.com>
22226 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22227 Add missing edge probabilities.
22229 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22231 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22232 Correct endianness.
22234 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22236 PR middle-end/79499
22237 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22238 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22239 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22240 prologue_seq sequences - if any.
22242 2017-08-02 Richard Biener <rguenther@suse.de>
22244 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22245 via vectors if supported, integer extracts via punning if supported
22246 or otherwise vector extracts.
22248 2017-08-02 Richard Biener <rguenther@suse.de>
22250 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22252 (bitmap_insert_into_set): ... this.
22254 2017-08-02 Richard Biener <rguenther@suse.de>
22256 PR tree-optimization/81633
22258 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22260 PR tree-optimization/71752
22261 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22263 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22265 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22266 (machine_function::call_ms2sysv_pad_out): Remove field.
22267 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22268 (ix86_compute_frame_layout): Likewise.
22270 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22273 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22274 attribute with interrupt attribute.
22276 2017-08-01 Andrew Pinski <apinski@cavium.com>
22278 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22279 BIT_INSERT_EXPR's operand 1
22280 to see if the types precision matches.
22282 2017-08-01 Martin Liska <mliska@suse.cz>
22284 PR middle-end/70140
22285 * builtins.c (expand_builtin_memcpy_args): Remove.
22286 (expand_builtin_memcpy): Call newly added function
22287 expand_builtin_memory_copy_args.
22288 (expand_builtin_memcpy_with_bounds): Likewise.
22289 (expand_builtin_mempcpy): Remove last argument.
22290 (expand_builtin_mempcpy_with_bounds): Likewise.
22291 (expand_builtin_memory_copy_args): New function created from
22292 expand_builtin_mempcpy_args with small modifications.
22293 (expand_builtin_mempcpy_args): Remove.
22294 (expand_builtin_stpcpy): Remove unused argument.
22295 (expand_builtin): Likewise.
22296 (expand_builtin_with_bounds): Likewise.
22298 2017-08-01 Martin Liska <mliska@suse.cz>
22301 Make mempcpy more optimal (PR middle-end/70140).
22303 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22306 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22307 __builtin_vec_cmpne verify both arguments are compatible vectors
22308 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22309 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22310 move computation of aligned to after checking the argument types.
22314 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22315 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22318 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22319 Doug Rupp <rupp@adacore.com>
22320 Olivier Hainque <hainque@adacore.com>
22322 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22323 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22325 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22326 for TARGET_OS_CPP_BUILTIN.
22327 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22328 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22330 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22331 passing required abi options to the assembler for EABI configurations.
22332 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22333 of .text.hot and .text.unlikely sections for kernel modules when
22334 using ARM style exceptions.
22335 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22336 options. Add EXTRA_CC1_SPEC.
22337 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22339 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22341 (ARM_TARGET2_DWARF_FORMAT): Define.
22342 * config/arm/t-vxworks: Adjust multilib control to removal of the
22343 Diab command line options.
22345 2017-08-01 Martin Liska <mliska@suse.cz>
22347 PR gcov-profile/81561
22348 * gcov.c (unblock): Make unblocking safe as we need to preserve
22349 index correspondence of blocks and block_lists.
22351 2017-08-01 Richard Biener <rguenther@suse.de>
22353 PR tree-optimization/81181
22354 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22355 (compute_antic): ... end of iteration here.
22357 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22359 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22360 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22361 (ftree-slp-vectorize): Likewise.
22362 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22363 can no longer be set independent of flag_tree_loop_vectorize.
22364 * omp-general.c (emp_max_vf): Likewise.
22365 * opts.c (enable_fdo_optimizations): Remove references to
22366 flag_tree_vectorize, these are now implicit.
22367 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22368 and leave it for the options machinery.
22370 2017-08-01 Martin Liska <mliska@suse.cz>
22372 PR middle-end/70140
22373 * builtins.c (expand_builtin_memcpy_args): Remove.
22374 (expand_builtin_memcpy): Call newly added function
22375 expand_builtin_memory_copy_args.
22376 (expand_builtin_memcpy_with_bounds): Likewise.
22377 (expand_builtin_mempcpy): Remove last argument.
22378 (expand_builtin_mempcpy_with_bounds): Likewise.
22379 (expand_builtin_memory_copy_args): New function created from
22380 expand_builtin_mempcpy_args with small modifications.
22381 (expand_builtin_mempcpy_args): Remove.
22382 (expand_builtin_stpcpy): Remove unused argument.
22383 (expand_builtin): Likewise.
22384 (expand_builtin_with_bounds): Likewise.
22386 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22389 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22390 print "ds:" only for immediates in generic address space.
22392 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22395 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22396 (ix86_function_ok_for_sibcall): Return false for naked functions.
22398 2017-08-01 Richard Biener <rguenther@suse.de>
22400 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22401 (compute_antic): Seed worklist with exit block predecessors.
22402 * cfganal.c (dfs_find_deadend): For a cycle return the source
22403 of the edge closing it.
22405 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22407 * config/aarch64/aarch64.c
22408 (aarch64_can_const_movi_rtx_p): Move 0 check.
22410 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22412 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22413 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22415 * match.pd: Ditto in address comparison pattern.
22417 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22419 PR tree-optimization/81627
22420 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22421 closed ssa form for store-store chain.
22423 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22425 PR tree-optimization/81620
22426 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22427 for store-store chain.
22429 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22431 PR tree-optimization/81588
22432 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22433 ranges[i].in_p, invert comparison code ccode. For >/>=,
22434 swap rhs1 and rhs2 and comparison code unconditionally,
22435 for </<= don't do that. Don't swap rhs1/rhs2 again if
22436 ranges[i].in_p, instead invert comparison code ccode if
22437 opcode or oe->rank is BIT_IOR_EXPR.
22440 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22441 a direct optab to conversion optab.
22442 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22443 with GET_MODE_INNER as last argument instead of optab_handler.
22444 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22445 vector extraction if possible and optab is available.
22446 * expr.c (store_constructor): Use convert_optab_handler instead
22447 of optab_handler. Use vector initialization from smaller
22448 vectors if possible and optab is available.
22449 * tree-vect-stmts.c (vectorizable_load): Likewise.
22450 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22451 now have two modes.
22452 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22453 of vec_init from half-sized vectors with the same element mode.
22454 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22455 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22456 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22457 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22458 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22459 after mode in gen_vec_extract* calls.
22460 (vec_extract<mode>): Renamed to ...
22461 (vec_extract<mode><ssescalarmodelower>): ... this.
22462 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22463 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22464 element mode after mode in gen_vec_init* calls.
22465 (VEC_INIT_HALF_MODE): New mode iterator.
22466 (vec_init<mode>): Renamed to ...
22467 (vec_init<mode><ssescalarmodelower>): ... this.
22468 (vec_init<mode><ssehalfvecmodelower>): New expander.
22469 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22470 (vec_extractv2sfsf): ... this.
22471 (vec_initv2sf): Renamed to ...
22472 (vec_initv2sfsf): ... this.
22473 (vec_extractv2si): Renamed to ...
22474 (vec_extractv2sisi): ... this.
22475 (vec_initv2si): Renamed to ...
22476 (vec_initv2sisi): ... this.
22477 (vec_extractv4hi): Renamed to ...
22478 (vec_extractv4hihi): ... this.
22479 (vec_initv4hi): Renamed to ...
22480 (vec_initv4hihi): ... this.
22481 (vec_extractv8qi): Renamed to ...
22482 (vec_extractv8qiqi): ... this.
22483 (vec_initv8qi): Renamed to ...
22484 (vec_initv8qiqi): ... this.
22485 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22486 (vec_init<mode>): Renamed to ...
22487 (vec_init<mode><VEC_base_l>): ... this.
22488 (vec_extract<mode>): Renamed to ...
22489 (vec_extract<mode><VEC_base_l>): ... this.
22490 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22491 (vec_initv2sfsf): ... this.
22492 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22493 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22494 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22495 element mode after mode in gen_vec_init* calls.
22496 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22497 (vec_init<mode><Vel>): ... this.
22498 (vec_extract<mode>): Renamed to ...
22499 (vec_extract<mode><Vel>): ... this.
22500 * config/aarch64/iterators.md (Vel): New mode attribute.
22501 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22502 Add element mode after mode in gen_vec_extract* calls.
22503 * config/s390/vector.md (non_vec_l): New mode attribute.
22504 (vec_extract<mode>): Renamed to ...
22505 (vec_extract<mode><non_vec_l>): ... this.
22506 (vec_init<mode>): Renamed to ...
22507 (vec_init<mode><non_vec_l>): ... this.
22508 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22509 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22511 * config/arm/iterators.md (V_elem_l): New mode attribute.
22512 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22513 (vec_extract<mode><V_elem_l>): ... this.
22514 (vec_extractv2di): Renamed to ...
22515 (vec_extractv2didi): ... this.
22516 (vec_init<mode>): Renamed to ...
22517 (vec_init<mode><V_elem_l>): ... this.
22518 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22519 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22520 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22521 Add element mode after gen_vec_extract* calls.
22522 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22523 (vec_init<mode><unitmode>): ... this.
22524 (vec_extract<mode>): Renamed to ...
22525 (vec_extract<mode><unitmode>): ... this.
22526 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22527 (vec_init<mode><unitmode>): ... this.
22528 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22529 (vec_initv2sfsf): ... this.
22530 (vec_extractv2sf): Renamed to ...
22531 (vec_extractv2sfsf): ... this.
22532 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22533 Add element mode after gen_vec_extract* calls.
22534 * config/mips/mips.md (unitmode): New mode iterator.
22535 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22536 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22537 * config/spu/spu.md (inner_l): New mode attribute.
22538 (vec_init<mode>): Renamed to ...
22539 (vec_init<mode><inner_l>): ... this.
22540 (vec_extract<mode>): Renamed to ...
22541 (vec_extract<mode><inner_l>): ... this.
22542 * config/sparc/sparc.md (veltmode): New mode iterator.
22543 (vec_init<VMALL:mode>): Renamed to ...
22544 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22545 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22546 (vec_initv2sisi): ... this.
22547 (vec_initv2sf): Renamed to ...
22548 (vec_initv2sfsf): ... this.
22549 (vec_extractv2sf): Renamed to ...
22550 (vec_extractv2sfsf): ... this.
22551 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22552 (vec_init<mode>): Renamed to ...
22553 (vec_init<mode><VEC_base_l>): ... this.
22554 (vec_extract<mode>): Renamed to ...
22555 (vec_extract<mode><VEC_base_l>): ... this.
22556 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22557 (vec_initv2sfsf): ... this.
22558 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22559 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22560 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22561 gen_vec_init* calls.
22563 2017-08-01 Richard Biener <rguenther@suse.de>
22565 PR tree-optimization/81297
22566 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22567 TREE_OVERFLOW from INTEGER_CSTs.
22569 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22571 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22573 2017-07-31 Carl Love <cel@us.ibm.com>
22575 * config/rs6000/rs6000-c: Add support for built-in functions
22576 vector signed char vec_xl_be (signed long long, signed char *);
22577 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22578 vector signed int vec_xl_be (signed long long, signed int *);
22579 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22580 vector signed long long vec_xl_be (signed long long, signed long long *);
22581 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22582 vector signed short vec_xl_be (signed long long, signed short *);
22583 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22584 vector double vec_xl_be (signed long long, double *);
22585 vector float vec_xl_be (signed long long, float *);
22586 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22587 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22588 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22590 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22591 (altivec_expand_builtin): Add switch statement to call
22592 altivec_expand_xl_be for each builtin.
22593 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22594 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22595 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22596 __builtin_vsx_le_be_v16qi.
22597 * doc/extend.texi: Update the built-in documentation file for the
22598 new built-in functions.
22600 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22603 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22605 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22607 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22609 * config.gcc: Add z14.
22610 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22611 CPU model numbers for z13s and z14.
22612 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22614 * config/s390/s390-opts.h (enum processor_type): Rename
22615 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22616 * config/s390/s390.c (processor_table): Add field for CPU name to
22617 be passed to Binutils.
22618 (s390_asm_output_machine_for_arch): Use the new field in
22619 processor_table for Binutils.
22620 (s390_expand_builtin): Replace arch12 with z14.
22621 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22622 (s390_get_sched_attrmask): Likewise.
22623 (s390_get_unit_mask): Likewise.
22624 * config/s390/s390.opt: Add z14 to processor_type enum.
22626 2017-07-31 Martin Jambor <mjambor@suse.cz>
22629 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22630 regardless of optimization level.
22632 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22633 Martin Liska <mliska@suse.cz>
22635 * predict.def: Remove old comment and adjust probability.
22636 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22637 PREDICT statements.
22639 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22642 * config/i386/i386.c (ix86_function_naked): New function.
22643 (ix86_can_use_return_insn_p): Return false for naked functions.
22644 (ix86_expand_prologue): Skip prologue for naked functions.
22645 (ix86_expand_epilogue): Skip epilogue for naked functions
22646 and emit trap instruction.
22647 (ix86_warn_func_return): New function.
22648 (ix86_attribute_table): Add "naked" attribute specification.
22649 (TARGET_WARN_FUNC_RETURN): Define.
22650 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22652 2017-07-31 Martin Liska <mliska@suse.cz>
22654 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22655 (dump_gimple_bb_header): Always dump BB info.
22656 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22658 2017-07-31 Martin Liska <mliska@suse.cz>
22661 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22662 also with current_function_decl non-null equality.
22664 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22667 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22668 change type to the element type, instead add eltype variable and
22669 use it where we are interested in the element type.
22671 PR tree-optimization/81603
22672 * ipa-polymorphic-call.c
22673 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22674 offset arithmetic in offset_int, bail out if the resulting bit offset
22675 doesn't fit into shwi.
22677 2017-07-31 Martin Liska <mliska@suse.cz>
22679 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22680 (gimplify_save_expr): Fix comment.
22682 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22685 * config/i386/i386.c (ix86_function_arg): Update arguments for
22687 (ix86_compute_frame_layout): Set the initial stack offset to
22688 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22689 INCOMING_FRAME_SP_OFFSET.
22690 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22691 stack before exception handler returns.
22692 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22693 the 'ERROR_CODE' for exception handler.
22695 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22697 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22698 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22699 (ASM_OUTPUT_REG_POP): Ditto.
22700 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22701 instead of asm_fprintf to output pure string.
22703 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22705 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22706 to imported_module_or_decl hook.
22707 (debug_nothing_tree_tree_tree_bool): Remove.
22708 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22709 * debug.c (do_nothing_debug_hooks): Use
22710 debug_nothing_tree_tree_tree_bool_bool instead of
22711 debug_nothing_tree_tree_tree_bool.
22712 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22713 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22714 * sdbout.c (sdb_debug_hooks): Likewise.
22715 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22716 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22718 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22719 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22720 attribute, don't add anything.
22722 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22724 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22725 (fold_build2_stat_loc): Likewise.
22726 (fold_build3_stat_loc): Likewise.
22727 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22728 (fold_build1_loc): Remove macro.
22729 (fold_build2_loc): Likewise.
22730 (fold_build3_loc): Likewise.
22732 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22734 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22735 (gimple_build_debug_bind_source_stat): Likewise.
22736 * gimple.h (gimple_build_debug_bind): Remove macro.
22737 (gimple_build_debug_bind_source): Likewise.
22739 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22741 * bitmap.c (bitmap_alloc): Adjust.
22742 (bitmap_gc_alloc): Likewise.
22743 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22745 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22747 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22748 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22749 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22750 (bitmap_gc_alloc_stat): Likewise.
22751 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22753 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22755 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22756 * rtl.h (shallow_copy_rtx): Remove macro.
22758 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22760 * emit-rtl.c (gen_raw_REG): Adjust.
22761 * gengenrtl.c (gendef): Likewise.
22762 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22763 * rtl.h (rtx_alloc): Remove macro.
22765 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22767 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22768 (build_tree_list_stat): Likewise.
22769 * tree.h (build_tree_list): Remove macro.
22770 (build_tree_list_vec): Likewise.
22772 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22774 * tree.c (make_vector_stat): Remove _stat from name.
22775 (build_vector_stat): Likewise.
22776 * tree.h (make_vector_stat): Remove macro.
22777 (build_vector_stat): Likewise.
22779 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22781 * tree.h (build_var_debug_value): Remove prototype.
22783 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22785 * tree.c (tree_cons_stat): Remove _stat from name.
22786 * tree.h (tree_cons): Remove macro.
22788 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22790 * tree.c (build_vl_exp_stat): Remove _stat from name.
22791 * tree.h (build_vl_exp): Remove macro.
22793 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22795 * tree.c (build_decl_stat): Remove _stat from name.
22796 * tree.h (build_decl): Remove macro.
22798 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22800 * gimple.c (gimple_build_with_ops_stat): Adjust.
22801 (gimple_alloc_stat): Remove _stat from name.
22802 * gimple.h (gimple_alloc): Remove macro.
22804 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22806 * tree.c (make_tree_vec_stat): Remove _stat from name.
22807 (grow_tree_vec_stat): Likewise.
22808 * tree.h (make_tree_vec_stat): Adjust prototype.
22809 (grow_tree_vec_stat): Likewise.
22810 (make_tree_vec): Remove macro.
22811 (grow_tree_vec): Likewise.
22813 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22815 * fold-const.c (fold_build1_stat_loc): Adjust.
22816 (fold_build2_stat_loc): Likewise.
22817 (fold_build3_stat_loc): Likewise.
22818 * tree.c (build0_stat): Remove _stat from name.
22819 (build1_stat): Likewise.
22820 (build2_stat): Likewise.
22821 (build3_stat): Likewise.
22822 (build4_stat): Likewise.
22823 (build5_stat): Likewise.
22824 * tree.h (build1_loc): Remove macro, and rename _stat function
22826 (build2_loc): Likewise.
22827 (build3_loc): Likewise.
22828 (build4_loc): Likewise.
22829 (build5_loc): Likewise.
22831 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22833 * tree.c (make_int_cst_stat): Remove _stat from name.
22834 * tree.h (make_int_cst_stat): Adjust prototype.
22835 (make_int_cst): Remove macro.
22837 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22839 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22840 * tree.h (make_tree_binfo_stat): Adjust prototype.
22841 (make_tree_binfo): Remove.
22843 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22845 * tree.c (copy_node_stat): Rename to copy_node.
22846 (build_distinct_type_copy): Adjust.
22847 * tree.h (copy_node_stat): Adjust prototype.
22848 (copy_node): Remove macro.
22850 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22852 * tree.c (make_node_stat): rename to make_node.
22853 (build_tree_list_stat): Adjust.
22854 (build0_stat): Likewise.
22855 (build2_stat): Likewise.
22856 (build3_stat): Likewise.
22857 (build4_stat): Likewise.
22858 (build5_stat): Likewise.
22859 (build_decl_stat): Likewise.
22860 * tree.h (make_node_stat): Adjust prototype.
22861 (make_node): remove macro.
22863 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22865 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22866 (PPC_FEATURE2_SCV): Likewise.
22867 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22869 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22871 * config/aarch64/aarch64.c
22872 (aarch64_internal_mov_immediate): Add new special pattern.
22873 * config/aarch64/aarch64.md (*movdi_aarch64):
22874 Add reg/32bit const mov case.
22876 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22877 Richard Sandiford <richard.sandiford@linaro.org>
22879 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22880 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22881 Add integer and movi cases.
22882 (movi-split-hf-df-sf split, fp16): New.
22883 (enabled): Added TARGET_FP_F16INST.
22884 * config/aarch64/iterators.md (GPF_HF): New.
22885 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22887 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22889 * config/aarch64/aarch64.c
22890 (aarch64_simd_container_mode): Add prototype.
22891 (aarch64_expand_mov_immediate): Add HI support.
22892 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22893 (aarch64_can_const_movi_rtx_p): New.
22894 (aarch64_preferred_reload_class):
22895 Remove restrictions of using FP registers for certain SIMD operations.
22896 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22897 (aarch64_valid_floating_const): Add integer move validation.
22898 (aarch64_simd_imm_scalar_p): Remove.
22899 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22900 (aarch64_legitimate_constant_p): Expand list of supported cases.
22901 * config/aarch64/aarch64-protos.h
22902 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22903 (aarch64_reinterpret_float_as_int): New.
22904 (aarch64_simd_imm_scalar_p): Remove.
22905 * config/aarch64/constraints.md (Uvi): New.
22906 (Dd): Split into Ds and new Dd.
22907 * config/aarch64/aarch64.md (*movsi_aarch64):
22909 (*movdi_aarch64): Add SIMD mov case.
22911 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22913 * tree-predcom.c: (struct chain): Handle store-store chain in which
22914 stores for elimination only store loop invariant values.
22915 (execute_pred_commoning_chain): Ditto.
22916 (prepare_initializers_chain_store_elim): Ditto.
22917 (prepare_finalizers): Ditto.
22918 (is_inv_store_elimination_chain): New function.
22919 (initialize_root_vars_store_elim_1): New function.
22921 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22923 * tree-predcom.c: Revise general description of the pass.
22924 (enum chain_type): New enum type for store elimination.
22925 (struct chain): New field supporting store elimination.
22926 (struct component): Ditto.
22927 (dump_chain): Dump store-stores chain.
22928 (release_chain): Release resources.
22929 (split_data_refs_to_components): Compute and create component
22930 contains only stores for elimination.
22931 (get_chain_last_ref_at): New function.
22932 (make_invariant_chain): Initialization.
22933 (make_rooted_chain): Specify chain type in parameter and record it.
22934 (add_looparound_copies): Skip for store-stores chain.
22935 (determine_roots_comp): Compute type of chain and pass it to
22937 (initialize_root_vars_store_elim_2): New function.
22938 (finalize_eliminated_stores): New function.
22939 (remove_stmt): Handle store for elimination.
22940 (execute_pred_commoning_chain): Execute predictive commoning on
22941 store-store chains.
22942 (determine_unroll_factor): Skip unroll for store-stores chain.
22943 (prepare_initializers_chain_store_elim): New function.
22944 (prepare_initializers_chain): Hanlde store-store chain.
22945 (prepare_finalizers_chain, prepare_finalizers): New function.
22946 (tree_predictive_commoning_loop): Return integer value indicating
22947 if loop is unrolled or lcssa form is corrupted.
22948 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22950 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22952 * tree-predcom.c (initialize_root): Delete.
22953 (execute_pred_commoning_chain): Initialize root vars and replace
22954 reference of non-combined chain directly, rather than call above
22957 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22959 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22960 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22962 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22964 * tree-predcom.c (struct chain): New field init_seq.
22965 (release_chain): Release init_seq.
22966 (prepare_initializers_chain): Record intialization stmts in above
22968 (insert_init_seqs): New function.
22969 (tree_predictive_commoning_loop): Call insert_init_seqs.
22971 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22973 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22975 2017-07-28 Richard Biener <rguenther@suse.de>
22977 * match.pd: Remove superfluous :c.
22978 * genmatch.c (simplify::id): Add member.
22979 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22981 (current_id): New global.
22982 (dt_node::parent): Move from ...
22983 (dt_operand::parent): ... here. Add for_id member.
22984 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22985 (decision_tree::find_node): Relax order requirement when
22986 merging DT_TRUE nodes to ones inbetween the current simplify
22987 and the one we try to merge with. Add diagnostic whenever
22988 we need to enforce pattern order by not merging.
22989 (decision_tree::insert): Set current_id.
22990 (decision_tree::print_node): Dump parent node and for_id.
22991 (parser::last_id): Add member.
22992 (parser::push_simplify): Assign unique id.
22993 (parser::parser): Initialize last_id.
22995 2017-07-28 Martin Liska <mliska@suse.cz>
22998 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22999 gimple_build_debug_bind.
23001 2017-07-28 Richard Biener <rguenther@suse.de>
23003 PR tree-optimization/81502
23004 * match.pd: Add pattern combining BIT_INSERT_EXPR with
23006 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23008 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23009 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23010 for BIT_FIELD_REF args.
23011 * fold-const.c (make_bit_field_ref): Likewise.
23012 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23014 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23017 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23018 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23019 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23020 Or it into SANITIZER_UNDEFINED.
23021 * ubsan.c: Include gimple-fold.h and varasm.h.
23022 (ubsan_expand_ptr_ifn): New function.
23023 (instrument_pointer_overflow): New function.
23024 (maybe_instrument_pointer_overflow): New function.
23025 (instrument_object_size): Formatting fix.
23026 (pass_ubsan::execute): Call instrument_pointer_overflow
23027 and maybe_instrument_pointer_overflow.
23028 * internal-fn.c (expand_UBSAN_PTR): New function.
23029 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23030 * sanitizer.def (__ubsan_handle_pointer_overflow,
23031 __ubsan_handle_pointer_overflow_abort): New builtins.
23032 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23033 * internal-fn.def (UBSAN_PTR): New internal function.
23034 * opts.c (sanitizer_opts): Add pointer-overflow.
23035 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23036 * fold-const.c (build_range_check): Compute pointer range check in
23037 integral type if pointer arithmetics would be needed. Formatting
23040 2017-07-28 Martin Liska <mliska@suse.cz>
23043 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23044 parameters that are of a variable-length.
23046 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23048 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23050 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23051 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23052 (CRT_CALL_STATIC_FUNCTION): Likewise.
23053 (ASM_DEFAULT_SPEC): New define.
23054 (ASM_SPEC32): Likewise.
23055 (ASM_SPEC64): Likewise.
23056 (ASM_SPEC_COMMON): Likewise.
23057 (ASM_SPEC): Likewise.
23058 (INVALID_64BIT): Likewise.
23059 (LINK_OS_DEFAULT_SPEC): Likewise.
23060 (LINK_OS_SPEC32): Likewise.
23061 (LINK_OS_SPEC64): Likewise.
23062 (POWERPC_LINUX): Likewise.
23063 (PTRDIFF_TYPE): Likewise.
23064 (RESTORE_FP_PREFIX): Likewise.
23065 (RESTORE_FP_SUFFIX): Likewise.
23066 (SAVE_FP_PREFIX): Likewise.
23067 (SAVE_FP_SUFFIX): Likewise.
23068 (SIZE_TYPE): Likewise.
23069 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23070 (TARGET_64BIT): Likewise.
23071 (TARGET_64BIT): Likewise.
23072 (TARGET_AIX): Likewise.
23073 (WCHAR_TYPE_SIZE): Likewise.
23074 (WCHAR_TYPE): Undefine.
23075 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23076 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23077 (CPP_OS_RTEMS_SPEC): Delete.
23078 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23079 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23081 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23083 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23085 PR tree-optimization/81578
23086 * tree-parloops.c (build_new_reduction): Bail out if
23087 reduction_code isn't one of the standard OpenMP reductions.
23088 Move the details printing after that decision.
23090 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23092 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23093 related to reload_in_progress.
23094 (splat_input_operand): Likewise.
23095 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23097 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23099 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23100 (TARGET_INSTANTIATE_DECLS): Likewise.
23101 (legitimate_indexed_address_p): Delete reload_in_progress code.
23102 (rs6000_debug_legitimate_address_p): Likewise.
23103 (rs6000_eliminate_indexed_memrefs): Likewise.
23104 (rs6000_emit_le_vsx_store): Likewise.
23105 (rs6000_emit_move_si_sf_subreg): Likewise.
23106 (rs6000_emit_move): Likewise.
23107 (register_to_reg_type): Likewise.
23108 (rs6000_pre_atomic_barrier): Likewise.
23109 (rs6000_machopic_legitimize_pic_address): Likewise.
23110 (rs6000_allocate_stack_temp): Likewise.
23111 (rs6000_address_for_fpconvert): Likewise.
23112 (rs6000_address_for_altivec): Likewise.
23113 (rs6000_secondary_memory_needed_rtx): Delete function.
23114 (rs6000_check_sdmode): Likewise.
23115 (rs6000_alloc_sdmode_stack_slot): Likewise.
23116 (rs6000_instantiate_decls): Likewise.
23117 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23118 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23119 Delete reload_in_progress.
23120 (*vec_reload_and_plus_<mptrsize>): Likewise.
23121 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23122 (vsx_div_v2di): Likewise.
23123 (vsx_udiv_v2di): Likewise.
23125 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23127 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23128 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23129 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23130 (rs6000_debug_reg_global): Delete print of LRA status.
23131 (rs6000_option_override_internal): Delete dead LRA related code.
23132 (rs6000_lra_p): Delete function.
23133 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23135 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23137 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23138 * config/riscv/rtems.h: New file.
23140 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23141 Sudakshina Das <sudi.das@arm.com>
23143 * config/aarch64/aarch64.md
23144 (define_split for and<mode>3nr_compare): Move
23145 non aarch64_logical_operand to a register.
23146 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23147 register immediate operand to a register.
23148 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23150 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23152 PR middle-end/81564
23153 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23155 2017-07-27 Richard Biener <rguenther@suse.de>
23157 PR tree-optimization/81573
23158 PR tree-optimization/81494
23159 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23160 multi defuse cycle case.
23162 2017-07-27 Richard Biener <rguenther@suse.de>
23164 PR tree-optimization/81571
23165 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23168 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23170 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23171 earlier and only if MASK_FPU is set. Adjust formatting.
23173 2017-07-27 Martin Liska <mliska@suse.cz>
23175 * opt-functions.awk: Add validation of value of Init.
23176 * optc-gen.awk: Pass new argument.
23178 2017-07-27 Martin Liska <mliska@suse.cz>
23180 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23181 Fix wrong condition.
23183 2017-07-27 Martin Liska <mliska@suse.cz>
23185 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23186 BBs and edges seen by autoFDO.
23188 2017-07-27 Richard Biener <rguenther@suse.de>
23190 PR tree-optimization/81502
23191 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23192 with incompatible but same sized type.
23193 (execute_update_addresses_taken): Likewise.
23195 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23197 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23198 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23200 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23203 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23204 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23205 Change s_operand to memory_operand.
23207 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23209 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23210 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23211 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23212 Emit instructions rather than returning an expression. Handle TFmode
23213 and KFmode by casting to TImode.
23214 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23215 (rs6000_emit_le_vsx_store): Likewise.
23216 * config/rs6000/vsx.md (VSX_TI): New iterator.
23217 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23218 (*vsx_le_undo_permute_<mode>): Likewise.
23219 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23220 emit the split sequence.
23221 (*vsx_le_perm_store_<mode>): Likewise.
23223 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23225 PR tree-optimization/81555
23226 PR tree-optimization/81556
23227 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23228 if true, force CHANGED for the recursive invocation.
23229 (reassociate_bb): Remember original length of ops array, pass
23230 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23232 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23233 attributes for noipa attribute. For naked attribute use
23234 lookup_attribute first before lookup_attribute_spec.
23235 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23237 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23238 for functions with noipa attribute.
23239 (cgraph_externally_visible_p): Return true for functions with noipa
23241 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23242 for functions with noipa attribute.
23243 * doc/extend.texi: Document noipa function attribute.
23244 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23245 also for functions with noipa attribute.
23246 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23248 2017-07-26 Andrew Pinski <apinski@cavium.com>
23250 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23251 vec_unalign_load_cost and vec_unalign_store_cost.
23253 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23255 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23256 -mvsx-small-integer option.
23257 (ISA_3_0_MASKS_IEEE): Likewise.
23258 (OTHER_VSX_VECTOR_MASKS): Likewise.
23259 (POWERPC_MASKS): Likewise.
23260 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23261 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23262 code, only testing for DImode being allowed in non-VSX floating
23264 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23265 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23267 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23268 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23269 TARGET_P8_VECTOR test.
23270 (rs6000_secondary_reload_simple_move): Likewise.
23271 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23272 since TARGET_P9_VECTOR was already tested.
23273 (rs6000_opt_masks): Remove -mvsx-small-integer.
23274 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23275 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23277 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23278 test for TARGET_VEXTRACTUB was used, and that uses
23280 (p9 extract splitter): Likewise.
23281 (vsx_extract_<mode>_di_p9): Likewise.
23282 (vsx_extract_<mode>_store_p9): Likewise.
23283 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23284 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23285 the elimination of TARGET_VSX_SMALL_INTEGER.
23286 (vsx_extract_<mode>_p8): Likewise.
23287 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23288 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23289 (vsx_set_<mode>_p9): Likewise.
23290 (vsx_set_v4sf_p9): Likewise.
23291 (vsx_set_v4sf_p9_zero): Likewise.
23292 (vsx_insert_extract_v4sf_p9): Likewise.
23293 (vsx_insert_extract_v4sf_p9_2): Likewise.
23294 * config/rs6000/rs6000.md (sign extend splitter): Change
23295 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23296 (floatsi<mode>2_lfiwax_mem): Likewise.
23297 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23298 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23299 since a test for TARGET_P9_VECTOR was used.
23300 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23301 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23302 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23303 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23304 TARGET_P8_VECTOR test.
23305 (fix_trunc<mode>si2_stfiwx): Likewise.
23306 (fix_trunc<mode>si2_internal): Likewise.
23307 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23308 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23310 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23311 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23312 TARGET_P8_VECTOR test.
23313 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23314 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23315 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23317 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23318 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23319 since a test for TARGET_P9_VECTOR was used.
23320 (splitter for loading small constants): Likewise.
23322 2017-07-26 Andrew Pinski <apinski@cavium.com>
23324 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23327 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23330 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23331 (fp_valid_at): Likewise.
23333 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23335 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23336 (qdf24xx_addrcost_table): Likewise.
23337 (cortexa57_tunings): Update to use generic_branch_cost.
23338 (cortexa72_tunings): Likewise.
23339 (cortexa73_tunings): Likewise.
23340 (qdf24xx_tunings): Likewise.
23342 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23344 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23345 (thunderx2t99_branch_cost): Likewise.
23346 (cortexa35_tunings): Update to use generic_branch_cost.
23347 (cortexa53_tunings): Likewise.
23348 (cortexa57_tunings): Likewise.
23349 (cortexa72_tunings): Likewise.
23350 (cortexa73_tunings): Likewise.
23351 (thunderx2t99_tunings): Likewise.
23353 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23355 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23356 (sparc_option_override): Honour MASK_FSMULD.
23357 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23358 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23359 * config/sparc/sparc.opt (mfsmuld): New option.
23360 * doc/invoke.texi (mfsmuld): Document option.
23362 2017-07-26 Marek Polacek <polacek@redhat.com>
23364 PR middle-end/70992
23365 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23367 2017-07-26 Richard Biener <rguenther@suse.de>
23369 * gimple-match-head.c (do_valueize): Return OP if valueize
23371 (get_def): New helper to get at the def stmt of a SSA name
23372 if valueize allows.
23373 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23374 do_valueize to get at the def stmt.
23375 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23377 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23379 PR middle-end/46932
23380 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23382 2017-07-26 Martin Liska <mliska@suse.cz>
23385 * function.c (expand_function_start): Make expansion of
23386 nonlocal_goto_save_area after parm_birth_insn.
23388 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23390 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23391 from all CPU target flags enable members.
23393 2017-07-26 Richard Biener <rguenther@suse.de>
23395 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23396 (decision_tree::gen): Make 'type' const.
23397 (write_predicate): Likewise.
23399 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23401 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23402 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23403 (rs6000_option_override_internal): Likewise.
23404 (rs6000_expand_vector_set): Likewise.
23405 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23406 (TARGET_UPPER_REGS_SF): Likewise.
23407 (TARGET_UPPER_REGS_DI): Likewise.
23408 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23409 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23410 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23411 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23412 (Splitters for DI constants in Altivec registers): Likewise.
23413 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23414 (vsx_set_v4sf_p9): Likewise.
23415 (vsx_set_v4sf_p9_zero): Likewise.
23416 (vsx_insert_extract_v4sf_p9): Likewise.
23417 (vsx_insert_extract_v4sf_p9_2): Likewise.
23419 2017-07-25 Carl Love <cel@us.ibm.com>
23421 * doc/extend.texi: Update the built-in documentation file for the
23422 existing built-in functions
23423 vector signed char vec_cnttz (vector signed char);
23424 vector unsigned char vec_cnttz (vector unsigned char);
23425 vector signed short vec_cnttz (vector signed short);
23426 vector unsigned short vec_cnttz (vector unsigned short);
23427 vector signed int vec_cnttz (vector signed int);
23428 vector unsigned int vec_cnttz (vector unsigned int);
23429 vector signed long long vec_cnttz (vector signed long long);
23430 vector unsigned long long vec_cnttz (vector unsigned long long);
23432 2017-07-25 Andrew Pinski <apinski@cavium.com>
23434 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23435 accesses where the use is for the first operand of a BIT_INSERT.
23437 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23440 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23441 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23443 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23445 * config/i386/gstabs.h: Delete.
23446 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23448 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23450 * config/i386/i386.c (ix86_decompose_address): Do not check for
23451 register RTX when looking at index_reg or base_reg.
23452 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23454 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23456 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23457 to update EH info here.
23459 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23461 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23463 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23465 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23467 2017-07-25 Torsten Duwe <duwe@suse.de>
23469 * common.opt: Introduce -fpatchable-function-entry
23470 command line option, and its variables function_entry_patch_area_size
23471 and function_entry_patch_area_start.
23472 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23473 including a two-value parser.
23474 * target.def (print_patchable_function_entry): New target hook.
23475 * targhooks.h (default_print_patchable_function_entry): New function.
23476 * targhooks.c (default_print_patchable_function_entry): Likewise.
23477 * toplev.c (process_options): Switch off IPA-RA if
23478 patchable function entries are being generated.
23479 * varasm.c (assemble_start_function): Look at the
23480 patchable-function-entry command line switch and current
23481 function attributes and maybe generate NOP instructions by
23482 calling the print_patchable_function_entry hook.
23483 * doc/extend.texi: Document patchable_function_entry attribute.
23484 * doc/invoke.texi: Document -fpatchable_function_entry
23485 command line option.
23486 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23488 * doc/tm.texi: Re-generate.
23490 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23493 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23494 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23496 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23498 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23500 2017-07-25 Richard Biener <rguenther@suse.de>
23502 PR tree-optimization/81455
23503 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23504 not walk in cycles when looking for guards.
23506 2017-07-25 Richard Biener <rguenther@suse.de>
23508 PR tree-optimization/81529
23509 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23510 when optimizing backedge uses.
23512 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23514 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23516 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23517 to dl_section_ref. On AIX, append an expression to subtract
23518 the size of the section length to dl_section_ref.
23520 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23522 * configure.ac: If any of the config.* scripts fail, exit 1.
23523 * configure: Regenerate.
23525 2017-07-25 Richard Biener <rguenther@suse.de>
23527 PR middle-end/81546
23528 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23529 of immediate uses, be more verbose on errors.
23531 2017-07-25 Richard Biener <rguenther@suse.de>
23533 PR tree-optimization/81510
23534 * tree-vect-loop.c (vect_is_simple_reduction): When the
23535 reduction stmt is not inside the loop bail out.
23537 2017-07-25 Richard Biener <rguenther@suse.de>
23539 PR tree-optimization/81303
23540 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23541 profitability check against LOOP_VINFO_NITERSM1.
23543 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23545 * domwalk.c (cmp_bb_postorder): Simplify.
23546 (sort_bbs_postorder): New function. Use it...
23547 (dom_walker::walk): ...here to optimize common cases.
23549 2017-07-25 Martin Liska <mliska@suse.cz>
23552 * ipa-visibility.c (function_and_variable_visibility): Make the
23553 redirection just on target that supports aliasing.
23554 Fix GNU coding style.
23556 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23559 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23561 * config/arm/rtems.h: Likewise.
23562 * config/bfin/rtems.h: Likewise.
23563 * config/i386/rtemself.h: Likewise.
23564 * config/lm32/rtems.h: Likewise.
23565 * config/m32c/rtems.h: Likewise.
23566 * config/m68k/rtemself.h: Likewise.
23567 * config/microblaze/rtems.h: Likewise.
23568 * config/mips/rtems.h: Likewise.
23569 * config/moxie/rtems.h: Likewise.
23570 * config/nios2/rtems.h: Likewise.
23571 * config/powerpcspe/rtems.h: Likewise.
23572 * config/rs6000/rtems.h: Likewise.
23573 * config/rtems.h: Likewise.
23574 * config/sh/rtems.h: Likewise.
23575 * config/sh/rtemself.h: Likewise.
23576 * config/sparc/rtemself.h: Likewise.
23578 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23581 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23582 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23583 * tree-ssa-structalias.c (alias_get_name): Same.
23585 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23588 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23589 instructions if no du chain is found.
23591 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23593 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23595 2017-07-25 Richard Biener <rguenther@suse.de>
23597 PR middle-end/81505
23598 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23601 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23603 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23604 upper-regs options.
23605 (ISA_2_7_MASKS_SERVER): Likewise.
23606 (ISA_3_0_MASKS_IEEE): Likewise.
23607 (OTHER_P8_VECTOR_MASKS): Likewise.
23608 (OTHER_VSX_VECTOR_MASKS): Likewise.
23609 (POWERPC_MASKS): Likewise.
23610 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23611 duplicate list of options.
23612 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23613 explicit -mupper-regs options.
23614 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23615 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23616 alias for -mupper-regs-df.
23617 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23618 (rs6000_init_hard_regno_mode_ok): Likewise.
23619 (rs6000_option_override_internal): Likewise.
23620 (rs6000_opt_masks): Likewise.
23621 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23622 options in terms of whether -mvsx or -mpower8-vector was used.
23623 (TARGET_UPPER_REGS_DI): Likewise.
23624 (TARGET_UPPER_REGS_SF): Likewise.
23625 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23626 -mupper-regs-* options.
23628 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23630 * passes.c (emergency_dump_function): Print some empty lines and a
23631 header before the RTL dump.
23633 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23635 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23637 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23640 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23641 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23643 2017-07-24 Carl Love <cel@us.ibm.com>
23645 * config/rs6000/rs6000-c.c: Add support for built-in functions
23646 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23647 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23648 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23649 vec_extract_fp_from_shortl): Add defines for the two builtins.
23650 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23651 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23653 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23654 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23655 * doc/extend.texi: Update the built-in documentation file for the
23656 new built-in function.
23658 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23661 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23663 * doc/generic.texi: Likewise.
23664 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23665 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23667 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23669 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23670 (aarch64_mls_elt_merge<mode>): Likewise.
23672 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23674 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23675 having __cxa_atexit.
23677 2017-07-23 Michael Collison <michael.collison@arm.com>
23679 * config/arm/arm.c (arm_option_override): Deprecate
23680 use of -mstructure-size-boundary.
23681 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23682 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23684 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23687 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23688 Reduce cost estimate for direct moves.
23690 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23693 * config/i386/i386.c (ix86_option_override_internal): Disable
23694 BMI, BMI2 and TBM instructions for -m16.
23696 2017-07-21 Carl Love <cel@us.ibm.com>
23698 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23699 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23700 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23701 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23702 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23703 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23704 VMULOSW): New enum "unspec" values.
23705 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23706 altivec_vmulosw): New patterns.
23707 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23708 VMULOSW): Add definitions.
23710 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23712 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23713 (qdf24xx): Likewise.
23714 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23715 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23716 (AARCH64_FL_V8_1): Renumber.
23717 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23718 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23719 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23720 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23721 rdma to feature modifiers list.
23723 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23725 PR middle-end/56727
23726 * ipa-visibility (function_and_variable_visibility): Convert
23727 recursive PLT call to direct call if appropriate.
23729 2017-07-21 Andrew Pinski <apinski@cavium.com>
23731 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23732 operand 1 to see if the types precision matches.
23733 * fold-const.c (operand_equal_p): Likewise.
23735 2017-07-21 Richard Biener <rguenther@suse.de>
23737 PR tree-optimization/81303
23738 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23739 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23740 (vect_peeling_hash_get_lowest_cost): Adjust.
23741 (vect_enhance_data_refs_alignment): Likewise. Use
23742 vect_get_peeling_costs_all_drs to compute the penalty for no
23743 peeling to match up costs.
23745 2017-07-21 Richard Biener <rguenther@suse.de>
23747 PR tree-optimization/81500
23748 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23749 we didn't identify a reduction path.
23751 2017-07-21 Tom de Vries <tom@codesourcery.com>
23752 Cesar Philippidis <cesar@codesourcery.com>
23754 PR gcov-profile/81442
23755 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23758 2017-07-21 Tom de Vries <tom@codesourcery.com>
23761 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23763 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23764 nvptx_override_options_after_change.
23766 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23768 * dwarf2out.c (output_file_names): Avoid double testing for
23769 dwarf_version >= 5.
23771 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23773 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23775 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23777 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23779 (try_crossjump_to_edge): Do not punt on partitioned functions.
23781 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23783 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23784 Put all BBs reachable only via paths crossing cold region to cold
23786 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23788 2016-07-21 Richard Biener <rguenther@suse.de>
23790 PR tree-optimization/81303
23791 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23792 into account prologue and epilogue iterations when raising
23793 min_profitable_iters to sth at least covering one vector iteration.
23795 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23797 * config/arm/arm.c (arm_test_cpu_arch_dat):
23800 2017-07-20 Nathan Sidwell <nathan@acm.org>
23802 Remove TYPE_METHODS.
23803 * tree.h (TYPE_METHODS): Delete.
23804 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23805 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23806 (dbxout_type_methods): Scan TYPE_FIELDS.
23807 (dbxout_type): Don't check TYPE_METHODS here.
23808 * function.c (use_register_for_decl): Always ignore register for
23809 class types when not optimizing.
23810 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23811 * tree.c (free_lang_data_in_type): Stitch out member functions and
23812 templates from TYPE_FIELDS.
23813 (build_distinct_type_copy, verify_type_variant,
23814 verify_type): Member fns are on TYPE_FIELDS.
23815 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23816 * tree-pretty-print.c (dump_generic_node): Likewise.
23818 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23821 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23822 V2TImode and V4TImode.
23823 (ix86_expand_vector_extract): Likewise.
23824 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23825 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23826 (ssescalarmode): Handle V4TImode and V2TImode.
23827 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23828 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23829 (VEXTRACTI128_MODE): New mode iterator.
23830 (splitter for *vec_extractv?ti first element): New.
23831 (VEC_INIT_MODE): New mode iterator.
23832 (vec_init<mode>): Consolidate 3 expanders into one using
23833 VEC_INIT_MODE mode iterator.
23835 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23837 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23838 non_spilled_static_chain_regno_p.
23840 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23842 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23844 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23846 * bb-reorder.c (connect_traces): Allow copying of blocks within
23849 2017-07-20 Richard Biener <rguenther@suse.de>
23851 * gimple.h (gimple_phi_result): Add gphi * overload.
23852 (gimple_phi_result_ptr): Likewise.
23853 (gimple_phi_arg): Likewise. Adjust index assert to only
23854 allow actual argument accesses rather than all slots available
23856 (gimple_phi_arg_def): Add gphi * overload.
23857 * tree-phinodes.c (make_phi_node): Initialize only actual
23859 (resize_phi_node): Clear memory not covered by old node,
23860 do not initialize excess argument slots.
23861 (reserve_phi_args_for_new_edge): Initialize new argument slot
23864 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23866 PR tree-optimization/81388
23868 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23870 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23871 by removing computation of may_be_zero.
23873 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23874 Tom de Vries <tom@codesourcery.com>
23876 PR middle-end/81030
23877 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23878 when gimple level profile disagrees with what RTL expander did.
23880 2017-07-20 Richard Biener <rguenther@suse.de>
23882 PR tree-optimization/61171
23883 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23884 (vect_analyze_stmt): Add slp instance parameter.
23885 (vectorizable_reduction): Likewise.
23886 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23887 (vect_is_simple_reduction): Deal with chains not detected
23888 as SLP reduction chain, specifically not properly associated
23889 chains containing a mix of plus/minus.
23890 (get_reduction_op): Remove.
23891 (get_initial_defs_for_reduction): Simplify, pass in whether
23892 this is a reduction chain, pass in the SLP node for the PHIs.
23893 (vect_create_epilog_for_reduction): Get the SLP instance as
23895 (vectorizable_reduction): Get the SLP instance as arg.
23896 During analysis remember the SLP node with the PHIs in the
23897 instance. Simplify getting at the vectorized reduction PHIs.
23898 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23899 through SLP instance.
23900 (vect_slp_analyze_operations): Likewise.
23901 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23902 (vect_transform_stmt): Likewise.
23904 2017-07-20 Tom de Vries <tom@codesourcery.com>
23906 PR tree-optimization/81489
23907 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23908 read of phi arg location to before loop that modifies phi.
23910 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23912 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23915 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23917 PR middle-end/81331
23918 * except.c (execute): Fix ordering issue.
23920 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23922 PR rtl-optimization/81423
23923 * combine.c (make_compound_operation_int): Don't try to optimize
23924 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23926 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23928 PR rtl-optimization/81423
23929 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23930 with a constant that is -1 in the truncated to mode.
23932 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23934 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23935 (determine_unlikely_bbs): ... here.
23936 * predict.h (propagate_unlikely_bbs_forward): Declare.
23937 * cfgexpand.c (pass_expand::execute): Use it.
23938 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23940 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23941 propagate_unlikely_bbs_forward.
23943 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23945 PR middle-end/81331
23946 * except.c (maybe_add_nop_after_section_switch): New function.
23949 2017-07-19 Tom de Vries <tom@codesourcery.com>
23951 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23953 2017-07-19 Tom de Vries <tom@codesourcery.com>
23955 * gimple.h (gimple_phi_arg): Make assert more strict.
23957 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23959 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23960 * config/rs6000/mmintrin.h: New file.
23961 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23963 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23965 PR tree-optimization/81346
23966 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23968 2017-07-19 Tom de Vries <tom@codesourcery.com>
23970 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23972 2017-07-19 Tom de Vries <tom@codesourcery.com>
23974 * config/nvptx/nvptx-modes.def: Add V2DImode.
23975 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23976 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23977 (nvptx_output_mov_insn): Handle lack of mov.b128.
23978 (nvptx_print_operand): Handle 'H' and 'L' codes.
23979 (nvptx_vector_mode_supported): Allow V2DImode.
23980 (nvptx_preferred_simd_mode): New function.
23981 (nvptx_data_alignment): New function.
23982 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23983 nvptx_preferred_simd_mode.
23984 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23986 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23988 2017-07-19 Tom de Vries <tom@codesourcery.com>
23990 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23991 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23992 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23993 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23994 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23995 (mov<VECIM>_insn): New define_insn.
23996 (define_expand "mov<VECIM>): New define_expand.
23998 2017-07-19 Tom de Vries <tom@codesourcery.com>
24000 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
24002 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24004 PR tree-optimization/81346
24005 * fold-const.h (fold_div_compare, range_check_type): Declare.
24006 * fold-const.c (range_check_type): New function.
24007 (build_range_check): Use range_check_type.
24008 (fold_div_compare): No longer static, rewritten into
24009 a match.pd helper function.
24010 (fold_comparison): Don't call fold_div_compare here.
24011 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24012 as helper function.
24014 2017-07-19 Nathan Sidwell <nathan@acm.org>
24016 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24017 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24018 * tree.c (find_decls_types_r, verify_type): Use
24019 TYPE_{MIN,MAX}_VALUE_RAW.
24020 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24021 (hash_tree): Likewise.
24022 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24024 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24027 2017-07-18 Tom de Vries <tom@codesourcery.com>
24029 PR middle-end/81464
24030 * omp-expand.c (expand_omp_for_static_chunk): Handle
24031 equal-argument loop exit phi.
24033 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24036 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24037 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24038 operand 2 predicate.
24039 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24040 operand 2 predicate.
24041 (ror,rol -> rorx splitters): Use const_int_operand as
24042 operand 2 predicate.
24044 2017-06-18 Richard Biener <rguenther@suse.de>
24046 PR tree-optimization/81410
24047 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24048 the gap in the ! slp_perm SLP case after each group.
24050 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24052 PR middle-end/81463
24053 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24056 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24058 PR middle-end/81462
24059 * predict.c (set_even_probabilities): Cleanup; do not affect
24060 probabilities that are already known.
24061 (combine_predictions_for_bb): Call even when count is set.
24063 2017-07-18 Nathan Sidwell <nathan@acm.org>
24065 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24068 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24071 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24072 optimization for loop niter analysis.
24074 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24077 * config/avr/avr.c (avr_optimize_casesi): Don't use
24078 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24080 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24082 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24083 body_cost_vec from _vect_peel_extended_info.
24084 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24085 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24088 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24090 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24092 2017-07-18 Richard Biener <rguenther@suse.de>
24094 PR tree-optimization/80620
24095 PR tree-optimization/81403
24096 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24097 info when re-using a VN table entry.
24099 2017-07-18 Richard Biener <rguenther@suse.de>
24101 PR tree-optimization/81418
24102 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24103 vectype_in. Verify that with lane-reducing reduction operations
24104 we have a single def-use cycle.
24106 2017-07-17 Carl Love <cel@us.ibm.com>
24108 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24110 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24111 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24112 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24113 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24114 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24115 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24116 VMULOSW): New enum "unspec" values.
24117 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24118 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24119 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24120 altivec_vmulosw): New patterns.
24121 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24122 VMULOSW): Add definitions.
24124 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24126 * config/alpha/alpha.c: Include predict.h.
24128 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24130 * tree-vrp.c (compare_assert_loc): Fix comparison function
24131 to return predictable results.
24133 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24135 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24137 (subdi3): Likewise.
24138 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24139 * doc/invoke.texi (mexpand-adddi): Update text.
24141 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24143 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24144 that also clobbers the CC register. The old expand code is moved
24146 (*arc_clzsi2): ... here.
24147 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24148 the CC register. The old expand code is moved to ...
24149 (arc_ctzsi2): ... here.
24151 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24153 * config/arc/arc.opt (mindexed-loads): Use initial value
24154 TARGET_INDEXED_LOADS_DEFAULT.
24155 (mauto-modify-reg): Use initial value
24156 TARGET_AUTO_MODIFY_REG_DEFAULT.
24157 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24158 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24159 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24160 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24162 2017-07-17 Martin Liska <mliska@suse.cz>
24165 * opts.c (finish_options): Do not allow -fgnu-tm
24166 w/ -fsanitize={kernel-,}address. Say sorry.
24168 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24171 * tree-loop-distribution.c (classify_partition): Only assert on
24172 numer of iterations.
24173 (merge_dep_scc_partitions): Delete prameter. Update function call.
24174 (distribute_loop): Remove code handling loop with unknown niters.
24175 (pass_loop_distribution::execute): Skip loop with unknown niters.
24177 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24180 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24181 function sort_partitions_by_post_order.
24183 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24185 PR tree-optimization/81374
24186 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24187 the max index of basic blocks, rather than number of basic blocks.
24189 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24191 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24193 (arc_legitimate_pic_operand_p): Likewise.
24194 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24196 (arc_needs_pcl_p): Likewise.
24197 (arc_legitimate_pc_offset_p): Likewise.
24198 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24199 function is also used in constrains.md.
24200 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24201 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24202 PLUS. Only return true/false in known cases, otherwise assert.
24203 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24204 is already called in arc_legitimate_constant_p.
24205 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24207 (LEGITIMATE_PIC_OPERAND_P): Use
24208 arc_raw_symbolic_reference_mentioned_p function.
24209 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24214 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24215 Andrew Burgess <andrew.burgess@embecosm.com>
24217 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24218 (arc_return_address_register): New function.
24219 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24220 (arc_handle_fndecl_attribute): Add naked attribute.
24221 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24222 (TARGET_WARN_FUNC_RETURN): Likewise.
24223 (arc_allocate_stack_slots_for_args): New function.
24224 (arc_warn_func_return): Likewise.
24225 (machine_function): Change type fn_type.
24226 (arc_compute_function_type): Consider new naked function type,
24227 change function return type.
24228 (arc_must_save_register): Adapt to handle new
24229 arc_compute_function_type's return type.
24230 (arc_expand_prologue): Likewise.
24231 (arc_expand_epilogue): Likewise.
24232 (arc_return_address_regs): Delete.
24233 (arc_return_address_register): New function.
24234 (arc_epilogue_uses): Use above function.
24235 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24236 (arc_function_type): Change encoding, add naked type.
24237 (ARC_INTERRUPT_P): Change to handle the new encoding.
24238 (ARC_FAST_INTERRUPT_P): Likewise.
24239 (ARC_NORMAL_P): Define.
24240 (ARC_NAKED_P): Likewise.
24241 (arc_compute_function_type): Delete prototype.
24242 * config/arc/arc.md (in_ret_delay_slot): Use
24243 arc_return_address_register function.
24244 (simple_return): Likewise.
24245 (p_return_i): Likewise.
24247 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24249 PR tree-optimization/81428
24250 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24251 can't be built for those types.
24253 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24255 Remove stuff dead since r239246.
24257 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24258 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24259 (avr_inform_devices): Remove dead stuff.
24261 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24263 * config/arm/arm_neon.h: Fix softp typo.
24265 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24267 PR tree-optimization/81365
24268 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24269 aggregate moves onto bb predecessor edges, make sure there are no
24270 loads that could alias the lhs in between the start of bb and the
24273 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24276 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24277 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24278 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24280 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24282 PR tree-optimization/81396
24283 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24284 (init_symbolic_number): Initialize it to 1.
24285 (perform_symbolic_merge): Add n_ops from both operands into the new
24287 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24288 without base_addr as useless if they need more than one operation.
24289 (bswap_replace): Handle !bswap case for NULL base_addr.
24291 2017-07-17 Tom de Vries <tom@codesourcery.com>
24294 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24297 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24299 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24300 conditional builtin define __FIX_LEON3FT_B2BST.
24302 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24304 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24305 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24308 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24310 PR rtl-optimization/81424
24311 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24312 to remove potential trapping from operands if -fnon-call-exceptions.
24314 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24316 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24317 profile_proability for scalling.
24318 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24320 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24322 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24324 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24326 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24327 fixpoint arithmetics.
24329 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24331 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24332 fixpoint arithmetics.
24334 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24336 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24337 fixpoint arithmetics.
24339 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24341 * profile-count.h (profile_probability::from_reg_br_prob_note,
24342 profile_probability::to_reg_br_prob_note): New functions.
24343 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24344 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24345 * predict.c (probability_reliable_p): Update.
24346 (edge_probability_reliable_p): Update.
24347 (br_prob_note_reliable_p): Update.
24348 (invert_br_probabilities): Update.
24349 (add_reg_br_prob_note): New function.
24350 (combine_predictions_for_insn): Update.
24351 * asan.c (asan_clear_shadow): Update.
24352 * cfgbuild.c (compute_outgoing_frequencies): Update.
24353 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24354 (update_br_prob_note): Update.
24355 (rtl_verify_edges): Update.
24356 (purge_dead_edges): Update.
24357 (fixup_reorder_chain): Update.
24358 * emit-rtl.c (try_split): Update.
24359 * ifcvt.c (cond_exec_process_insns): Update.
24360 (cond_exec_process_if_block): Update.
24361 (dead_or_predicable): Update.
24362 * internal-fn.c (expand_addsub_overflow): Update.
24363 (expand_neg_overflow): Update.
24364 (expand_mul_overflow): Update.
24365 * loop-doloop.c (doloop_modify): Update.
24366 * loop-unroll.c (compare_and_jump_seq): Update.
24367 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24368 * predict.h: Update.
24369 * reorg.c (mostly_true_jump): Update.
24371 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24372 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24373 * config/arc/arc.c: (emit_unlikely_jump): Update.
24374 * config/arm/arm.c: (emit_unlikely_jump): Update.
24375 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24376 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24377 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24378 (ix86_print_operand): Update.
24379 (ix86_split_fp_branch): Update.
24380 (predict_jump): Update.
24381 * config/ia64/ia64.c (ia64_print_operand): Update.
24382 * config/mmix/mmix.c (mmix_print_operand): Update.
24383 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24384 (rs6000_expand_split_stack_prologue): Update.
24385 * config/rs6000/rs6000.c: Update.
24386 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24387 (s390_expand_vec_movstr): Update.
24388 (s390_expand_cs_tdsi): Update.
24389 (s390_expand_split_stack_prologue): Update.
24390 * config/sh/sh.c (sh_print_operand): Update.
24391 (expand_cbranchsi4): Update.
24392 (expand_cbranchdi4): Update.
24393 * config/sparc/sparc.c (output_v9branch): Update.
24394 * config/spu/spu.c (get_branch_target): Update.
24395 (ea_load_store_inline): Update.
24396 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24397 * config/tilepro/tilepro.c: Update.
24399 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24401 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24402 (gimplify_save_expr): Likewise.
24404 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24406 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24408 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24410 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24412 * timevar.def (TV_IPA_FNSUMMARY): Define.
24414 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24416 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24417 to back store errata sensitive sequence from being generated.
24418 (sqrtdf2_fix): Likewise.
24420 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24422 * tree-ssa-threadupdate.c (compute_path_counts,
24423 update_joiner_offpath_counts): Use profile_probability.
24425 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24428 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24430 * config/arm/arm-c.c (arm_cpu_builtins): Define
24431 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24433 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24435 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24436 array entries to represent __ieee128 versions of the
24437 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24438 scalar_extract_sig, and scalar_insert_exp built-in functions.
24439 (altivec_resolve_overloaded_builtin): Add special case handling
24440 for the __builtin_scalar_insert_exp function, as represented by
24441 the P9V_BUILTIN_VEC_VSIEDP constant.
24442 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24443 exponent support for __ieee128 argument.
24444 (VSESQP): Add scalar extract signature support for __ieee128
24446 (VSTDCNQP): Add scalar test negative support for __ieee128
24448 (VSIEQP): Add scalar insert exponent support for __int128 argument
24449 with __ieee128 result.
24450 (VSIEQPF): Add scalar insert exponent support for __ieee128
24451 argument with __ieee128 result.
24452 (VSTDCQP): Add scalar test data class support for __ieee128
24454 (VSTDCNQP): Add overload support for scalar test negative with
24455 __ieee128 argument.
24456 (VSTDCQP): Add overload support for scalar test data class
24457 __ieee128 argument.
24458 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24459 UNSPEC_VSX_SXSIGDP.
24460 (UNSPEC_VSX_SIEXPQP): New constant.
24461 (xsxexpqp): New insn for VSX scalar extract exponent quad
24463 (xsxsigqp): New insn for VSX scalar extract significand quad
24465 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24466 precision with floating point argument.
24467 (xststdcqp): New expand for VSX scalar test data class quad
24469 (xststdcnegqp): New expand for VSX scalar test negative quad
24471 (xststdcqp): New insn to match expansions for VSX scalar test data
24472 class quad precision and VSX scalar test negative quad precision.
24473 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24474 special case operand checking to enforce that second operand of
24475 VSX scalar test data class with quad precision argument is a 7-bit
24477 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24478 prototypes and descriptions of __ieee128 versions of
24479 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24480 scalar_test_data_class, and scalar_test_neg built-in functions.
24482 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24484 PR tree-optimization/81162
24485 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24486 replace a negate with an add.
24488 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24490 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24492 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24494 * config/arm/arm-c.c (arm_cpu_builtins): Define
24495 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24497 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24499 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24500 (armv8-r): Set ARM Cortex-R52 as default CPU.
24501 * config/arm/arm-tables.opt: Regenerate.
24502 * config/arm/arm-tune.md: Regenerate.
24503 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24505 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24506 extension for -mcpu=cortex-r52.
24508 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24510 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24511 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24512 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24513 (fp-armv8): Define it as FP_ARMv8 only.
24514 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24515 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24517 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24518 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24519 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24520 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24521 than TARGET_FPU_ARMV8.
24522 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24523 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24524 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24526 * config/arm/neon.md (neon_vrint): Likewise.
24527 (neon_vcvt): Likewise.
24528 (neon_<fmaxmin_op><mode>): Likewise.
24529 (<fmaxmin><mode>3): Likewise.
24530 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24531 * config/arm/predicates.md (arm_cond_move_operator): Check against
24532 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24534 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24536 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24537 to top of function.
24539 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24541 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24542 loop in comment with memset.
24544 2017-07-14 Martin Liska <mliska@suse.cz>
24546 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24547 * dwarf2out.c (is_java): Remove the function.
24548 (output_pubname): Remove usage of the function.
24549 (lower_bound_default): Remove usage of DW_LANG_Java.
24550 (gen_compile_unit_die): Likewise.
24551 * gcc.c: Remove compiler defaults for .java and .zip files.
24552 * gimple-expr.c (remove_suffix): Change as there's no longer
24553 extension than 4-letter one.
24554 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24555 (gimplify_save_expr): Likewise.
24556 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24557 as it's possible even for other languages than Java.
24558 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24559 * lto-opts.c (lto_write_options): Remove reference to Java.
24560 * opts.c (strip_off_ending): Update file extension handling.
24561 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24562 * tree-eh.c (lower_resx): Likewise.
24563 * tree.c (free_lang_data_in_type): Remove dead code.
24564 (find_decls_types_r): Likewise.
24565 (build_common_builtin_nodes): Remove Java from a comment.
24566 (verify_type): Remove dead code.
24567 * varasm.c (assemble_external): Remove Java from a comment.
24569 2017-07-14 Martin Liska <mliska@suse.cz>
24571 * opts.c (finish_options): Add quotes.
24572 (common_handle_option): Likewise.
24574 2017-07-14 Martin Liska <mliska@suse.cz>
24576 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24577 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24578 Remove N_SO_PASCAL.
24579 * dwarf2out.c (lower_bound_default): Do not handle
24581 (gen_compile_unit_die): Likewise.
24582 * gcc.c: Remove default extension binding for GNU Pascal.
24583 * stmt.c: Remove Pascal language from a comment.
24584 * xcoffout.c: Likewise.
24586 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24589 * diagnostic-show-locus.c (fixit_cmp): New function.
24590 (layout::layout): Sort m_fixit_hints.
24591 (column_range::column_range): Assert that the values are valid.
24592 (struct char_span): New struct.
24593 (correction::overwrite): New method.
24594 (struct source_line): New struct.
24595 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24596 calls in terms of classes source_line and char_span, and
24597 correction::overwrite.
24598 (selftest::test_overlapped_fixit_printing_2): New function.
24599 (selftest::diagnostic_show_locus_c_tests): Call it.
24601 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24603 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24604 early if there is no lhs.
24606 2017-07-13 Martin Liska <mliska@suse.cz>
24608 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24609 (gen_reference_type_die): Likewise.
24610 * stor-layout.c: Remove Pascal-related comment.
24612 2017-07-13 Martin Liska <mliska@suse.cz>
24614 * opts.c (finish_options): Add quotes to error messages.
24615 (parse_sanitizer_options): Likewise.
24617 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24619 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24621 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24623 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24625 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24627 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24629 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24631 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24634 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24635 provides the hardware capability bits, define the macro
24636 __BUILTIN_CPU_SUPPORTS__.
24637 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24638 if GLIBC does not provide the hardware capability bits. Add a
24639 gcc_unreachable call if the built-in cpu function is neither
24640 __builtin_cpu_is nor __builtin_cpu_supports.
24641 (rs6000_get_function_versions_dispatcher): Change the warning
24642 that an old GLIBC is used which does not export the capability
24643 bits to be an error.
24644 * doc/extend.texi (target_clones attribute): Document the
24645 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24646 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24647 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24648 the macros defined by GCC if the newer GLIBC is available.
24650 2017-07-12 Jeff Law <law@redhat.com>
24652 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24653 remaining includes slightly.
24654 * config/riscv/riscv-builtins.c: Include profile-count.h.
24656 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24659 * config/avr/avr.c (avr_set_current_function): In diagnostic
24660 messages: Quote keywords and (parts of) identifiers.
24661 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24664 2017-07-12 Carl Love <cel@us.ibm.com>
24666 * config/rs6000/rs6000-c.c: Add support for built-in functions
24667 vector bool char vec_revb (vector bool char);
24668 vector bool short vec_revb (vector short char);
24669 vector bool int vec_revb (vector bool int);
24670 vector bool long long vec_revb (vector bool long long);
24671 * doc/extend.texi: Update the built-in documentation file for the
24672 new built-in functions.
24674 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24676 * config/s390/s390.md: Remove movcc splitter.
24678 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24680 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24681 load/store on condition.
24683 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24686 * config/avr/avr.c (avr_encode_section_info)
24687 [progmem && !TREE_READONLY]: Error if progmem object needs
24690 2017-07-11 Michael Collison <michael.collison@arm.com>
24692 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24695 2017-07-11 Carl Love <cel@us.ibm.com>
24697 * config/rs6000/rs6000-c.c: Add support for builtins
24698 vector unsigned int vec_parity_lsbb (vector signed int);
24699 vector unsigned int vec_parity_lsbb (vector unsigned int);
24700 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24701 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24702 vector unsigned long long vec_parity_lsbb (vector signed long long);
24703 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24704 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24705 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24706 * doc/extend.texi: Update the built-in documentation file for the
24707 new built-in functions.
24709 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24711 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24712 (layout::m_primary_loc): New field.
24713 (layout::layout): Initialize new field. Move location filtering
24714 logic from here to...
24715 (layout::maybe_add_location_range): ...this new method. Add
24716 support for filtering to just the lines already specified by other
24718 (layout::will_show_line_p): New method.
24719 (gcc_rich_location::add_location_if_nearby): New method.
24720 (selftest::test_add_location_if_nearby): New test function.
24721 (selftest::diagnostic_show_locus_c_tests): Call it.
24722 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24725 2017-07-11 Tom de Vries <tom@codesourcery.com>
24727 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24728 (bb_first_real_insn): New function.
24729 (nvptx_single): Add extra initialization of broadcasted condition
24732 2017-07-11 Nathan Sidwell <nathan@acm.org>
24734 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24736 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24738 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24739 Binutils doc as TEXI will mess them up.
24740 * doc/invoke.texi (AVR Options): Same here.
24742 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24744 * config/sparc/sparc.opt (mfix-ut700): New option.
24745 (mfix-gr712rc): Likewise.
24746 (sparc_fix_b2bst): New variable.
24747 * doc/invoke.texi (SPARC options): Document them.
24748 (ARM options): Fix warnings.
24749 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24750 instructions to prevent sequences that can trigger the store-store
24751 errata for certain LEON3FT processors.
24752 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24753 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24754 * config/sparc/sparc.md (fix_b2bst): New attribute.
24755 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24757 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24760 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24762 (*rsqrtsf2_sse): Ditto.
24764 (div<mode>3): Macroize insn from divdf3 and divsf3
24765 using MODEF mode iterator.
24767 2017-07-10 Martin Sebor <msebor@redhat.com>
24769 PR tree-optimization/80397
24770 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24771 instead of testing for equality to INTEGER_TYPE.
24773 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24775 * config.gcc: Remove uclibc from arc target spec.
24777 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24779 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24781 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24784 * lto-wrapper.c (remove_option): New function.
24785 (merge_and_complain): Merge PIC/PIE options more realistically.
24787 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24789 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24793 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24794 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24795 * config.in: Regenerate.
24796 * configure: Regenerate.
24797 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24798 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24799 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24800 (TARGET_GASISR_PROLOGUES): ...target mask.
24801 * common/config/avr/avr-common.c
24802 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24803 Set -mgas-isr-prologues.
24804 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24805 INSERT_PASS_BEFORE for it.
24806 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24807 * config/avr/avr.c (avr_option_override)
24808 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24809 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24810 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24811 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24812 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24814 (make_avr_pass_pre_proep): New function.
24815 (emit_push_sfr) <treg>: Add argument to function and use it
24816 instead of TMP_REG.
24817 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24818 and set machine->gasisr.yes.
24819 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24820 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24821 __gcc_isr.n_pushed to .L__stack_usage.
24822 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24823 (avr_asm_final_postscan_insn): ...this new static function.
24824 * config/avr/avr.h (machine_function)
24825 <is_no_gccisr, use_L__stack_usage>: New fields.
24826 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24827 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24828 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24829 (gasisr, *gasisr): New expander and insn.
24830 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24831 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24832 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24834 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24836 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24839 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24841 Move jump-tables out of .text again.
24844 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24845 (ASM_OUTPUT_ADDR_VEC): New function.
24846 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24847 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24848 INSN_ADDRESSes as asm comment.
24849 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24850 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24851 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24852 * config/avr/avr.md (*tablejump): Adjust comment.
24853 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24854 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24856 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24857 (avr_output_addr_vec): New proto.
24858 (avr_log_t) <insn_addresses>: New field.
24860 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24863 * config/i386/i386.c (ix86_function_arg_advance): Set
24864 outgoing_args_on_stack to true if there are outgoing arguments
24866 (ix86_function_arg): Likewise.
24867 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24868 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24869 * config/i386/i386.h (machine_function): Add
24870 outgoing_args_on_stack.
24872 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24874 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24875 supporting pthreds.
24876 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24878 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24880 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24881 (REAL_H): Remove $(MACHMODE_H).
24882 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24884 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24885 $(MACHMODE_H) and double-int.h.
24886 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24888 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24891 2017-07-07 Andrew Pinski <apinski@cavium.com>
24893 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24894 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24896 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24898 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24899 Add warning if GCC was not configured to link against a GLIBC that
24900 exports the hardware capability bits.
24901 (make_resolver_func): Make resolver function private and not a
24902 COMDAT function. Create the name with clone_function_name instead
24903 of make_unique_name.
24906 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24907 correct operand in doing the split.
24909 2017-07-07 Carl Love <cel@us.ibm.com>
24911 * config/rs6000/rs6000-c: Add support for built-in function
24912 vector unsigned short vec_pack_to_short_fp32 (vector float,
24914 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24915 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24916 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24917 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24918 (convert_4f32_8i16): Add define_expand.
24919 * doc/extend.texi: Update the built-in documentation file for the
24920 new built-in function.
24922 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24924 * config/sparc/m8.md: New file.
24925 * config/sparc/sparc.md: Include m8.md.
24927 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24929 * config/sparc/sparc.opt: New option -mvis4b.
24930 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24931 (sparc_option_override): Handle VIS4B.
24932 (enum sparc_builtins): Define
24933 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24934 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24935 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24936 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24937 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24938 (check_constant_argument): New function.
24939 (sparc_vis_init_builtins): Define builtins
24940 __builtin_vis_dictunpack{8,16,32},
24941 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24942 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24943 __builtin_vis_fpcmpde{8,16,32}shl and
24944 __builtin_vis_fpcmpur{8,16,32}shl.
24945 (sparc_expand_builtin): Check that the constant operands to
24946 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24947 constant and in range.
24948 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24950 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24951 (SPARC_IMM5_P): Likewise.
24952 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24953 (enabled): Handle vis4b.
24954 (UNSPEC_DICTUNPACK): New unspec.
24955 (UNSPEC_FPCMPSHL): Likewise.
24956 (UNSPEC_FPUCMPSHL): Likewise.
24957 (UNSPEC_FPCMPDESHL): Likewise.
24958 (UNSPEC_FPCMPURSHL): Likewise.
24959 (cpu_feature): New CPU feature `vis4b'.
24960 (dictunpack{8,16,32}): New insns.
24961 (FPCSMODE): New mode iterator.
24962 (fpcscond): New code iterator.
24963 (fpcsucond): Likewise.
24964 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24965 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24966 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24967 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24968 * config/sparc/constraints.md: Define constraints `q' for unsigned
24969 2-bit integer constants and `t' for unsigned 5-bit integer
24971 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24973 (imm5_operand_dictunpack16): Likewise.
24974 (imm5_operand_dictunpack32): Likewise.
24975 (imm2_operand): Likewise.
24976 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24977 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24978 ditunpack* and fpcmp*shl builtins.
24980 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24982 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24983 * config.in: Add HAVE_AS_SPARC6 define.
24984 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24986 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24988 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24989 (CPP_CPU_SPEC): Handle m8.
24990 (ASM_CPU_SPEC): Likewise.
24991 * config/sparc/sparc-opts.h (enum processor_type): Add
24993 * config/sparc/sparc.c (m8_costs): New struct.
24994 (sparc_option_override): Handle TARGET_CPU_m8.
24995 (sparc32_initialize_trampoline): Likewise.
24996 (sparc64_initialize_trampoline): Likewise.
24997 (sparc_issue_rate): Likewise.
24998 (sparc_register_move_cost): Likewise.
24999 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
25000 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
25001 (ASM_CPU64_DEFAULT_SPEC): Likewise.
25002 (CPP_CPU_SPEC): Handle M8.
25003 (ASM_CPU_SPEC): Likewise.
25004 (AS_M8_FLAG): Define.
25005 * config/sparc/sparc.md: Add m8 to the cpu attribute.
25006 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25007 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25009 * configure: Regenerate.
25010 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25013 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25015 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25017 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25018 ("*movdi_insn_sp32"): Do not set v3pipe.
25019 ("*movsi_insn"): Likewise.
25020 ("*movdi_insn_sp64"): Likewise.
25021 ("*movsf_insn"): Likewise.
25022 ("*movdf_insn_sp32"): Likewise.
25023 ("*movdf_insn_sp64"): Likewise.
25024 ("*zero_extendsidi2_insn_sp64"): Likewise.
25025 ("*sign_extendsidi2_insn"): Likewise.
25026 ("*mov<VM32:mode>_insn"): Likewise.
25027 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25028 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25029 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25030 ("<vlop:code><VL:mode>3"): Likewise.
25031 ("*not_<vlop:code><VL:mode>3"): Likewise.
25032 ("*nand<VL:mode>_vis"): Likewise.
25033 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25034 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25035 ("one_cmpl<VL:mode>2"): Likewise.
25036 ("faligndata<VM64:mode>_vis"): Likewise.
25037 ("alignaddrsi_vis"): Likewise.
25038 ("alignaddrdi_vis"): Likweise.
25039 ("alignaddrlsi_vis"): Likewise.
25040 ("alignaddrldi_vis"): Likewise.
25041 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25042 ("bmaskdi_vis"): Likewise.
25043 ("bmasksi_vis"): Likewise.
25044 ("bshuffle<VM64:mode>_vis"): Likewise.
25045 ("cmask8<P:mode>_vis"): Likewise.
25046 ("cmask16<P:mode>_vis"): Likewise.
25047 ("cmask32<P:mode>_vis"): Likewise.
25048 ("pdistn<P:mode>_vis"): Likewise.
25049 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25051 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25053 * config/sparc/sparc.md ("subtype"): New insn attribute.
25054 ("*wrgsr_sp64"): Set insn subtype.
25055 ("*rdgsr_sp64"): Likewise.
25056 ("alignaddrsi_vis"): Likewise.
25057 ("alignaddrdi_vis"): Likewise.
25058 ("alignaddrlsi_vis"): Likewise.
25059 ("alignaddrldi_vis"): Likewise.
25060 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25061 ("fexpand_vis"): Likewise.
25062 ("fpmerge_vis"): Likewise.
25063 ("faligndata<VM64:mode>_vis"): Likewise.
25064 ("bshuffle<VM64:mode>_vis"): Likewise.
25065 ("cmask8<P:mode>_vis"): Likewise.
25066 ("cmask16<P:mode>_vis"): Likewise.
25067 ("cmask32<P:mode>_vis"): Likewise.
25068 ("fchksm16_vis"): Likewise.
25069 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25070 ("fmean16_vis"): Likewise.
25071 ("fp<plusminus_insn>64_vis"): Likewise.
25072 ("<plusminus_insn>v8qi3"): Likewise.
25073 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25074 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25075 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25076 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25077 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25078 ("*movqi_insn"): Likewise.
25079 ("*movhi_insn"): Likewise.
25080 ("*movsi_insn"): Likewise.
25081 ("movsi_pic_gotdata_op"): Likewise.
25082 ("*movdi_insn_sp32"): Likewise.
25083 ("*movdi_insn_sp64"): Likewise.
25084 ("movdi_pic_gotdata_op"): Likewise.
25085 ("*movsf_insn"): Likewise.
25086 ("*movdf_insn_sp32"): Likewise.
25087 ("*movdf_insn_sp64"): Likewise.
25088 ("*zero_extendhisi2_insn"): Likewise.
25089 ("*zero_extendqihi2_insn"): Likewise.
25090 ("*zero_extendqisi2_insn"): Likewise.
25091 ("*zero_extendqidi2_insn"): Likewise.
25092 ("*zero_extendhidi2_insn"): Likewise.
25093 ("*zero_extendsidi2_insn_sp64"): Likewise.
25094 ("ldfsr"): Likewise.
25095 ("prefetch_64"): Likewise.
25096 ("prefetch_32"): Likewise.
25097 ("tie_ld32"): Likewise.
25098 ("tie_ld64"): Likewise.
25099 ("*tldo_ldub_sp32"): Likewise.
25100 ("*tldo_ldub1_sp32"): Likewise.
25101 ("*tldo_ldub2_sp32"): Likewise.
25102 ("*tldo_ldub_sp64"): Likewise.
25103 ("*tldo_ldub1_sp64"): Likewise.
25104 ("*tldo_ldub2_sp64"): Likewise.
25105 ("*tldo_ldub3_sp64"): Likewise.
25106 ("*tldo_lduh_sp32"): Likewise.
25107 ("*tldo_lduh1_sp32"): Likewise.
25108 ("*tldo_lduh_sp64"): Likewise.
25109 ("*tldo_lduh1_sp64"): Likewise.
25110 ("*tldo_lduh2_sp64"): Likewise.
25111 ("*tldo_lduw_sp32"): Likewise.
25112 ("*tldo_lduw_sp64"): Likewise.
25113 ("*tldo_lduw1_sp64"): Likewise.
25114 ("*tldo_ldx_sp64"): Likewise.
25115 ("*mov<VM32:mode>_insn"): Likewise.
25116 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25117 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25119 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25121 * config/sparc/sparc.md ("type"): New insn type viscmp.
25122 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25124 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25125 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25126 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25127 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25129 ("n7_vis_logical_11cycle"): Likewise.
25130 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25131 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25132 * config/sparc/niagara.md ("niag_vis"): Likewise.
25133 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25134 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25136 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25138 * config/sparc/sparc.md: New instruction type `bmask'.
25139 (bmaskdi_vis): Use the `bmask' type.
25140 (bmasksi_vis): Likewise.
25141 * config/sparc/ultra3.md (us3_array): Likewise.
25142 * config/sparc/niagara7.md (n7_array): Likewise.
25143 * config/sparc/niagara4.md (n4_array): Likewise.
25144 * config/sparc/niagara2.md (niag2_vis): Likewise.
25145 (niag3_vis): Likewise.
25146 * config/sparc/niagara.md (niag_vis): Likewise.
25148 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25150 * ipa-comdats.c: Remove optimize check from gate.
25151 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25152 for functions not optimized.
25153 (ipa_fn_summary_read): Skip optimize check.
25154 (ipa_fn_summary_write): Likewise.
25155 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25157 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25159 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25160 (check_callers): Give up on caller which is not optimized.
25161 (inline_small_functions): Likewise.
25162 (ipa_inline): Do not give up when not optimizing.
25163 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25164 away unoptimizes cdtors.
25165 (whole_program_function_and_variable_visibility): Do
25166 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25167 * ipa.c (process_references): Do not check optimize.
25168 (symbol_table::remove_unreachable_nodes): Update optimize check.
25169 (set_writeonly_bit): Update optimize check.
25170 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25171 (pass_ipa_single_use::gate): Remove.
25173 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25175 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25176 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25177 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25178 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25179 permute_load, permute_store, adjust_extract, adjust_splat,
25180 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25181 replace_swap_with_copy, dump_swap_insn_table,
25182 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25183 recombine_lvx_pattern, recombine_stvx_pattern,
25184 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25185 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25186 to file rs6000-p8swap.c.
25187 * config/rs6000/rs6000-p8swap.c: New file.
25188 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25189 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25190 and rs6000*-*-* targets.
25192 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25194 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25196 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25198 * lto-wrapper.c (merge_and_complain): Do not merge
25199 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25200 fsigned_zeros, ftrapping_math, fwrapv.
25201 (append_compiler_options): Do not track these options.
25202 (append_linker_options): Likewie
25204 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25206 * cgraphunit.c (cgraph_node::finalize_function): When
25207 !flag_toplevel_reorde set no_reorder flag.
25208 (varpool_node::finalize_decl): Likewise.
25209 (symbol_table::compile): Drop no toplevel reorder path.
25211 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25213 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25214 edges; zero probability is not better than uninitialized.
25216 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25218 * asan.h (asan_sanitize_allocas_p): Declare.
25219 * asan.c (asan_sanitize_allocas_p): New function.
25220 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25221 (handle_builtin_alloca): Likewise.
25222 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25223 if !asan_sanitize_allocas_p.
25224 * params.def (asan-instrument-allocas): Add new option.
25225 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25226 * opts.c (common_handle_option): Disable allocas sanitization for
25229 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25231 * asan.c: Include gimple-fold.h.
25232 (get_last_alloca_addr): New function.
25233 (handle_builtin_stackrestore): Likewise.
25234 (handle_builtin_alloca): Likewise.
25235 (asan_emit_allocas_unpoison): Likewise.
25236 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25237 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25238 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25239 (instrument_builtin_call): Pass gimple iterator to
25240 get_mem_refs_of_builtin_call.
25241 (last_alloca_addr): New global.
25242 * asan.h (asan_emit_allocas_unpoison): Declare.
25243 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25244 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25245 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25246 if function calls alloca.
25247 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25248 * gimple-fold.h (replace_call_with_value): Declare.
25249 * internal-fn.c: Include asan.h.
25250 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25251 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25253 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25255 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25256 (C_SELFTEST_FLAGS): New.
25257 (CPP_SELFTEST_FLAGS): New.
25258 (SELFTEST_DEPS): New, from deps of s-selftest.
25259 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25260 (CPP_SELFTEST_DEPS): New.
25261 (selftest): Add dependency on s-selftest-c++.
25262 (s-selftest): Rename to...
25263 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25264 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25265 than SELFTEST_FLAGS.
25266 (selftest-gdb): Rename to...
25267 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25269 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25270 (selftest-valgrind): Rename to...
25271 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25273 (selftest-valgrind): Reintroduce as an alias for
25274 selftest-c-valgrind.
25275 (s-selftest-c++): New.
25276 (selftest-c++-gdb): New.
25277 (selftest-c++-valgrind): New.
25279 2017-07-06 Olivier Hainque <hainque@adacore.com>
25281 * gcc.c (process_command): When deciding if undefined variables
25282 should be ignored when processing specs, accept "gcc -v" as well.
25284 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25286 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25287 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25289 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25291 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25292 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25293 * config/arm/arm-tables.opt: Regenerate.
25294 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25296 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25298 2017-07-06 Carl Love <cel@us.ibm.com>
25300 * ChangeLog: Clean up from mid air collision
25302 2017-07-06 Carl Love <cel@us.ibm.com>
25304 * config/rs6000/rs6000-c.c: Add support for built-in functions
25305 vector signed int vec_subc (vector signed int, vector signed int);
25306 vector signed __int128 vec_subc (vector signed __int128,
25307 vector signed __int128);
25308 vector unsigned __int128 vec_subc (vector unsigned __int128,
25309 vector unsigned __int128);
25310 vector signed int vec_sube (vector signed int, vector signed int,
25311 vector signed int);
25312 vector unsigned int vec_sube (vector unsigned int,
25313 vector unsigned int,
25314 vector unsigned int);
25315 vector signed __int128 vec_sube (vector signed __int128,
25316 vector signed __int128,
25317 vector signed__int128);
25318 vector unsigned __int128 vec_sube (vector unsigned __int128,
25319 vector unsigned __int128,
25320 vector unsigned __int128);
25321 vector signed int vec_subec (vector signed int, vector signed int,
25322 vector signed int);
25323 vector unsigned int vec_subec (vector unsigned int,
25324 vector unsigned int,
25325 vector unsigned int);
25326 vector signed __int128 vec_subec (vector signed __int128,
25327 vector signed __int128,
25328 vector signed__int128);
25329 vector unsigned __int128 vec_subec (vector unsigned __int128,
25330 vector unsigned __int128,
25331 vector unsigned __int128);
25332 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25333 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25334 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25335 BU_ALTIVEC_OVERLOAD_X definitions.
25336 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25337 * doc/extend.texi: Update the built-in documentation file for the new
25338 built-in functions.
25340 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25343 * diagnostic-show-locus.c (layout::layout): Use start and finish
25344 spelling location for the start and finish of each range.
25345 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25346 Add unused aspect param.
25347 * input.c (expand_location_1): Add "aspect" param, and use it
25348 to access the correct part of the location.
25349 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25351 (expand_location_to_spelling_point): Likewise.
25352 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25353 param, and pass it to expand_location_1.
25355 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25357 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25358 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25359 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25360 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25361 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25362 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25363 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25364 _mm_maskz_getmant_ss): New intrinsics.
25365 (__builtin_ia32_getexpss128_mask): Changed to ...
25366 __builtin_ia32_getexpss128_round ... this.
25367 (__builtin_ia32_getexpsd128_mask): Changed to ...
25368 __builtin_ia32_getexpsd128_round ... this.
25369 * config/i386/i386-builtin-types.def
25370 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25371 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25372 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25373 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25374 __builtin_ia32_getmantss_mask_round): New builtins.
25375 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25376 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25377 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25378 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25379 * config/i386/sse.md
25380 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25381 avx512f_sgetexp<mode><mask_scalar_name>
25382 <round_saeonly_scalar_name> ... this.
25383 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25384 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25385 vgetexp<ssescalarmodesuffix>
25386 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25387 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25388 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25389 avx512f_vgetmant<mode><mask_scalar_name>
25390 <round_saeonly_scalar_name> ... this.
25391 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25392 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25393 vgetmant<ssescalarmodesuffix>
25394 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25395 %0<mask_scalar_operand4>, %1, %2
25396 <round_saeonly_scalar_mask_op4>, %3} ... this.
25397 * config/i386/subst.md (mask_scalar_operand4,
25398 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25399 round_saeonly_scalar_nimm_predicate): New subst attributes.
25401 2017-07-06 Julia Koval <julia.koval@intel.com>
25403 * config/i386/i386.c (ix86_erase_embedded_rounding):
25404 Remove code for old rounding pattern.
25406 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25408 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25410 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25412 * doc/sourcebuild.texi (Test Directives, Variants of
25413 dg-require-support): Add documentation for dg-require-stack-check.
25415 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25417 * config/i386/subst.md (mask_scalar, round_scalar,
25418 round_saeonly_scalar): New meta-templates.
25419 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25420 round_scalar_mask_operand3, round_scalar_mask_op3,
25421 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25422 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25423 round_saeonly_scalar_constraint,
25424 round_saeonly_scalar_prefix): New subst attribute.
25425 * config/i386/sse.md
25426 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25427 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25428 <round_scalar_name> ... this.
25429 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25430 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25431 <round_scalar_name> ... this.
25432 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25433 <sse>_vm<code><mode>3<mask_scalar_name>
25434 <round_saeonly_scalar_name> ... this.
25435 (v<plusminus_mnemonic><ssescalarmodesuffix>
25436 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25437 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25438 v<plusminus_mnemonic><ssescalarmodesuffix>
25439 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25440 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25441 (v<multdiv_mnemonic><ssescalarmodesuffix>
25442 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25443 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25444 v<multdiv_mnemonic><ssescalarmodesuffix>
25445 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25446 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25447 (v<maxmin_float><ssescalarmodesuffix>
25448 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25449 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25450 v<maxmin_float><ssescalarmodesuffix>
25451 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25452 %0<mask_scalar_operand3>, %1, %<iptr>2
25453 <round_saeonly_scalar_mask_op3>} ... this.
25455 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25457 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25458 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25460 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25461 Alan Hayward <alan.hayward@arm.com>
25462 David Sherwood <david.sherwood@arm.com>
25464 * combine.c (simplify_if_then_else): Remove "enum" before
25466 * compare-elim.c (can_eliminate_compare): Likewise.
25467 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25469 (aarch64_lookup_simd_builtin_type): Likewise.
25470 (aarch64_simd_builtin_type): Likewise.
25471 (aarch64_init_simd_builtin_types): Likewise.
25472 (aarch64_simd_expand_args): Likewise.
25473 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25475 (aarch64_reverse_mask): Likewise.
25476 (aarch64_simd_emit_reg_reg_move): Likewise.
25477 (aarch64_gen_adjusted_ldpstp): Likewise.
25478 (aarch64_ccmp_mode_to_code): Likewise.
25479 (aarch64_operands_ok_for_ldpstp): Likewise.
25480 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25481 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25483 (aarch64_min_divisions_for_recip_mul): Likewise.
25484 (aarch64_reassociation_width): Likewise.
25485 (aarch64_get_condition_code_1): Likewise.
25486 (aarch64_simd_emit_reg_reg_move): Likewise.
25487 (aarch64_simd_attr_length_rglist): Likewise.
25488 (aarch64_reverse_mask): Likewise.
25489 (aarch64_operands_ok_for_ldpstp): Likewise.
25490 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25491 (aarch64_gen_adjusted_ldpstp): Likewise.
25492 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25494 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25495 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25496 (arm_lookup_simd_builtin_type): Likewise.
25497 (arm_simd_builtin_type): Likewise.
25498 (arm_init_simd_builtin_types): Likewise.
25499 (arm_expand_builtin_args): Likewise.
25500 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25501 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25502 (ft32_setup_incoming_varargs): Likewise.
25503 (ft32_function_arg): Likewise.
25504 (ft32_function_arg_advance): Likewise.
25505 (ft32_pass_by_reference): Likewise.
25506 (ft32_arg_partial_bytes): Likewise.
25507 (ft32_valid_pointer_mode): Likewise.
25508 (ft32_addr_space_pointer_mode): Likewise.
25509 (ft32_addr_space_legitimate_address_p): Likewise.
25510 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25512 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25513 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25514 (iamcu_alignment): Likewise.
25515 (canonicalize_vector_int_perm): Likewise.
25516 (ix86_noce_conversion_profitable_p): Likewise.
25517 (ix86_mpx_bound_mode): Likewise.
25518 (ix86_operands_ok_for_move_multiple): Likewise.
25519 * config/microblaze/microblaze-protos.h
25520 (microblaze_expand_conditional_branch_reg): Likewise.
25521 * config/microblaze/microblaze.c
25522 (microblaze_expand_conditional_branch_reg): Likewise.
25523 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25525 (rs6000_reassociation_width): Likewise.
25526 (rs6000_invalid_binary_op): Likewise.
25527 (fusion_p9_p): Likewise.
25528 (emit_fusion_p9_load): Likewise.
25529 (emit_fusion_p9_store): Likewise.
25530 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25532 (riscv_hard_regno_mode_ok_p): Likewise.
25533 (riscv_address_insns): Likewise.
25534 (riscv_split_symbol): Likewise.
25535 (riscv_legitimize_move): Likewise.
25536 (riscv_function_value): Likewise.
25537 (riscv_hard_regno_nregs): Likewise.
25538 (riscv_expand_builtin): Likewise.
25539 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25540 (riscv_build_integer): Likewise.
25541 (riscv_split_integer): Likewise.
25542 (riscv_legitimate_constant_p): Likewise.
25543 (riscv_cannot_force_const_mem): Likewise.
25544 (riscv_regno_mode_ok_for_base_p): Likewise.
25545 (riscv_valid_base_register_p): Likewise.
25546 (riscv_valid_offset_p): Likewise.
25547 (riscv_valid_lo_sum_p): Likewise.
25548 (riscv_classify_address): Likewise.
25549 (riscv_legitimate_address_p): Likewise.
25550 (riscv_address_insns): Likewise.
25551 (riscv_load_store_insns): Likewise.
25552 (riscv_force_binary): Likewise.
25553 (riscv_split_symbol): Likewise.
25554 (riscv_force_address): Likewise.
25555 (riscv_legitimize_address): Likewise.
25556 (riscv_move_integer): Likewise.
25557 (riscv_legitimize_const_move): Likewise.
25558 (riscv_legitimize_move): Likewise.
25559 (riscv_address_cost): Likewise.
25560 (riscv_subword): Likewise.
25561 (riscv_output_move): Likewise.
25562 (riscv_canonicalize_int_order_test): Likewise.
25563 (riscv_emit_int_order_test): Likewise.
25564 (riscv_function_arg_boundary): Likewise.
25565 (riscv_pass_mode_in_fpr_p): Likewise.
25566 (riscv_pass_fpr_single): Likewise.
25567 (riscv_pass_fpr_pair): Likewise.
25568 (riscv_get_arg_info): Likewise.
25569 (riscv_function_arg): Likewise.
25570 (riscv_function_arg_advance): Likewise.
25571 (riscv_arg_partial_bytes): Likewise.
25572 (riscv_function_value): Likewise.
25573 (riscv_pass_by_reference): Likewise.
25574 (riscv_setup_incoming_varargs): Likewise.
25575 (riscv_print_operand): Likewise.
25576 (riscv_elf_select_rtx_section): Likewise.
25577 (riscv_save_restore_reg): Likewise.
25578 (riscv_for_each_saved_reg): Likewise.
25579 (riscv_register_move_cost): Likewise.
25580 (riscv_hard_regno_mode_ok_p): Likewise.
25581 (riscv_hard_regno_nregs): Likewise.
25582 (riscv_class_max_nregs): Likewise.
25583 (riscv_memory_move_cost): Likewise.
25584 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25585 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25586 (rl78_addr_space_address_mode): Likewise.
25587 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25589 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25590 (rs6000_reassociation_width): Likewise.
25591 (rs6000_invalid_binary_op): Likewise.
25592 (fusion_p9_p): Likewise.
25593 (emit_fusion_p9_load): Likewise.
25594 (emit_fusion_p9_store): Likewise.
25595 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25596 (ok_for_simple_move_operands): Likewise.
25597 (ok_for_simple_move_strict_operands): Likewise.
25598 (ok_for_simple_arith_logic_operands): Likewise.
25599 (visium_legitimize_reload_address): Likewise.
25600 (visium_select_cc_mode): Likewise.
25601 (output_cbranch): Likewise.
25602 (visium_split_double_move): Likewise.
25603 (visium_expand_copysign): Likewise.
25604 (visium_expand_int_cstore): Likewise.
25605 (visium_expand_fp_cstore): Likewise.
25606 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25607 (visium_function_arg): Likewise.
25608 (visium_function_arg_advance): Likewise.
25609 (visium_libcall_value): Likewise.
25610 (visium_setup_incoming_varargs): Likewise.
25611 (visium_legitimate_constant_p): Likewise.
25612 (visium_legitimate_address_p): Likewise.
25613 (visium_legitimize_address): Likewise.
25614 (visium_secondary_reload): Likewise.
25615 (visium_register_move_cost): Likewise.
25616 (visium_memory_move_cost): Likewise.
25617 (prepare_move_operands): Likewise.
25618 (ok_for_simple_move_operands): Likewise.
25619 (ok_for_simple_move_strict_operands): Likewise.
25620 (ok_for_simple_arith_logic_operands): Likewise.
25621 (visium_function_value_1): Likewise.
25622 (rtx_ok_for_offset_p): Likewise.
25623 (visium_legitimize_reload_address): Likewise.
25624 (visium_split_double_move): Likewise.
25625 (visium_expand_copysign): Likewise.
25626 (visium_expand_int_cstore): Likewise.
25627 (visium_expand_fp_cstore): Likewise.
25628 (visium_split_cstore): Likewise.
25629 (visium_select_cc_mode): Likewise.
25630 (visium_split_cbranch): Likewise.
25631 (output_cbranch): Likewise.
25632 (visium_print_operand_address): Likewise.
25633 * expmed.c (flip_storage_order): Likewise.
25634 * expmed.h (emit_cstore): Likewise.
25635 (flip_storage_order): Likewise.
25636 * genrecog.c (validate_pattern): Likewise.
25637 * hsa-gen.c (gen_hsa_addr): Likewise.
25638 * internal-fn.c (expand_arith_overflow): Likewise.
25639 * ira-color.c (allocno_copy_cost_saving): Likewise.
25640 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25641 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25642 (process_invariant_for_inheritance): Likewise.
25643 * lra-eliminations.c (move_plus_up): Likewise.
25644 * omp-low.c (lower_oacc_reductions): Likewise.
25645 * simplify-rtx.c (simplify_subreg): Likewise.
25646 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25647 (TARGET_CHKP_BOUND_MODE): Likewise..
25648 * targhooks.c (default_chkp_bound_mode): Likewise.
25649 (default_setup_incoming_vararg_bounds): Likewise.
25650 * targhooks.h (default_chkp_bound_mode): Likewise.
25651 (default_setup_incoming_vararg_bounds): Likewise.
25652 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25653 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25654 (have_whole_vector_shift): Likewise.
25655 * tree-vect-stmts.c (vectorizable_load): Likewise.
25656 * doc/tm.texi: Regenerate.
25658 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25660 Graceful degrade if Binutils PR21472 is not available.
25663 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25664 .rodata in flash test fails.
25665 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25666 * confgure: Regenerate.
25667 * config.in: Regenerate.
25668 * config/avr/avr.c (avr_asm_named_section)
25669 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25670 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25671 (avr_asm_init_sections): Same.
25673 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25675 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25676 (fma<VH:mode>4_intrinsic): Likewise.
25677 (*fmsub<VCVTF:mode>4): Likewise.
25678 (*fmsub<VH:mode>4_intrinsic): Likewise.
25680 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25683 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25684 Don't depend on "optimize > 0".
25685 (out_movhi_r_mr, out_movqi_mr_r): Same.
25686 (out_movhi_mr_r, out_movqi_r_mr): Same.
25687 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25688 io_address_operand on "optimize > 0".
25690 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25692 * tree-loop-distribution.c: Add general explanantion on the pass.
25693 (generate_loops_for_partition): Mark distributed loop.
25694 (pg_add_dependence_edges): New parameter. Handle alias data
25695 dependence specially and record it in the parameter if asked.
25696 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25697 (init_partition_graph_vertices, add_partition_graph_edge): New.
25698 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25699 (free_partition_graph_vdata, build_partition_graph): New.
25700 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25701 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25702 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25703 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25704 (version_for_distribution_p, finalize_partitions): New.
25705 (distribute_loop): Handle alias data dependence specially. Factor
25706 out loop fusion code as functions and call these functions.
25708 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25710 * tree-loop-distribution.c (classify_partition): New parameter and
25711 better handle reduction statement.
25712 (rdg_build_partitions): Revise comment.
25713 (distribute_loop): Compute statements in all partitions and pass it
25714 to classify_partition.
25716 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25718 * tree-loop-distribution.c (enum partition_type): New.
25719 (struct partition): New field type.
25720 (partition_merge_into): Add parameter. Update partition type.
25721 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25722 (build_rdg_partition_for_vertex): Compute partition type.
25723 (rdg_build_partitions): Dump partition type.
25724 (distribute_loop): Update calls to partition_merge_into.
25726 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25728 * tree-loop-distribution.c (struct ddr_hasher): New.
25729 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25731 (classify_partition): Call get_data_dependence.
25732 (pg_add_dependence_edges): Ditto.
25733 (distribute_loop): Release data dependence hash table.
25735 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25737 * tree-loop-distribution.c (ref_base_address): Delete.
25738 (similar_memory_accesses): Rename ...
25739 (share_memory_accesses): ... to this. Check if partitions access
25740 the same memory reference.
25741 (distribute_loop): Call share_memory_accesses.
25743 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25745 * tree-loop-distribution.c (struct partition): New field recording
25746 its data reference.
25747 (partition_alloc, partition_free): Init and release data refs.
25748 (partition_merge_into): Merge data refs.
25749 (build_rdg_partition_for_vertex): Collect data refs for partition.
25750 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25752 (distribute_loop): Remve data refs from vertice data of partition
25755 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25757 * tree-loop-distribution.c (params.h): Include header file.
25758 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25759 (datarefs_vec): New global var.
25760 (create_rdg_vertices): Use datarefs_vec directly.
25761 (free_rdg): Don't free data references.
25762 (build_rdg): Update use. Don't free data references.
25763 (distribute_loop): Compute global variable for data references.
25764 Bail out if there are too many data references.
25766 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25768 * tree-loop-distribution.c (loop_nest): New global var.
25769 (build_rdg): Use loop directly, rather than loop nest.
25770 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25772 (distribute_loop): Compute global variable loop nest. Update use.
25774 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25776 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25777 (partition_merge_into): New parameter. Dump reason for fusion.
25778 (distribute_loop): Update use of partition_merge_into.
25780 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25782 * tree-loop-distribution.c (bb_top_order_index): New.
25783 (bb_top_order_index_size, bb_top_order_cmp): New.
25784 (stmts_from_loop): Use topological order.
25785 (pass_loop_distribution::execute): Compute and release topological
25786 order for basic blocks.
25788 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25790 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25793 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25795 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25796 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25797 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25798 * internal-fn.def (LOOP_DIST_ALIAS): New.
25799 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25800 (fold_loop_internal_call): ... this.
25801 (vect_loop_dist_alias_call): New function.
25802 (set_uid_loop_bbs): Call fold_loop_internal_call.
25803 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25806 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25809 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25810 Require dead FLAGS_REG at the beginning of a peephole.
25812 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25815 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25816 arguments in the call to __builtin_ia32_sbb_u32.
25817 (_subborrow_u64): Swap _X and _Y arguments in the call to
25818 __builtin_ia32_sbb_u64.
25820 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25823 * tree-vrp.c (compare_assert_loc): Turn into a function template
25824 with stable template parameter. Only test if a->e is NULL,
25825 !a->e == !b->e has been verified already. Use e == NULL or
25826 e != NULL instead of e or ! e tests. If stable is true, don't use
25827 iterative_hash_expr, on the other side allow a or b or both NULL
25828 and sort the NULLs last.
25829 (process_assert_insertions): Sort using compare_assert_loc<false>
25830 instead of compare_assert_loc, later sort using
25831 compare_assert_loc<true> before calling process_assert_insertions_for
25832 in a loop. Use break instead of continue once seen NULL pointer.
25834 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25836 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25837 Cortex-R7 and Cortex-R8 processors.
25839 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25841 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25842 uninitialized while src is not.
25844 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25846 * common/config/arm/arm-common.c: Adjust include path for
25848 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25849 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25850 (arm-cpu-data.h): Likewise.
25851 (arm-cpu-cdata.h): Likewise.
25852 * config/arm/arm-cpu.h: Delete.
25853 * config/arm/arm-cpu-cdata.h: Delete.
25854 * config/arm/arm-cpu-data.h: Delete.
25856 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25858 * config/arm/arm-cpus.in (cortex-a55): New.
25859 (cortex-a75): Likewise.
25860 (cortex-a75.cortex-a55): Likewise.
25861 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25863 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25864 * config/arm/arm-cpu-cdata.h: Regenerate.
25865 * config/arm/arm-cpu-data.h: Regenerate.
25866 * config/arm/arm-cpu.h: Regenerate.
25867 * config/arm/arm-tables.opt: Regenerate.
25868 * config/arm/arm-tune.md: Regenerate.
25870 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25872 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25874 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25876 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25879 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25881 PR tree-optimization/81292
25882 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25883 full_string_p, also call adjust_related_strinfos if the adjustment
25884 is simple, otherwise invalidate related strinfos.
25886 2017-07-04 Martin Liska <mliska@suse.cz>
25889 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25890 newly created variable as DECL_IGNORED_P.
25892 2017-07-04 Martin Liska <mliska@suse.cz>
25895 * ipa-inline.c (inline_small_functions):
25896 Use xstrdup_for_dump.
25898 2017-07-04 Tom de Vries <tom@codesourcery.com>
25900 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25902 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25905 * config/darwin.c (darwin_function_switched_text_sections):
25908 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25910 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25912 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25914 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25916 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25918 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25919 min_profitable_iters, and th as inclusive lower bounds.
25920 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25921 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25922 for min_profitable_iters and min_profitable_estimate.
25923 (vect_transform_loop): Treat th as an inclusive lower bound.
25924 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25926 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25929 * config/darwin.c (darwin_function_switched_text_sections):
25930 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25931 in two pieces, and suppress the use of buf.
25933 2017-07-03 Nathan Sidwell <nathan@acm.org>
25935 * hash-table.h (hash_table_mod1): Fix indentation.
25937 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25939 PR middle-end/81290
25940 * predict.c (force_edge_cold): Be more careful about propagation
25942 * profile-count.h (profile_probability::guessed,
25943 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25945 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25947 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25949 * doc/invoke.texi (rcpc architecture extension): Document it.
25951 2017-07-03 Richard Biener <rguenther@suse.de>
25953 PR tree-optimization/60510
25954 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25955 the scalar reduction PHI and use it.
25956 (vectorizable_reduction): Properly guard the single_defuse_cycle
25957 path for non-SLP reduction chains where we cannot use it.
25958 Rework reduc_def/index and vector type deduction. Rework
25959 vector operand gathering during reduction op code-gen.
25960 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25961 chains dissolve the chain and leave it to non-SLP reduction
25964 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25966 * tree-data-ref.h (dr_alignment): Declare.
25967 * tree-data-ref.c (dr_alignment): New function.
25968 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25969 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25971 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25973 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25975 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25976 and base_misalignment fields.
25977 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25978 * tree-data-ref.c: Include builtins.h.
25979 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25980 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25981 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25982 * tree-vect-data-refs.c: Include tree-cfg.h.
25983 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25984 fields instead of calculating an alignment here.
25985 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25986 innermost_loop_behavior fields.
25988 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25990 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25992 (DR_STEP_ALIGNMENT): New macro.
25993 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25994 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25995 (create_data_ref): Print it.
25996 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25997 to tell whether the step preserves vector (mis)alignment.
25998 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25999 Move the check for an integer step and generalise to all INTEGER_CST.
26000 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
26001 Print the outer step alignment.
26003 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26005 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
26006 with offset_alignment.
26007 (DR_ALIGNED_TO): Delete.
26008 (DR_OFFSET_ALIGNMENT): New macro.
26009 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26010 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26011 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26012 (create_data_ref): Likewise.
26013 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26014 (vect_analyze_data_refs): Likewise.
26015 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26016 creating dummy innermost behavior.
26018 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26020 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26021 with a "innermost_loop_behavior *" and refeence tree.
26022 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26023 (create_data_ref): Update call accordingly.
26024 * tree-predcom.c (find_looparound_phi): Likewise.
26026 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26028 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26029 fields with dr_wrt_vec_loop.
26030 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26031 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26032 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26033 (vect_dr_behavior): New function.
26034 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26035 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26036 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26037 track whether the step preserves the misalignment.
26038 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26039 Use vect_dr_behavior.
26040 (vect_setup_realignment): Update call accordingly.
26041 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26042 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26043 call to vect_create_addr_base_for_vector_ref.
26044 (vect_create_cond_for_align_checks): Likewise.
26045 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26046 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26047 (vect_recog_mask_conversion_pattern): Likewise.
26048 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26049 (new_stmt_vec_info): Remove redundant zeroing.
26051 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26053 * common/config/arm/arm-common.c (arm_be8_option): New function.
26054 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26055 (ISA_ARMv6): Add isa_bit_be8.
26056 * config/arm/arm.h (arm_be8_option): Add prototype.
26057 (BE8_SPEC_FUNCTION): New define.
26058 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26059 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26060 (mlittle-endian): Similarly.
26061 (mbe8, mbe32): New options.
26062 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26063 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26065 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26067 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26069 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26071 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26072 (cleanup_tree_cfg_bb): Use it.
26073 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26075 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26077 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26080 * loop-doloop.c (add_test): Update profile.
26082 2017-07-03 Martin Liska <mliska@suse.cz>
26085 * sanopt.c (rewrite_usage_of_param): New function.
26086 (sanitize_rewrite_addressable_params): Likewise.
26087 (pass_sanopt::execute): Call rewrite_usage_of_param.
26089 2017-07-03 Richard Biener <rguenther@suse.de>
26091 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26092 back to using VIEW_CONVERT_EXPR.
26094 2017-07-03 Martin Liska <mliska@suse.cz>
26097 * doc/extend.texi: Document when a resolver function is
26098 generated for target_clones.
26100 2017-07-03 Martin Liska <mliska@suse.cz>
26102 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26103 and shadow memory of auto variables which are subject of
26104 use-after-scope sanitization.
26105 (asan_expand_mark_ifn): Add do set only when is_poison.
26107 2016-07-03 Richard Biener <rguenther@suse.de>
26109 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26111 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26112 (vectorizable_reduction): Perform reduction PHI creation when
26113 visiting a reduction PHI and adjust and simplify code generation
26114 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26115 (vect_transform_loop): Visit reduction PHIs.
26116 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26117 defs into the SLP tree.
26118 (vect_build_slp_tree): Reduction defs terminate the recursion.
26119 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26121 (vect_get_vec_defs_for_stmt_copy): Export.
26122 (vect_get_vec_defs): Likewise.
26123 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26125 (vect_get_vec_defs_for_stmt_copy): Declare.
26126 (vect_get_vec_defs): Likewise.
26128 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26130 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26131 parameter with a "loop" parameter and use it instead of the
26132 loop containing DR_STMT. Don't check simple_iv when doing
26133 BB analysis. Describe the two analysis modes in the comment.
26135 2017-07-03 Tom de Vries <tom@codesourcery.com>
26137 PR tree-optimization/69468
26138 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26139 (find_same_succ_bb): Handle ignore_edge_flags.
26141 2017-07-03 Tom de Vries <tom@codesourcery.com>
26143 PR tree-optimization/81192
26144 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26146 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26148 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26150 2017-07-03 Tom de Vries <tom@codesourcery.com>
26152 PR tree-optimization/81192
26153 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26154 BB_SAME_SUCC (bb) == NULL.
26156 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26158 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26161 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26163 * dumpfile.c: Include profile-count.h
26164 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26166 (insert_cond_bb): Update profile.
26167 * tree-cfg.h (insert_cond_bb): Update prototype.
26168 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26169 * tree-dump.c: Do not include tree-cfg.
26171 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26173 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26175 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26177 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26180 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26182 * tree-complex.c (expand_complex_div_wide): update profile.
26184 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26185 Alan Hayward <alan.hayward@arm.com>
26186 David Sherwood <david.sherwood@arm.com>
26188 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26189 (CORETYPES_H): New define.
26190 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26191 (insn-modes-inline.h, s-modes-inline-h): New rules.
26192 (generated_files): Add insn-modes-inline.h.
26193 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26194 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26195 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26196 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26197 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26198 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26199 (build/gencodes.o, build/genconditions.o): Likewise.
26200 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26201 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26202 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26203 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26204 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26205 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26206 * coretypes.h: Include everything up to real.h for generators.
26207 Include insn-modes.h first. Include wide-int-print.h after
26208 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26209 * machmode.h: Don't include insn-modes.h here.
26210 * function-tests.c: Remove includes of signop.h, machmode.h,
26211 double-int.h and wide-int.h.
26213 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26215 * optc-save-gen.awk: Likewise.
26216 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26217 * godump.c: Remove include of wide-int-print.h.
26218 * pretty-print.h: Likewise.
26219 * wide-int-print.cc: Likewise.
26220 * wide-int.cc: Likewise.
26221 * hash-map-tests.c: Remove include of signop.h.
26222 * hash-set-tests.c: Likewise.
26223 * rtl-tests.c: Likewise.
26224 * mkconfig.sh: Remove include of machmode.h.
26225 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26227 (emit_insn_modes_inline_h): ...this new function. Emit the code
26228 into an insn-modes-inline.h header file, adding appropriate
26229 include guards and end comments.
26230 (emit_insn_modes_c_header): Remove include of machmode.h.
26231 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26233 (main): Handle -i flag and call emit_insn_modes_inline_h when
26236 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26238 * tree-ssa-strlen.c (strinfo): Rename the length field to
26239 nonzero_chars. Add a full_string_p field.
26240 (compare_nonzero_chars, zero_length_string_p): New functions.
26241 (get_addr_stridx): Add an offset_out parameter.
26242 Use compare_nonzero_chars.
26243 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26244 (new_strinfo): Update after above changes to strinfo.
26245 (set_endptr_and_length): Set full_string_p.
26246 (get_string_length): Update after above changes to strinfo.
26247 (unshare_strinfo): Update call to new_strinfo.
26248 (maybe_invalidate): Likewise.
26249 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26250 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26251 as a uhwi instead of an shwi. Update after above changes to
26252 strinfo and new_strinfo.
26253 (zero_length_string): Assert that chainsi contains full strings.
26254 Use zero_length_string_p. Update call to new_strinfo.
26255 (adjust_related_strinfos): Update after above changes to strinfo.
26256 Copy full_string_p from origsi.
26257 (adjust_last_stmt): Use zero_length_string_p.
26258 (handle_builtin_strlen): Update after above changes to strinfo and
26259 new_strinfo. Install the lhs as the string length if the previous
26260 entry didn't describe a full string.
26261 (handle_builtin_strchr): Update after above changes to strinfo
26263 (handle_builtin_strcpy): Likewise.
26264 (handle_builtin_strcat): Likewise.
26265 (handle_builtin_malloc): Likewise.
26266 (handle_pointer_plus): Likewise.
26267 (handle_builtin_memcpy): Likewise. Track nonzero characters
26268 that aren't necessarily followed by a nul terminator.
26269 (handle_char_store): Likewise.
26271 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26273 PR tree-optimization/80769
26274 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26275 for malloc and calloc. Document the new invariant that all related
26276 strinfos have delayed lengths or none do.
26277 (verify_related_strinfos): Move earlier in file.
26278 (set_endptr_and_length): New function, split out from...
26279 (get_string_length): ...here. Also set the lengths of related
26281 (zero_length_string): Assert that chainsi has known (rather than
26283 (adjust_related_strinfos): Likewise.
26285 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26287 PR tree-optimization/81136
26288 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26289 assert that two references with the same misalignment have the same
26290 compile-time misalignment if those compile-time misalignments
26293 2017-07-01 Andi Kleen <ak@linux.intel.com>
26295 * print-tree.c (print_node): Print all attributes.
26297 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26299 * cfg.c (scale_bbs_frequencies): New function.
26300 * cfg.h (scale_bbs_frequencies): Declare it.
26301 * cfgloopanal.c (single_likely_exit): Cleanup.
26302 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26304 (scale_loop_profile): Likewise.
26305 (loop_version): Likewise.
26306 (create_empty_loop_on_edge): Update.
26307 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26308 scale_loop_frequencies, scale_loop_profile, loopify,
26309 loop_version): Update prototypes.
26310 * modulo-sched.c (sms_schedule): Update.
26311 * predict.c (unlikely_executed_edge_p): Also check probability.
26312 (probably_never_executed_edge_p): Fix typo.
26313 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26314 * tree-parloops.c (gen_parallel_loop): Update.
26315 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26316 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26317 * tree-ssa-loop-split.c (split_loop): Update.
26318 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26319 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26320 (vect_loop_versioning): Update.
26321 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26323 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26325 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26327 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26329 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26330 to keep profile consistent.
26332 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26334 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26335 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26336 * profile-count.h (max_safe_multiplier): Make unsigned.
26337 (profile_count::guessed_zero): New.
26339 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26341 * bb-reorder.c (fix_up_crossing_landing_pad,
26342 fix_crossing_conditional_branches): Use make_single_succ_edge
26343 to keep profile consistent.
26345 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26347 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26350 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26353 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26354 the right scopes, make sure cond_jump isn't preserved between multiple
26355 iterations. Search for fallthru edge whenever there are 3+ edges and
26356 use find_fallthru_edge for it.
26358 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26360 Patch by Alexander Monakov <amonakov@ispras.ru>
26361 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26362 probabilities consistently.
26364 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26366 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26367 emit_cmp_and_jump_insns.
26369 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26372 * tree-inline.c (expand_call_inline): Combine profile statuses.
26374 2017-06-30 Andrew Pinski <apinski@cavium.com>
26376 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26377 fold_stmt returned true.
26379 2017-06-30 Nathan Sidwell <nathan@acm.org>
26381 * ggc.h (empty_string): Delete.
26382 * cfgexpand.c (expand_asm_stmt): Use plain "".
26383 * optabs.c (expand_asm_memory_barrier): Likewise.
26384 * stringpool.c (empty_string): Delete.
26385 (digit_vector, digit_string): Delete.
26386 (ggc_alloc_string): Use plain "", don't optimize single digit
26387 strings. Use ggc_alloc_atomic.
26389 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26391 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26392 comparison set and one other set, use the cost of the non-comparison
26395 2017-06-30 Nathan Sidwell <nathan@acm.org>
26397 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26400 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26402 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26403 loops. Remove now unneeded calls to gimple_switch_set_label() that
26404 just set removed labels to NULL_TREE.
26406 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26408 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26409 (set_range_info): ...here. Only call set_range_info_raw if domain
26411 (set_nonzero_bits): Call set_range_info_raw.
26412 * tree-ssanames.h (set_range_info_raw): New.
26414 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26417 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26418 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26419 of nonimmediate_operand and <store_mask_constraint> instead of m
26420 for the input operand. For V8FI iterator, always split if input
26421 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26422 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26423 <store_mask_predicate> instead of register_operand and
26424 <store_mask_constraint> instead of v for the input operand. Make
26425 sure both operands aren't MEMs for if not <mask_applied>.
26427 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26429 * lto-wrapper.c (copy_file) Close both file descriptors before
26432 2017-06-30 Martin Liska <mliska@suse.cz>
26435 * multiple_target.c (create_dispatcher_calls): Make ifunc
26436 also for function that don't have calls or are not referenced.
26438 2017-06-30 Richard Biener <rguenther@suse.de>
26440 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26441 analyze the first scalar stmt. Move vector type computation
26442 for the BB case here from ...
26443 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26444 live operation processing in the SLP case properly.
26446 2017-06-30 Richard Biener <rguenther@suse.de>
26448 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26450 2017-06-30 Martin Liska <mliska@suse.cz>
26453 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26454 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26456 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26458 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26461 2017-06-30 Richard Biener <rguenther@suse.de>
26463 PR tree-optimization/81249
26464 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26465 condition reduction result to original scalar type.
26467 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26469 * profile-count.h (enum profile_quality): Fix typos and whitespace
26472 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26474 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26475 type for branch probabilities.
26477 2017-06-29 Julian Brown <julian@codesourcery.com>
26478 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26480 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26481 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26482 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26483 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26485 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26487 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26488 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26489 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26490 CC usage from generic code to here.
26491 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26492 CC usage into the target macros.
26494 2017-06-29 Maya Rashish <coypu@sdf.org>
26496 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26499 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26501 * arm/arm-builtins.c: Include profile-count.h
26502 * except.c (sjlj_emit_function_enter): Use
26503 profile_probability::unlikely.
26505 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26507 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26508 and tocrel_offset be pointer args rather than implicitly using
26510 (legitimate_constant_pool_address_p, rs6000_emit_move,
26511 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26512 tocrel_offset and use in toc_relative_expr_p call.
26513 (print_operand, print_operand_address): Use static tocrel_base_oac
26514 and tocrel_offset_oac.
26515 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26518 2017-06-29 Maya Rashish <coypu@sdf.org>
26520 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26522 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26524 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26525 objects, take into account only the alignment of 'op0' and 'mode1' if
26528 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26530 * ccmp.c (ccmp_tree_comparison_p): New function.
26531 (ccmp_candidate_p): Update to use above function.
26532 (get_compare_parts): New function.
26533 (expand_ccmp_next): Update to use new functions.
26534 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26536 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26537 take mode as argument.
26538 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26539 * expr.c (expand_expr_real_1): Pass mode as argument.
26541 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26543 * combine.c (combine_instructions): Print insns to dump_file, together
26546 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26548 * asan.c (asan_emit_stack_protection): Update.
26549 (create_cond_insert_point): Update.
26550 * auto-profile.c (afdo_propagate_circuit): Update.
26551 * basic-block.h (struct edge_def): Turn probability to
26552 profile_probability.
26553 (EDGE_FREQUENCY): Update.
26554 * bb-reorder.c (find_traces_1_round): Update.
26555 (better_edge_p): Update.
26556 (sanitize_hot_paths): Update.
26557 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26558 (make_single_succ_edge): Update.
26559 (check_bb_profile): Update.
26560 (dump_edge_info): Update.
26561 (update_bb_profile_for_threading): Update.
26562 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26563 probabilitycount to 0.
26564 * cfgbuild.c (compute_outgoing_frequencies): Update.
26565 * cfgcleanup.c (try_forward_edges): Update.
26566 (outgoing_edges_match): Update.
26567 (try_crossjump_to_edge): Update.
26568 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26569 (expand_gimple_tailcall): Update.
26570 (construct_init_block): Use make_single_succ_edge.
26571 (construct_exit_block): Use make_single_succ_edge.
26572 * cfghooks.c (verify_flow_info): Update.
26573 (redirect_edge_succ_nodup): Update.
26574 (split_edge): Update.
26575 (account_profile_record): Update.
26576 * cfgloopanal.c (single_likely_exit): Update.
26577 * cfgloopmanip.c (scale_loop_profile): Update.
26578 (set_zero_probability): Remove.
26579 (duplicate_loop_to_header_edge): Update.
26580 * cfgloopmanip.h (loop_version): Update prototype.
26581 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26582 (force_nonfallthru_and_redirect): Update.
26583 (update_br_prob_note): Update.
26584 (rtl_verify_edges): Update.
26585 (purge_dead_edges): Update.
26586 (rtl_lv_add_condition_to_bb): Update.
26587 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26588 * cgraphunit.c (init_lowered_empty_function): Update.
26589 (cgraph_node::expand_thunk): Update.
26590 * cilk-common.c: Include profile-count.h
26591 * dojump.c (inv): Remove.
26592 (jumpifnot): Update.
26593 (jumpifnot_1): Update.
26594 (do_jump_1): Update.
26596 (do_jump_by_parts_greater_rtx): Update.
26597 (do_compare_rtx_and_jump): Update.
26598 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26599 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26600 * dwarf2cfi.c: Include profile-count.h
26601 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26602 (sjlj_emit_dispatch_table): Likewise.
26603 * explow.c: Include profile-count.h
26604 * expmed.c (emit_store_flag_force): Update.
26605 (do_cmp_and_jump): Update.
26606 * expr.c (compare_by_pieces_d::generate): Update.
26607 (compare_by_pieces_d::finish_mode): Update.
26608 (emit_block_move_via_loop): Update.
26609 (store_expr_with_bounds): Update.
26610 (store_constructor): Update.
26611 (expand_expr_real_2): Update.
26612 (expand_expr_real_1): Update.
26613 * expr.h (try_casesi, try_tablejump): Update prototypes.
26614 * gimple-pretty-print.c (dump_probability): Update.
26615 (dump_profile): New.
26616 (dump_gimple_label): Update.
26617 (dump_gimple_bb_header): Update.
26618 * graph.c (draw_cfg_node_succ_edges): Update.
26619 * hsa-gen.c (convert_switch_statements): Update.
26620 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26621 (find_if_case_1): Update.
26622 (find_if_case_2): Update.
26623 * internal-fn.c (expand_arith_overflow_result_store): Update.
26624 (expand_addsub_overflow): Update.
26625 (expand_neg_overflow): Update.
26626 (expand_mul_overflow): Update.
26627 (expand_vector_ubsan_overflow): Update.
26628 * ipa-cp.c (good_cloning_opportunity_p): Update.
26629 * ipa-split.c (split_function): Use make_single_succ_edge.
26630 * ipa-utils.c (ipa_merge_profiles): Update.
26631 * loop-doloop.c (add_test): Update.
26632 (doloop_modify): Update.
26633 * loop-unroll.c (compare_and_jump_seq): Update.
26634 (unroll_loop_runtime_iterations): Update.
26635 * lra-constraints.c (lra_inheritance): Update.
26636 * lto-streamer-in.c (input_cfg): Update.
26637 * lto-streamer-out.c (output_cfg): Update.
26638 * mcf.c (adjust_cfg_counts): Update.
26639 * modulo-sched.c (sms_schedule): Update.
26640 * omp-expand.c (expand_omp_for_init_counts): Update.
26641 (extract_omp_for_update_vars): Update.
26642 (expand_omp_ordered_sink): Update.
26643 (expand_omp_for_ordered_loops): Update.
26644 (expand_omp_for_generic): Update.
26645 (expand_omp_for_static_nochunk): Update.
26646 (expand_omp_for_static_chunk): Update.
26647 (expand_cilk_for): Update.
26648 (expand_omp_simd): Update.
26649 (expand_omp_taskloop_for_outer): Update.
26650 (expand_omp_taskloop_for_inner): Update.
26651 * omp-simd-clone.c (simd_clone_adjust): Update.
26652 * optabs.c (expand_doubleword_shift): Update.
26653 (expand_abs): Update.
26654 (emit_cmp_and_jump_insn_1): Update.
26655 (expand_compare_and_swap_loop): Update.
26656 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26657 * predict.c (predictable_edge_p): Update.
26658 (edge_probability_reliable_p): Update.
26659 (set_even_probabilities): Update.
26660 (combine_predictions_for_insn): Update.
26661 (combine_predictions_for_bb): Update.
26662 (propagate_freq): Update.
26663 (estimate_bb_frequencies): Update.
26664 (force_edge_cold): Update.
26665 * profile-count.c (profile_count::dump): Add missing space into dump.
26666 (profile_count::debug): Add newline.
26667 (profile_count::differs_from_p): Explicitly convert to unsigned.
26668 (profile_count::stream_in): Update.
26669 (profile_probability::dump): New member function.
26670 (profile_probability::debug): New member function.
26671 (profile_probability::differs_from_p): New member function.
26672 (profile_probability::differs_lot_from_p): New member function.
26673 (profile_probability::stream_in): New member function.
26674 (profile_probability::stream_out): New member function.
26675 * profile-count.h (profile_count_quality): Rename to ...
26676 (profile_quality): ... this one.
26677 (profile_probability): New.
26678 (profile_count): Update.
26679 * profile.c (compute_branch_probabilities): Update.
26680 * recog.c (peep2_attempt): Update.
26681 * sched-ebb.c (schedule_ebbs): Update.
26682 * sched-rgn.c (find_single_block_region): Update.
26683 (compute_dom_prob_ps): Update.
26684 (schedule_region): Update.
26685 * sel-sched-ir.c (compute_succs_info): Update.
26686 * stmt.c (struct case_node): Update.
26687 (do_jump_if_equal): Update.
26688 (get_outgoing_edge_probs): Update.
26689 (conditional_probability): Update.
26690 (emit_case_dispatch_table): Update.
26691 (expand_case): Update.
26692 (expand_sjlj_dispatch_table): Update.
26693 (emit_case_nodes): Update.
26694 * targhooks.c: Update.
26695 * tracer.c (better_p): Update.
26696 (find_best_successor): Update.
26697 * trans-mem.c (expand_transaction): Update.
26698 * tree-call-cdce.c: Update.
26699 * tree-cfg.c (gimple_split_edge): Upate.
26700 (move_sese_region_to_fn): Upate.
26701 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26702 * tree-eh.c (lower_resx): Upate.
26703 (cleanup_empty_eh_move_lp): Upate.
26704 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26705 * tree-inline.c (copy_edges_for_bb): Update.
26706 (copy_cfg_body): Update.
26707 * tree-parloops.c (gen_parallel_loop): Update.
26708 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26709 (gimple_gen_time_profiler): Update.
26710 * tree-ssa-dce.c (remove_dead_stmt): Update.
26711 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26712 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26713 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26714 (unloop_loops): Update.
26715 (try_peel_loop): Update.
26716 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26717 * tree-ssa-loop-split.c (connect_loops): Update.
26718 (split_loop): Update.
26719 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26720 (hoist_guard): Update.
26721 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26722 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26723 (value_replacement): Update.
26724 * tree-ssa-reassoc.c (branch_fixup): Update.
26725 * tree-ssa-tail-merge.c (replace_block_by): Update.
26726 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26727 (create_edge_and_update_destination_phis): Update.
26728 (compute_path_counts): Update.
26729 (recompute_probabilities): Update.
26730 (update_joiner_offpath_counts): Update.
26731 (freqs_to_counts_path): Update.
26732 (duplicate_thread_path): Update.
26733 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26734 (struct switch_conv_info): Update.
26735 (gen_inbound_check): Update.
26736 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26737 (vect_do_peeling): Update.
26738 (vect_loop_versioning): Update.
26739 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26740 (optimize_mask_stores): Update.
26741 * ubsan.c (ubsan_expand_null_ifn): Update.
26742 * value-prof.c (gimple_divmod_fixed_value): Update.
26743 (gimple_divmod_fixed_value_transform): Update.
26744 (gimple_mod_pow2): Update.
26745 (gimple_mod_pow2_value_transform): Update.
26746 (gimple_mod_subtract): Update.
26747 (gimple_mod_subtract_transform): Update.
26748 (gimple_ic): Update.
26749 (gimple_stringop_fixed_value): Update.
26750 (gimple_stringops_transform): Update.
26751 * value-prof.h: Update.
26753 2017-06-29 Carl Love <cel@us.ibm.com>
26755 * config/rs6000/rs6000-c.c: Add support for built-in functions
26756 vector signed int vec_signed (vector float);
26757 vector signed long long vec_signed (vector double);
26758 vector signed int vec_signed2 (vector double, vector double);
26759 vector signed int vec_signede (vector double);
26760 vector signed int vec_signedo (vector double);
26761 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26762 instruction generator.
26763 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26764 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26765 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26767 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26768 vunsignede_v2df): Add define_expands.
26769 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26770 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26771 VEC_UNSIGNEDO): Add definitions.
26772 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26773 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26774 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26775 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26776 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26777 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26778 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26779 vec_unsignede, vec_unsignedo): Add builtin defines.
26780 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26782 * doc/extend.texi: Update the built-in documentation file for the
26783 new built-in functions.
26785 2017-06-29 Richard Biener <rguenther@suse.de>
26787 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26788 reduction chains to LOOP_VINFO_REDUCTIONS.
26789 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26790 SLP reductions after processing reduction chains.
26792 2017-06-29 Nathan Sidwell <nathan@acm.org>
26794 * builtins.c (fold_builtin_FUNCTION): Use
26795 lang_hooks.decl_printable_name.
26797 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26799 PR middle-end/81194
26800 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26801 with only one label.
26802 * stmt.c (expand_case): Assert NCASES is greater than one.
26804 2017-06-29 Richard Biener <rguenther@suse.de>
26806 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26808 (group_case_labels): Likewise.
26809 (find_taken_edge): Push sanity checking on val to workers...
26810 (find_taken_edge_cond_expr): ... here
26811 (find_taken_edge_switch_expr): ... and here, handle cases
26812 with just a default label.
26813 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26814 (group_case_labels): Likewise.
26815 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26816 group_case_labels does anything cleanup the CFG again.
26818 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26820 PR tree-optimization/81196
26821 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26822 exit condition comparing two IVs.
26824 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26826 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26827 profile to the dummy entry at the end of the list of architectures.
26828 * config/arm/arm-cpu-cdata.h: Regenerated.
26830 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26831 Michael Collison <michael.collison@arm.com>
26834 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26836 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26837 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26838 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26839 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26840 of shift when the shift amount is masked with constant equal to
26841 the size of the mode.
26842 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26845 2017-06-29 Martin Liska <mliska@suse.cz>
26847 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26849 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26852 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26855 (edge_growth_cache_entry::edge_growth_cache_entry): New
26857 (reset_edge_growth_cache): Update to use constructor.
26859 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26861 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26862 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26863 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26865 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26867 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26868 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26870 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26872 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26873 (*-linux-uclibc*): Add t-uclibc tmake_file.
26874 * config/t-musl: New.
26875 * config/t-uclibc: New.
26877 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26879 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26881 (gen_comm_data): Emit architectural setting of arch_prof.
26882 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26884 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26885 (armv8-m.base, armv8-m.main): Likewise.
26886 * arm-protos.h (arm_build_target): Add profile field.
26887 (arch_option): Likewise.
26888 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26890 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26891 arm_active_target.profile.
26893 2017-06-28 Richard Biener <rguenther@suse.de>
26895 PR middle-end/81227
26896 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26897 TYPE_OVERFLOW_WRAPS.
26898 * match.pd (negate_expr_p): Likewise.
26899 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26900 fold_build2, not fold_binary.
26902 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26904 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26905 Convert memory address to Pmode.
26906 (aarch64_print_operand): Assert MEM operands are always Pmode.
26908 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26911 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26912 Remove redundant if.
26913 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26914 * config/arm/aarch-common-protos.h
26915 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26916 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26918 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26921 * multiple_target.c (create_dispatcher_calls): Set the default
26922 clone to be static, not public.
26924 2017-06-28 Richard Biener <rguenther@suse.de>
26926 * tree-vect-loop.c (vectorizable_reduction): Move special
26927 cond reduction IV var creation ...
26928 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26929 parameter. Use STMT_VINFO_VECTYPE.
26930 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26933 2017-06-28 Martin Liska <mliska@suse.cz>
26936 * ipa-visibility.c (non_local_p): Handle visibility.
26938 2017-06-28 Martin Liska <mliska@suse.cz>
26941 * common.opt: Add IntegerRange to various options.
26942 * opt-functions.awk (integer_range_info): New function.
26943 * optc-gen.awk: Add integer_range_info to cl_options struct.
26944 * opts-common.c (decode_cmdline_option): Handle
26945 CL_ERR_INT_RANGE_ARG.
26946 (cmdline_handle_error): Likewise.
26947 * opts.c (print_filtered_help): Show valid interval in
26948 when --help is provided.
26949 * opts.h (struct cl_option): Add range_min and range_max fields.
26950 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26952 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26954 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26955 (x * C EQ/NE y * C): New transformation.
26957 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26959 * genmultilib (combination_space): Accept '+' in option names.
26961 2017-06-28 Martin Liska <mliska@suse.cz>
26964 * asan.c (instrument_derefs): Bail out inner references
26965 that are hard register variables.
26967 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26970 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26971 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26973 2017-06-28 Richard Biener <rguenther@suse.de>
26975 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26976 (vect_get_slp_defs): Adjust.
26977 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26979 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26981 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26982 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26983 (vectorizable_reduction): Adjust.
26984 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26986 (vect_get_slp_defs): Likewise.
26987 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26988 (vectorizable_bswap): Adjust.
26989 (vectorizable_call): Likewise.
26990 (vectorizable_conversion): Likewise.
26991 (vectorizable_assignment): Likewise.
26992 (vectorizable_shift): Likewise.
26993 (vectorizable_operation): Likewise.
26994 (vectorizable_store): Likewise.
26995 (vectorizable_condition): Likewise.
26996 (vectorizable_comparison): Likewise.
26998 2017-06-28 Michael Collison <michael.collison@arm.com>
27001 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
27003 (arm_gen_movmemqi): Removed unused variable 'i'.
27004 Convert 'for' loop into 'while' loop.
27005 (arm_expand_prologue): Remove last unnecessary set of insn.
27006 (thumb_pop): Remove unused variable 'pushed_words'.
27007 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27009 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27011 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27012 * config/s390/s390-protos.h: Add prototype of
27013 s390_rel_address_ok_p.
27014 * config/s390/s390.c (s390_got_symbol): New function.
27015 (s390_rel_address_ok_p): New function.
27016 (legitimize_pic_address): Use s390_rel_address_ok_p.
27017 (s390_load_got): Use s390_got_symbol.
27018 (s390_option_override): Issue error if
27019 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27020 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27022 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27024 2017-06-27 Andrew Pinski <apinski@cavium.com>
27026 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27027 (X * copysign (1.0, X)): New pattern.
27028 (X * copysign (1.0, -X)): New pattern.
27029 (copysign (-1.0, CST)): New pattern.
27031 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27033 * genmultilib (combination_space): Remove variable.
27034 Validate reuse rules against regular expression for any sequence
27035 of multilib options in any order.
27037 2017-06-27 Michael Collison <michael.collison@arm.com>
27039 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27040 call aarch64_split_simd_combine.
27041 * (aarch64_combine_internal<mode>): Delete pattern.
27042 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27043 Allow register and subreg operands.
27045 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27047 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27048 specific need, just fallback on defaults.
27049 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27051 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27052 Olivier Hainque <hainque@adacore.com>
27054 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27056 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27057 targets. Pick a default if no particular attempt applied.
27058 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27061 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27063 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27064 (x86_64-wrs-vxworks7): Likewise.
27066 2017-06-27 Marek Polacek <polacek@redhat.com>
27069 * ubsan.c (instrument_null): Check get_base_address's result for null.
27071 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27073 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27075 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27077 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27078 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27079 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27080 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27081 New function types.
27082 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27083 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27084 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27085 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27086 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27087 BUILT_IN_FEUPDATEENV): New builtins.
27088 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27089 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27090 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27091 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27093 (builtin_structptr_types): Adjust size.
27094 * tree.c (builtin_structptr_types): Add four entries.
27096 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27097 Olivier Hainque <hainque@adacore.com>
27099 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27100 (TLS_SYM): New local macro, forcing reference to __tls__ on
27101 link command lines for VxWorks 7 RTPs, triggering initialization
27103 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27104 OS features TLS support, true for RTPs on VxWorks 7.
27105 * config/vxworks.c (vxworks_override_options): Setup emutls
27108 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27110 * predict.c (test_prediction_value_range): Use -1U instead of -1
27111 to avoid narrowing conversion warning.
27112 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27113 to avoid narrowing conversion warning.
27114 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27116 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27118 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27120 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27121 64bit configurations.
27122 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27123 (SIZE_TYPE): Likewise.
27124 * config/vxworks.c (vxworks_emutls_var_fields): Use
27125 long_unsigned_type_node instead of unsigned_type_node as the offset
27126 field type, which is "pointer" mode in emutls.c.
27128 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27131 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27133 PR middle-end/81207
27134 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27135 gimple_vuse copying separately from gimple_vdef copying.
27137 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27139 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27142 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27143 Olivier Hainque <hainque@adacore.com>
27145 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27146 for all vxworks7 targets.
27147 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27148 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27149 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27150 variations for VX6/VX7 and 32/64bits later on in ...
27151 (VXWORKS_LIB_SPEC): Leverage new macros.
27152 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27153 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27155 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27157 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27158 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27160 2017-06-26 Carl Love <cel@us.ibm.com>
27162 * config/rs6000/rs6000-c.c: Add support for built-in functions
27163 vector bool char vec_reve (vector bool char);
27164 vector signed char vec_reve (vector signed char);
27165 vector unsigned char vec_reve (vector unsigned char);
27166 vector bool int vec_reve (vector bool int);
27167 vector signed int vec_reve (vector signed int);
27168 vector unsigned int vec_reve (vector unsigned int);
27169 vector bool long long vec_reve (vector bool long long);
27170 vector signed long long vec_reve (vector signed long long);
27171 vector unsigned long long vec_reve (vector unsigned long long);
27172 vector bool short vec_reve (vector bool short);
27173 vector signed short vec_reve (vector signed short);
27174 vector double vec_reve (vector double);
27175 vector float vec_reve (vector float);
27176 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27177 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27178 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27179 (altivec_vreve): New pattern.
27180 * config/rs6000/altivec.h (vec_reve): New define.
27181 * doc/extend.texi (vec_rev): Update the built-in documentation file
27182 for the new built-in functions.
27184 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27186 PR tree-optimization/71815
27187 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27189 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27191 (slsr_process_phi): Likewise.
27192 (replace_uncond_cands_and_profitable_phis): Don't replace a
27193 multiply candidate with a stride of 1 (copy or cast).
27194 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27196 (lowest_cost_path): Likewise.
27197 (total_savings): Likewise.
27199 2017-06-26 Richard Biener <rguenther@suse.de>
27202 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27203 Use def_builtin_pure for all gather builtins.
27205 2017-06-26 Richard Biener <rguenther@suse.de>
27207 PR tree-optimization/81203
27208 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27209 non-dominating BBs.
27211 2017-06-26 Marek Polacek <polacek@redhat.com>
27214 * doc/invoke.texi: Document -Wmultistatement-macros.
27216 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27218 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27219 arm_neon_ok_no_float_abi effective target.
27221 2017-06-26 Richard Biener <rguenther@suse.de>
27223 PR tree-optimization/80928
27224 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27225 (copy_bbs): Set BB_DUPLICATED flag early.
27226 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27228 (execute_on_shrinking_pred): Likewise.
27229 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27230 BB_DUPLICATED blocks.
27231 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27232 iterate over all PHIs considering removal of *gsi.
27234 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27236 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27239 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27241 * config/rs6000/rs6000-string.c: (expand_block_clear,
27242 do_load_for_compare, select_block_compare_mode,
27243 compute_current_alignment, expand_block_compare,
27244 expand_strncmp_align_check, expand_strn_compare,
27245 expand_block_move, rs6000_output_load_multiple)
27246 Move functions related to string/block move/compare
27247 to a separate file.
27248 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27249 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27250 for this function which is now used in two files.
27251 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27252 * config.gcc: Add rs6000-string.o to extra_objs for
27253 targets powerpc*-*-* and rs6000*-*-*.
27255 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27258 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27259 32-bit, since indexed is not valid for DImode.
27260 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27261 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27262 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27263 (define_peephole2 for Altivec d-form store): Likewise.
27266 * multiple_target.c (create_dispatcher_calls): Only create the
27267 dispatcher call if the function is the default clone of a
27268 versioned function.
27270 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27272 PR middle-end/80902
27273 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27274 a call, force the call to not be a tail call.
27276 2017-06-23 Jeff Law <law@redhat.com>
27278 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27281 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27283 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27284 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27285 handling for early expansion of vector shifts (sl,sr,sra,rl).
27286 (builtin_function_type): Add vector shift right instructions
27287 to the unsigned argument list.
27289 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27291 rtl-optimizatoin/79286
27292 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27293 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27294 trap. PIC register plus a const unspec without offset can never trap.
27296 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27298 * tree.h (builtin_structptr_type): New type.
27299 (builtin_structptr_types): Declare new array.
27300 * tree.c (builtin_structptr_types): New array.
27301 (free_lang_data, build_common_tree_nodes): Use it.
27303 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27306 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27309 2017-06-22 Matt Turner <mattst88@gmail.com>
27311 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27312 Lake models to skylake case. Assume skylake for unknown
27313 models with clflushopt.
27315 2017-06-22 Jeff Law <law@redhat.com>
27317 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27318 frame sizes that do not satisfy aarch64_uimm12_shift.
27320 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27322 * profile-count.h (apply_probability,
27323 apply_scale, probability_in): Fix checks for zero.
27325 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27327 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27328 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27330 2016-06-22 Richard Biener <rguenther@suse.de>
27332 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27333 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27334 REDUC_MAX_EXPR support.
27335 (vectorizable_reduction): Likewise.
27336 (vect_create_epilog_for_reduction): Likewise.
27338 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27340 * match.pd (A / (1 << B) -> A >> B): New.
27341 * generic-match-head.c: Include optabs-tree.h.
27342 * gimple-match-head.c: Likewise.
27343 * optabs-tree.h (target_supports_op_p): New.
27344 * optabs-tree.c (target_supports_op_p): New.
27346 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27348 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27349 $gcc_cv_ld --help output.
27350 (gcc_cv_ld_demangle): Likewise.
27351 (gcc_cv_ld_eh_frame_hdr): Likewise.
27352 (gcc_cv_ld_pie): Likewise.
27353 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27354 (gcc_cv_ld_buildid): Likewise.
27355 (gcc_cv_ld_sysroot): Likewise.
27356 (ld_bndplt_support): Likewise.
27357 (ld_pushpopstate_support): Likewise.
27358 * configure: Regenerate.
27359 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27361 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27364 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27365 operands indexes to avoid gap between operands and match_dups.
27367 2017-06-21 Andrew Pinski <apinski@cavium.com>
27369 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27370 Increment Arith_shift and Arith_shift_reg by 1.
27371 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27373 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27374 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27375 (aarch64_strip_extend): Add new argument and test for it.
27376 (aarch64_cheap_mult_shift_p): New function.
27377 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27378 add a cost if it is true.
27379 Update calls to aarch64_strip_extend.
27380 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27382 2017-06-21 Andrew Pinski <apinski@cavium.com>
27384 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27386 (thunderxt88): Likewise.
27387 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27388 (thunderx_prefetch_tune): New variable.
27389 (thunderx2t99_prefetch_tune): Update for the correct values.
27390 (thunderxt88_tunings): New variable.
27391 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27392 generic_prefetch_tune.
27393 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27395 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27397 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27398 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27399 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27400 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27401 (aarch64_atomic_cas<mode>, GPI): Likewise.
27403 2017-06-21 Martin Liska <mliska@suse.cz>
27405 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27406 statements on cold and hot labels.
27407 * predict.c (tree_estimate_probability_bb): Remove the
27408 prediction from this place.
27410 2017-06-21 Martin Liska <mliska@suse.cz>
27412 PR tree-optimization/79489
27413 * gimplify.c (maybe_add_early_return_predict_stmt): New
27415 (gimplify_return_expr): Call the function.
27416 * predict.c (tree_estimate_probability_bb): Remove handling
27418 * predict.def: Update comment about early return predictor.
27419 * gimple-predict.h (is_gimple_predict): New function.
27420 * predict.def: Change default value of early return to 66.
27421 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27423 * passes.def: Put pass_strip_predict_hints to the beginning of
27426 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27428 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27429 FUNCTION_DECL declarations.
27430 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27432 (dwaf2out_decl): Likewise.
27433 * godump.c (go_early_global_decl): Skip call to the real debug hook
27434 for FUNCTION_DECL declarations.
27435 * passes.c (rest_of_decl_compilation): Skip call to the
27436 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27437 -fdump-go-spec is passed.
27439 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27441 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27442 Reorder for compactness.
27443 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27444 (def_builtin_pure, def_builtin_pure2): New functions.
27445 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27447 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27449 * match.pd (nop_convert): New predicate.
27450 ((A +- CST1) +- CST2): Allow some NOP conversions.
27452 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27455 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27456 with ctors/dtors if GOVD_SHARED is set.
27458 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27460 * config/aarch64/aarch64.md (movti_aarch64):
27461 Emit mov rather than orr.
27462 (movtf_aarch64): Likewise.
27463 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27464 Emit mov rather than orr.
27466 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27468 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27469 Swap alternatives, make integer dup more expensive.
27471 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27473 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27474 Return true for non-tls symbols.
27476 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27478 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27479 (cortex-a75): Likewise.
27480 (cortex-a75.cortex-a55): Likewise.
27481 * config/aarch64/aarch64-tune.md: Regenerate.
27482 * doc/invoke.texi (-mtune): Document new values for -mtune.
27484 2017-06-21 Tom de Vries <tom@codesourcery.com>
27486 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27487 stack_size feature.
27488 (Effective-Target Keywords, Other attributes): Suggest using
27489 dg-add-options stack_size feature to get stack limit in stack_size
27490 effective target documentation.
27492 2017-06-21 Julian Brown <julian@codesourcery.com>
27493 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27495 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27496 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27497 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27499 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27500 attribute type list for neon_multiply.
27501 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27502 attribute type list for neon_multiply.
27503 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27504 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27505 attribute type list for neon_multiply.
27506 * config/arm/types.md (crypto_pmull): Add.
27507 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27508 attribute type list.
27510 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27512 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27515 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27517 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27518 to make sure not to dereference a NULL cost_classes_ptr pointer.
27520 2017-06-20 Carl Love <cel@us.ibm.com>
27522 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27523 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27524 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27525 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27526 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27527 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27528 VMULOSW): New enum "unspec" values.
27529 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27530 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27531 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27532 altivec_vmulosw): New patterns.
27533 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27534 VMULOSW): Add definitions.
27536 2017-06-20 Julia Koval <julia.koval@intel.com>
27538 * config/i386/i386.c: Fix rounding expand for new pattern.
27539 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27541 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27543 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27544 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27546 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27548 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27551 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27553 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27554 architecture, then by implementer ID.
27555 * config/aarch64/aarch64-tune.md: Regenerate.
27557 2017-06-20 Richard Biener <rguenther@suse.de>
27559 PR middle-end/81097
27560 * fold-const.c (split_tree): Fold to type before negating.
27562 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27564 * diagnostic-show-locus.c
27565 (selftest::test_fixit_deletion_affecting_newline): New function.
27566 (selftest::diagnostic_show_locus_c_tests): Call it.
27568 2017-06-20 Andreas Schwab <schwab@suse.de>
27571 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27574 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27576 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27577 __ARM_FEATURE_COPROC according to support.
27579 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27581 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27582 Rewritten to avoid overflow for > 32-bit pointers.
27585 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27586 by removing enum keyword.
27587 (ubsan_type_descriptor): Likewise. Formatting fix.
27590 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27591 splitter): Require TARGET_SSE2 in the condition.
27593 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27596 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27597 for doing vector set of SFmode on ISA 3.0.
27598 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27599 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27601 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27602 SFmode value into a V4SF variable that was extracted from another
27603 V4SF variable without converting the element to double precision
27604 and back to single precision vector format.
27605 (vsx_insert_extract_v4sf_p9_2): Likewise.
27607 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27609 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27610 in UWHI to avoid undefined overflow.
27613 * ubsan.h (enum ubsan_encode_value_phase): New.
27614 (ubsan_encode_value): Change second argument to
27615 enum ubsan_encode_value_phase with default value of
27616 UBSAN_ENCODE_VALUE_GENERIC.
27617 * ubsan.c (ubsan_encode_value): Change second argument to
27618 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27619 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27620 create_tmp_var_raw instead of create_tmp_var and use a
27622 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27623 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27624 ubsan_encode_value callers.
27627 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27628 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27629 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27631 2017-06-19 Richard Biener <rguenther@suse.de>
27633 PR middle-end/81118
27634 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27635 estimates if we changed anything.
27637 2017-06-19 Richard Biener <rguenther@suse.de>
27639 PR tree-optimization/80887
27640 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27641 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27642 simplified lookups, then reset mprts_hook.
27643 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27645 (try_to_simplify): Likewise.
27647 2017-06-19 Martin Liska <mliska@suse.cz>
27650 * gimplify.c (gimplify_switch_expr):
27651 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27653 2017-06-19 Martin Liska <mliska@suse.cz>
27655 * doc/install.texi: Document that PGO runs in 4 stages.
27657 2017-06-19 Martin Liska <mliska@suse.cz>
27660 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27661 to dispatcher function name.
27662 * multiple_target.c (replace_function_decl): New function.
27663 (create_dispatcher_calls): Redirect both edges and references.
27665 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27667 * profile-count.c (profile_count::dump): Dump quality.
27668 (profile_count::differs_from_p): Update for unsigned val.
27669 * profile-count.h (profile_count_quality): New enum.
27670 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27672 2017-06-19 Richard Biener <rguenther@suse.de>
27674 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27675 struct function as arg.
27676 (estimate_numbers_of_iterations): Export overload with loop arg.
27677 (free_numbers_of_iterations_estimates_loop): Use an overload of
27678 free_numbers_of_iterations_estimates instead.
27679 * tree-cfg.c (remove_bb): Adjust.
27680 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27681 * tree-parloops.c (gen_parallel_loop): Likewise.
27682 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27684 (tree_unroll_loops_completely): Likewise.
27685 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27686 Use an overload instead and export.
27687 (estimated_loop_iterations): Adjust.
27688 (max_loop_iterations): Likewise.
27689 (likely_max_loop_iterations): Likewise.
27690 (estimate_numbers_of_iterations): Take struct function as arg
27692 (loop_exits_before_overflow): Adjust.
27693 (free_numbers_of_iterations_estimates_loop): Use an overload.
27694 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27695 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27697 2017-06-19 Richard Biener <rguenther@suse.de>
27700 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27701 RANGE_EXPR conservatively.
27703 2017-06-16 Carl Love <cel@us.ibm.com>
27705 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27706 definitions for vec_float, vec_float2, vec_floato,
27707 vec_floate built-ins.
27708 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27709 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27711 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27712 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27713 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27714 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27715 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27716 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27717 vec_floato): Add builtin defines.
27718 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27719 Update the built-in documentation file for the new built-in
27722 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27724 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27725 (mthumb): Mark as the negative of -marm.
27727 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27729 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27731 (ARM Options, -mtune): Document that this accepts the same
27732 extension options as -mcpu.
27733 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27735 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27737 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27738 permitted extensions.
27740 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27742 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27743 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27744 (armv8-m.main): Add option +nodsp.
27745 * config/arm/arm-cpu-cdata.h: Regenerated.
27747 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27749 * config/arm/t-fuchsia: New file.
27750 * config.gcc (arm*-*-fuchsia*): Use it.
27752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27754 * config/arm/t-symbian: Rewrite for new option infrastructure.
27756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27758 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27759 (MULTILIB_REQUIRED): Likewise.
27761 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27763 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27764 (MULTILIB_RESUE): Likewise.
27765 (MULTILIB_MATCHES): Likewise.
27766 (MULTLIB_REQUIRED): Likewise.
27768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27770 * config/arm/t-rtems: Rewrite for new option framework.
27772 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27774 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27775 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27776 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27777 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27778 * config/arm/t-multilib: ... here.
27779 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27780 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27781 armv7-a and armv8*-a when A-profile libraries have not been built.
27782 * config/arm/t-rmprofile: Rewrite.
27784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27786 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27787 with a backslash. Remove the backslash after substituting unescaped
27789 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27791 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27793 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27794 the driver to pass through the default setting of -mfloat-abi.
27795 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27797 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27798 (all_feat_combs): New rule.
27799 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27801 * config/arm/t-aprofile: Rewrite.
27803 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27805 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27806 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27807 fpu is not specified by the user/command-line.
27808 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27809 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27810 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27811 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27812 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27813 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27817 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27818 * config/arm/t-arm-elf: Rewritten.
27820 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27822 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27823 have some floating-point instructions.
27824 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27825 (TARGET_MAYBE_HARD_FLOAT): New macro.
27826 * config/arm/arm-builtins.c (arm_init_builtins): Use
27827 TARGET_MAYBE_HARD_FLOAT.
27828 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27830 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27832 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27833 (configargs.h): Include it.
27834 (arm_print_hint_for_fpu_option): New function.
27835 (arm_parse_fpu_option): New function.
27836 (candidate_extension): New class.
27837 (arm_canon_for_multilib): New function.
27838 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27839 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27840 (ARCH_CANONICAL_SPECS): New macro.
27841 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27845 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27846 are set after handling multilib fragments. Set target_cpu_default2
27849 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27851 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27853 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27854 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27856 (check_arch): Likewise.
27857 * config/arm/arm.c (arm_configure_build_target): Handle
27858 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27859 options in the default.
27861 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27863 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27864 when an option is an alias of another.
27865 * config/arm/parsecpu.awk (optalias): New parser token.
27866 (gen_comm_data): Mark non-alias options as such. Emit entries
27867 for extension aliases.
27868 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27869 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27870 (armv6kz, armv6zk, armv6t2): Likewise.
27871 (armv7): Make vfpv3-d16 an alias.
27872 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27874 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27875 Sort in canonical order.
27876 (armv8-a): Sort in canonical order.
27877 (armv8.1-a, armv8.2-a): Likewise.
27878 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27880 (cortex-a9): Sort in canonical order.
27881 * config/arm/arm.c (selftests.h): Include it.
27882 (arm_test_cpu_arch_data): New function.
27883 (arm_run_self_tests): New function.
27884 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27885 (targetm): Move declaration to the end of the file.
27886 * arm-cpu-cdata.h: Regenerated.
27888 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27890 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27891 call to target_mode_check describing the type of option passed.
27892 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27893 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27894 arm_parse_cpu_option_name to match parameters against list of
27896 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27897 arm_arch_core_flags data structure.
27898 * config/arm/arm-cpu_cdata.h: Regenerated.
27900 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27902 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27904 (arm_print_hint_for_cpu_option): Likewise.
27905 (arm_print_hint_for_arch_option): Likewise.
27906 (arm_parse_cpu_option_name): Likewise.
27907 (arm_parse_arch_option_name): Likewise.
27908 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27909 of entries in the all_fpus list.
27910 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27911 (arm_parse_cpu_option_name): Declare.
27912 (arm_parse_arch_option_name): Declare.
27913 (arm_parse_option_features): Declare.
27914 (arm_intialize_isa): Declare.
27915 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27917 (gen_comm_data): ... here. Make definitions non-static.
27918 * config/arm/arm-cpu-data.h: Regenerated.
27919 * config/arm/arm-cpu-cdata.h: Regenerated.
27921 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27923 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27924 (cpu_arch_extension): New structure.
27925 (cpu_arch_option, arch_option, cpu_option): New structures.
27926 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27927 architecture types.
27928 (gen_data): Generate new format data tables.
27929 * config/arm/arm.c (cpu_tune): New structure.
27930 (cpu_option, processors): Delete.
27931 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27932 (arm_print_hint_for_cpu_option): ... this and ...
27933 (arm_print_hint_for_arch_option): ... this.
27934 (arm_parse_arch_cpu_name): Delete. Replace with ...
27935 (arm_parse_cpu_option_name): ... this and ...
27936 (arm_parse_arch_option_name): ... this.
27937 (arm_unrecognized_feature): Change type of target parameter to
27939 (arm_parse_arch_cpu_features): Delete. Replace with ...
27940 (arm_parse_option_features): ... this.
27941 (arm_configure_build_target): Rework to use new configuration data
27943 (arm_print_tune_info): Rework for new configuration data tables.
27944 * config/arm/arm-cpu-data.h: Regenerated.
27945 * config/arm/arm-cpu.h: Regenerated.
27947 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27949 * Makefile.in (OBJS): Move sbitmap.o from here ...
27950 (OBJS-libcommon): ... to here.
27952 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27954 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27955 (ISA_ALL_CRYPTO): New macro.
27956 (ISA_ALL_SIMD): New macro
27957 (ISA_ALL_FP): New macro.
27958 * config/arm/arm.c (fpu_bitlist): Update initializer.
27959 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27961 (arm9e): Add fpu. Add option for nofp
27962 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27963 (arm926ej-s, arm1026ej-s): Likewise.
27964 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27965 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27966 neon-fp16, neon-vfpv4, nofp and nosimd.
27967 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27968 (cortex-a8): Add fpu. Add option for nofp.
27969 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27970 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27971 (cortex-r4f): Add fpu.
27972 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27973 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27975 (cortex-r8): Likewise.
27976 (cortex-m4): Add fpu. Add option for nofp.
27977 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27978 (cortex-a17.cortex-a7): Likewise.
27979 (cortex-a32): Add fpu. Add options for crypto and nofp.
27980 (cortex-a35, cortex-a53): Likewise.
27981 (cortex-a57): Add fpu. Add option for crypto.
27982 (cortex-a72, cortex-a73): Likewise.
27983 (exynos-m1): Likewise.
27984 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27985 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27986 (cortex-m33): Add fpu. Add option for nofp.
27987 * config/arm/arm-cpu-cdata.h: Regenerated
27988 * config/arm/arm-cpu-data.h: Regenerated.
27990 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27992 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27993 (armv5te, armv5tej): Likewise.
27994 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27995 (armv7): Add options fp and vfpv3-d16.
27996 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27997 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27999 (armv7ve): Likewise.
28000 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
28001 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
28002 (armv8-a): Add nocrypto option.
28003 (armv8.1-a, armv8.2-a): Likewise.
28004 (armv8-m.main): add options fp, fp.dp and nofp.
28006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28008 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28010 (armv8-a+crc): Delete.
28011 (armv8.1-a): Add options simd, crypto and nofp.
28012 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28013 (armv8.2-a+fp16): Delete.
28014 (armv8-m.main): Add option dsp.
28015 (armv8-m.main+dsp): Delete.
28016 (cortex-a8): Add fpu. Add nofp option.
28017 (cortex-a9): Add fpu. Add nofp and nosimd options.
28018 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28019 link to main cpu and architecture data structures.
28020 (gen_comm_data): Only put isa attributes from the main architecture
28022 (option): New statement for architecture and CPU entries.
28023 * arm.c (struct cpu_option): New structure.
28024 (struct processors): Add entry for options.
28025 (arm_unrecognized_feature): New function.
28026 (arm_parse_arch_cpu_name): Ignore any characters after the first
28028 (arm_parse_arch_cpu_feature): New function.
28029 (arm_configure_build_target): Separate out any CPU and architecture
28030 features and parse separately. Don't error out if -mfpu=auto is
28031 used with only an architecture string.
28032 (arm_print_asm_arch_directives): New function.
28033 (arm_file_start): Call it.
28034 * config/arm/arm-cpu-cdata.h: Regenerated.
28035 * config/arm/arm-cpu-data.h: Likewise.
28036 * config/arm/arm-tables.opt: Likewise.
28038 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28040 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28041 assembler when it is not -mfpu=auto.
28043 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28045 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28046 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28047 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28048 (ASM_CPU_SPEC): Rewrite.
28049 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28050 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28051 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28052 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28053 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28054 copied string is NUL-terminated. Also strip any characters prefixed
28056 (arm_rewrite_selected_arch): New function.
28057 (arm_rewrite_march): New function.
28059 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28061 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28062 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28063 (march, mcpu, mtune): Convert to string-based options.
28064 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28065 (arm_parse_arch_cpu_name): New function.
28066 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28067 identify selected architecture or CPU.
28068 (arm_option_save): New function.
28069 (TARGET_OPTION_SAVE): Redefine.
28070 (arm_option_restore): Restore string options.
28071 (arm_option_print): Print string options.
28073 2017-06-16 Martin Sebor <msebor@redhat.com>
28075 PR tree-optimization/80933
28076 PR tree-optimization/80934
28077 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28078 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28079 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28080 (gimple_fold_builtin): Call them.
28082 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28084 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28085 as unlikely; update profile.
28087 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28089 * predict.c (force_edge_cold): Handle declaring edges impossible
28092 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28094 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28096 (try_unroll_loop_completely): Fix reporting.
28098 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28100 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28102 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28105 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28106 if given a non-constant argument for an intrinsic which requires a
28109 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28111 * profile.c (compare_freqs): New function.
28112 (branch_prob): Sort edge list.
28113 (find_spanning_tree): Assume that the list is priority sorted.
28115 2017-06-16 Richard Biener <rguenther@suse.de>
28117 PR tree-optimization/81090
28118 * passes.def (pass_record_bounds): Remove.
28119 * tree-pass.h (make_pass_record_bounds): Likewise.
28120 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28121 make_pass_record_bounds): Likewise.
28122 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28123 not free niter estimates at the beginning but at the end.
28124 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28126 2017-06-16 Richard Biener <rguenther@suse.de>
28128 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28129 initializer to workaround ICE in host GCC 4.8.
28131 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28133 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28135 (clone_inlined_nodes): Update.
28137 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28139 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28140 prefetch settings, and enable prefetching by default at -O3.
28142 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28144 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28145 Set flag_prefetch_loop_arrays according to tuning data.
28147 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28149 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28150 New tune structure.
28151 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28152 [Unrelated to main purpose of the patch] Place the pointer field last
28153 to enable type checking errors when tune structure are wrongly merged.
28154 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28155 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28156 (thunderx2t99_prefetch_tune): New tune constants.
28157 (tune_params *_tunings): Update all tunings (no functional change).
28158 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28159 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28160 from tunings structures.
28162 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28165 * ubsan.c (instrument_null): Add T argument, use it instead
28166 of computing it based on IS_LHS.
28167 (instrument_object_size): Likewise.
28168 (pass_ubsan::execute): Adjust instrument_null and
28169 instrument_object_size callers to pass gimple_get_lhs or
28170 gimple_assign_rhs1 result to it. Use instrument_null instead of
28171 calling get_base_address and instrument_mem_ref. Handle
28172 aggregate call arguments for object-size sanitization.
28174 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28176 PR tree-optimization/81089
28177 * tree-vrp.c (is_masked_range_test): Validate operands of
28180 2017-06-15 Martin Sebor <msebor@redhat.com>
28183 * dumpfile.c (dump_register): Avoid calling memset to initialize
28184 a class with a default ctor.
28185 * gcc.c (struct compiler): Remove const qualification.
28186 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28187 * hash-table.h: Ditto.
28188 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28190 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28191 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28193 * params.h (struct param_info): Make struct members non-const.
28194 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28195 with default initialization.
28196 * vec.h (vec_copy_construct, vec_default_construct): New helper
28198 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28199 with vec_copy_construct.
28200 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28201 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28202 * doc/invoke.texi (-Wclass-memaccess): Document.
28204 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28206 * emit-rtl.h (is_leaf): Update comment about local
28207 register allocator.
28209 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28212 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28213 for a variable to have a section before checking if the section has a
28215 Set section to.persistent if persistent attribute is set.
28216 Warn if .persistent attribute is used on an automatic variable.
28218 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28220 PR rtl-optimization/80474
28221 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28223 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28225 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28228 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28230 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28231 (casesi): Emit bounds checking as RTL.
28232 (casesi_internal_mips16_<mode>): Remove bounds checking.
28234 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28236 * config/xtensa/xtensa.c (xtensa_option_override): Append
28237 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28238 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28239 xtensa_doloop_hooks): Define unconditionally.
28240 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28242 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28243 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28244 for it in xtensa_option_override.
28245 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28246 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28248 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28250 * doc/cppopts.texi: Document '-' special value to -MF.
28252 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28254 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28255 (cortex_a53_fconst): Likewise.
28256 (cortex_a53_fpmul): Likewise.
28257 (cortex_a53_f_load_64): Likewise.
28258 (cortex_a53_f_load_many): Likewise.
28259 (cortex_a53_advsimd_alu): Likewise.
28260 (cortex_a53_advsimd_alu_q): Likewise.
28261 (cortex_a53_advsimd_mul): Likewise.
28262 (cortex_a53_advsimd_mul_q): Likewise.
28263 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28264 Add missing fmul, r2f_cvt and fconst cases.
28266 2017-06-14 Richard Biener <rguenther@suse.de>
28268 PR middle-end/81088
28269 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28271 (fold_binary_loc): When associating do not treat pre-existing
28272 TREE_OVERFLOW on literal constants as a reason to allow
28273 TREE_OVERFLOW on associated literal constants.
28275 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28277 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28278 (MASK_FEATURES): New macro.
28279 * config/sparc/sparc.c (sparc_option_override): Remove the special
28280 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28282 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28284 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28285 a division of 0 if non-call exceptions are enabled.
28287 2017-06-14 Andrew Pinski <apinski@cavium.com>
28288 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28291 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28292 Improve vector initialization code gen for only variable case.
28294 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28296 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28298 2017-06-14 Richard Biener <rguenther@suse.de>
28300 PR tree-optimization/81083
28301 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28304 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28306 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28307 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28308 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28309 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28310 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28311 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28313 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28315 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28316 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28318 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28320 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28322 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28324 * config/rs6000/t-rtems: Don't handle SPE.
28326 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28328 * config/rs6000/t-linux: Don't handle SPE.
28330 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28332 * config/rs6000/eabispe.h: Delete file.
28334 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28336 * config/rs6000/t-spe: Delete file.
28338 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28340 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28341 (rs6000_legitimate_offset_address_p): Return false for anything in
28342 V2SImode or V2SFmode.
28344 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28346 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28347 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28348 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28350 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28351 (rs6000_legitimate_offset_address_p): Ditto.
28352 (rs6000_emit_move): Ditto.
28353 (rs6000_init_builtins): Remove V4HI_type_node.
28355 2017-06-13 Martin Liska <mliska@suse.cz>
28358 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28359 (gate_asan): Likewise.
28360 * asan.h (asan_no_sanitize_address_p): Remove the function.
28361 (sanitize_flags_p): New function.
28362 * builtins.def: Fix coding style.
28363 * common.opt: Use renamed enum value.
28364 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28365 * doc/extend.texi: Document no_sanitize attribute.
28366 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28367 to SANITIZE_UNDEFINED_NONDEFAULT.
28368 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28369 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28370 Use sanitize_flags_p.
28371 * gimplify.c (gimplify_function_tree): Likewise.
28372 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28373 * opts.c (parse_no_sanitize_attribute): New function.
28374 (common_handle_option): Use renamed enum value.
28375 * opts.h (parse_no_sanitize_attribute): Declare.
28376 * tree.c (sanitize_flags_p): New function.
28377 * tree.h: Declared here.
28378 * tsan.c: Use sanitize_flags_p.
28379 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28380 (instrument_mem_ref): Likewise.
28381 (instrument_bool_enum_load): Likewise.
28382 (do_ubsan_in_current_function): Remove the function.
28383 (pass_ubsan::execute): Use sanitize_flags_p.
28384 * ubsan.h: Remove do_ubsan_in_current_function
28385 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28386 (dump_function_to_file): Use it here.
28388 2017-06-13 Martin Jambor <mjambor@suse.cz>
28390 PR tree-optimization/80803
28391 PR tree-optimization/81063
28392 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28393 (propagate_subaccesses_across_link): Enqueue subtree whenever
28394 necessary instead of relying on the caller.
28396 2017-06-13 Martin Jambor <mjambor@suse.cz>
28398 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28399 that have a first_link.
28400 (sort_and_splice_var_accesses): Do not check first_link before
28402 (subtree_mark_written_and_enqueue): Likewise.
28403 (propagate_all_subaccesses): Likewise and do not stop at first
28404 parent with a first_link.
28406 2017-06-13 Martin Jambor <mjambor@suse.cz>
28408 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28411 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28413 * match.pd: New pattern.
28415 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28417 * tree-vrp.c (is_masked_range_test): New function.
28418 (register_edge_assert_for): Determine ranges for
28421 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28423 PR tree-optimization/67328
28424 * fold-const.c (maskable_range_p): New function.
28425 (build_range_check): Generate bittests if possible.
28427 2017-06-13 Martin Liska <mliska@suse.cz>
28429 * gimple-pretty-print.c (dump_probability): Add new argument.
28430 (dump_edge_probability): Dump both probability and count.
28431 (dump_gimple_label): Likewise.
28432 (dump_gimple_bb_header): Likewise.
28434 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28437 * config/avr/avr-devices.c: Fix indentation.
28438 * config/avr/gen-avr-mmcu-specs.c: Dito.
28440 2017-06-13 Richard Biener <rguenther@suse.de>
28442 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28443 instead get vector type from stmt_info.
28444 (vectorizable_reduction): Adjust. Remove dead code.
28446 2017-06-13 Richard Biener <rguenther@suse.de>
28448 PR middle-end/81065
28449 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28450 case of C * (x * C2 + C3).
28451 (fold_addr_of_array_ref_difference): Properly fold index difference.
28453 2017-06-12 David S. Miller <davem@davemloft.net>
28456 * config/sparc/sparc.md (return expander): Emit frame blockage if
28457 function uses alloca.
28459 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28461 * combine.c (make_field_assignment): Check len rather than the mode
28462 precision when calling force_to_mode.
28464 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28466 Support multilibs and devices that see flash in RAM address range.
28469 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28470 (avr_mcu_t) <flash_pm_offset>: New field.
28471 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28472 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28473 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28474 (AVR_TINY_PM_OFFSET): Remove macro.
28475 * config/avr/avr.opt (-mshort-calls): New option.
28476 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28477 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28478 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28479 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28480 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28481 instead of avr_arch->have_jmp_call.
28482 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28483 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28484 avr_arch->flash_pm_offset to define.
28485 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28486 new field flash_pm_offset. Add entry for avrxmega3.
28487 (avr_texinfo): Add entry for avrxmega3.
28488 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28489 attiny212, attiny214,
28490 attiny412, attiny414, attiny416, attiny417,
28491 attiny814, attiny816, attiny817,
28492 attiny1614, attiny1616, attiny1617,
28493 attiny3214, attiny3216, attiny3217.
28494 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28495 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28496 (avr_print_operand_address) [AVR_TINY]: Same.
28497 (avr_asm_init_sections) <readonly_data_section>: Only patch
28498 callback if avr_arch->flash_pm_offset = 0.
28499 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28500 for rodata if avr_arch->flash_pm_offset != 0.
28501 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28502 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28503 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28504 (m_options): Append opt_rcall.
28505 (m_dirnames): Append dir_rcall.
28506 * config/avr/t-multilib: Regenerate.
28508 * configure.ac [target=avr]: Check whether avrxmega3 default
28509 linker description file works as needed.
28510 * configure: Regenerate.
28511 * doc/avr-mmcu.texi: Regenerate.
28512 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28513 <__AVR_ARCH__>: Document avrxmega3 and 103.
28514 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28515 <__AVR_SHORT_CALLS__>: Document it.
28516 <__AVR_PM_BASE_ADDRESS__>: Document it.
28517 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28518 (AVR Variable Attributes) <progmem>: Document this is
28519 not needed for avrxmega3.
28520 (AVR Named Address Spaces) <__flash>: Dito.
28522 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28524 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28526 2017-06-12 Doug Rupp <rupp@adacore.com>
28528 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28529 Append vxworks-stdint.h to the tm_file list.
28530 * config/vxworks-stdint.h: New file.
28532 2017-06-12 Martin Liska <mliska@suse.cz>
28534 PR tree-optimization/81041
28535 * tree-profile.c (gimple_gen_ic_func_profiler):
28536 Create an extra BB in profile-generate
28537 (gimple_gen_time_profiler): Likewise.
28539 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28541 PR tree-optimization/81003
28542 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28543 (update_range_test): Use it instead of force_gimple_operand_gsi.
28545 2017-06-12 Richard Biener <rguenther@suse.de>
28547 PR tree-optimization/81053
28548 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28549 with backedge value not defined in loop. Simplify def stmt
28552 2017-06-11 Tom de Vries <tom@codesourcery.com>
28555 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28557 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28558 nvptx_cannot_force_const_mem.
28560 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28562 * opts.c (finish_options): Move test for flag_split_stack after
28563 it has been initialized.
28565 2017-06-11 Jason Merrill <jason@redhat.com>
28567 * tree.h (id_equal): New.
28568 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28569 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28570 instead of strcmp of IDENTIFIER_POINTER.
28572 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28574 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28575 (mark_all_inlined_calls_cdtor): Fix formating.
28576 (inline_transform): Rescale profile before inlining.
28578 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28580 * cgraph.h (cgraph_edge::clone): Update prototype.
28581 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28582 (cgraph_node::create_clone): Update.
28583 (cgraph_node::create_version_clone): Update.
28584 * tree-inline.c (copy_bb): Update.
28585 (expand_call_inline): Update.
28587 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28589 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28590 factored out from ...
28591 (rs6000_emit_prologue): ... here.
28593 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28595 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28596 factored out from ...
28597 (rs6000_emit_prologue): ... here.
28599 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28601 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28603 (handle_missing_profiles): Fix computation of tp_first_run.
28604 (counts_to_freqs): Do not touch freqs when count is 0.
28606 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28608 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28611 2017-06-10 Tom de Vries <tom@codesourcery.com>
28613 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28614 attributes): Document signal effective target.
28616 2017-06-10 Tom de Vries <tom@codesourcery.com>
28618 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28619 Document effective target stack_size.
28621 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28623 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28624 to the edit_context if they can be auto-applied.
28626 2017-06-9 Ian Lance Taylor <iant@golang.org>
28628 * opts.c (finish_options): If -fsplit-stack, disable implicit
28629 -forder-blocks-and-partition.
28630 * doc/invoke.texi (Optimize Options): Document that when using
28631 -fsplit-stack -forder-blocks-and-partition is not implicitly
28634 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28636 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28637 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28638 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28639 * builtins.def (abort, trap, unreachable): Declare cold.
28640 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28641 * tree-core.h (ECF_COLD): New.
28642 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28643 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28645 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28647 * predict.c (unlikely_executed_stmt_p): Cleanup.
28649 2017-06-09 Richard Biener <rguenther@suse.de>
28651 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28652 model if the ref is always written to.
28654 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28656 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28658 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28660 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28663 2017-06-09 Tom de Vries <tom@codesourcery.com>
28666 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28667 "target cannot support label values" when encountering LABEL_REF.
28669 2017-06-09 Martin Liska <mliska@suse.cz>
28671 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28672 (gimple_gen_ic_func_profiler): Emit direct comparison
28673 of __gcov_indirect_call_callee with NULL.
28674 (gimple_gen_time_profiler): Change probability from
28675 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28677 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28679 * profile.c (edge_gcov_counts): Turn to pointer.
28680 (compute_branch_probabilities, compute_branch_probabilities): Update.
28681 (branch_prob): Do not clear edge_gcov_count.
28682 * profile.h (edge_gcov_counts): Turn to pointer.
28683 (edge_gcov_count): Update.
28685 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28687 * gimple.h (gimple_check_failed): Mark cold.
28689 2017-06-09 Richard Biener <rguenther@suse.de>
28691 PR tree-optimization/66623
28692 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28693 refactor check_reduction into two parts, properly computing
28694 whether we have to check reduction validity for outer loop
28697 2017-06-09 Richard Biener <rguenther@suse.de>
28699 PR tree-optimization/79483
28700 * graphite-scop-detection.c (order): New global.
28701 (get_order): Compute bb to order mapping that satisfies code
28702 generation constraints.
28703 (cmp_pbbs): New helper.
28704 (build_scops): Start domwalk at entry block, sort generated
28707 2017-06-09 Richard Biener <rguenther@suse.de>
28709 PR middle-end/81007
28710 * ipa-polymorphic-call.c
28711 (ipa_polymorphic_call_context::restrict_to_inner_class):
28712 Skip FIELD_DECLs with error_mark_node type.
28713 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28716 2017-06-09 Martin Liska <mliska@suse.cz>
28718 * predict.c (struct branch_predictor): New struct.
28719 (test_prediction_value_range): New test.
28720 (predict_c_tests): New function.
28721 * selftest-run-tests.c (selftest::run_tests): Run the function.
28722 * selftest.h: Declare new tests.
28724 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28727 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28728 gen_add3_insn did not fail.
28729 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28730 r0, construct that number in a temporary reg and add that reg to r0.
28731 If asked to put the result in r0 as well, fail.
28733 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28735 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28736 for early expansion of vec_eqv.
28738 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28740 PR middle-end/81005
28741 * ubsan.c (instrument_null): Avoid pointless code temporary.
28742 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28744 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28748 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28751 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28754 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28756 * predict.c (unlikely_executed_edge_p): Move ahead.
28757 (probably_never_executed_edge_p): Use it.
28759 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28761 PR middle-end/79988
28762 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28763 gimple_call_builtin_p call.
28765 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28767 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28768 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28769 rtl_check_failed_type2, rtl_check_failed_code1,
28770 rtl_check_failed_code2, rtl_check_failed_code_mode,
28771 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28772 rtvec_check_failed_bounds, rtl_check_failed_flag,
28773 _fatal_insn_not_found, _fatal_insn): Likewise.
28774 * tree.h (tree_contains_struct_check_failed,
28775 tree_check_failed, tree_not_check_failed,
28776 tree_class_check_failed, tree_range_check_failed,
28777 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28778 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28779 tree_operand_check_failed, omp_clause_check_failed,
28780 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28783 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28785 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28786 flag_branch_probabilities.
28787 * ipa-inline.c (edge_badness): Likewise.
28788 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28789 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28790 * predict.c (maybe_hot_frequency_p): Likewise.
28791 (probably_never_executed): Likewise.
28792 * sched-ebb.c (schedule_ebbs): Likewise.
28793 * sched-rgn.c (find_single_block_region): Likewise.
28794 * tracer.c (tail_duplicate): Likewise.
28796 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28798 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28799 longer requires x_flag_profile_use.
28801 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28803 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28804 instead of flag_reorder_blocks_and_partition.
28805 * dbxout.c (dbxout_function_end): Likewise.
28806 * dwarf2out.c (gen_subprogram_die): Likewise.
28807 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28808 * hw-doloop.c (reorg_loops): Likewise.
28809 * varasm.c (assemble_start_function,
28810 assemble_end_function): Likewise.
28811 (decide_function_section): Do not check for
28812 flag_reorder_blocks_and_partition.
28814 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28816 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28818 (chkp_get_hard_register_fake_addr_expr): Ditto.
28819 (chkp_build_addr_expr): Add check for hard reg case.
28820 (chkp_parse_array_and_component_ref): Ditto.
28821 (chkp_find_bounds_1): Ditto.
28822 (chkp_process_stmt): Don't generate bounds store for
28825 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28827 * predict.c (maybe_hot_bb_p): Do not check profile status.
28828 (maybe_hot_edge_p): Likewise.
28829 (probably_never_executed): Check for zero counts even if profile
28831 (unlikely_executed_edge_p): New function.
28832 (unlikely_executed_stmt_p): New function.
28833 (unlikely_executed_bb_p): New function.
28834 (set_even_probabilities): Use unlikely predicates.
28835 (combine_predictions_for_bb): Likewise.
28836 (predict_paths_for_bb): Likewise.
28837 (predict_paths_leading_to_edge): Likewise.
28838 (determine_unlikely_bbs): New function.
28839 (estimate_bb_frequencies): Use it.
28840 (compute_function_frequency): Use zero counts even if profile is
28842 * profile-count.h: Fix typo.
28844 2017-08-08 Julia Koval <julia.koval@intel.com>
28846 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28847 _mm512_mask_cvtsepi16_storeu_epi8,
28848 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28849 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28850 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28851 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28852 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28853 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28854 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28855 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28856 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28857 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28858 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28859 __builtin_ia32_pmovuswb256mem_mask,
28860 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28861 __builtin_ia32_pmovwb512mem_mask): New builtins.
28863 2017-08-08 Julia Koval <julia.koval@intel.com>
28865 PR target/73350,80862
28866 * config/i386/subst.md (round): Fix round pattern.
28867 * config/i386/i386.c (ix86_erase_embedded_rounding):
28868 Fix erasing rounding for the fixed pattern.
28870 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28872 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28874 2017-06-08 Martin Liska <mliska@suse.cz>
28876 PR gcov-profile/80911
28877 * gcov.c (block_info::block_info): New constructor.
28879 2017-06-07 Carl Love <cel@us.ibm.com>
28881 * config/rs6000/rs6000-c: The return type of the following
28882 built-in functions was implemented as int not long long. Fix sign
28883 of return value for the unsigned version of vec_mulo and vec_mule.
28884 vector unsigned long long vec_bperm (vector unsigned long long,
28885 vector unsigned char)
28886 vector signed long long vec_mule (vector signed int,
28888 vector unsigned long long vec_mule (vector unsigned int,
28889 vector unsigned int)
28890 vector signed long long vec_mulo (vector signed int,
28892 vector unsigned long long vec_mulo (vector unsigned int,
28893 vector unsigned int)
28894 * doc/extend.texi: Fix the documentation for the built-in
28897 2017-06-07 Carl Love <cel@us.ibm.com>
28900 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28903 2017-06-07 Carl Love <cel@us.ibm.com>
28905 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28906 support, Generate doublehv for signed int/float for BE case only.
28908 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28910 * doc/invoke.texi (mcx16): Rewrite.
28912 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28914 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28915 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28916 *mov<mode>_softfloat, and an anonymous splitter): Use
28917 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28919 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28921 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28923 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28924 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28925 (rs6000_debug_reg_global): Adjust.
28926 (rs6000_init_hard_regno_mode_ok): Adjust.
28927 (rs6000_dbx_register_number): Adjust.
28928 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28929 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28930 Remove SPE_ACC and SPEFSCR.
28931 (REG_ALLOC_ORDER): Ditto.
28932 (FRAME_POINTER_REGNUM): Change to 111.
28933 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28934 (REG_CLASS_NAMES): Ditto.
28935 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28936 (REGISTER_NAMES): Ditto.
28937 (ADDITIONAL_REG_NAMES): Ditto.
28938 (rs6000_reg_names): Ditto.
28939 * config/rs6000/rs6000.md: Renumber some register number
28942 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28944 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28946 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28947 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28949 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28950 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28951 Delete the SPE high registers.
28952 (REG_ALLOC_ORDER): Ditto.
28953 (enum reg_class): Remove SPE_HIGH_REGS.
28954 (REG_CLASS_NAMES): Ditto.
28955 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28956 (REGISTER_NAMES): Ditto.
28957 (rs6000_reg_names): Ditto.
28958 * doc/tm.texi.in: Remove SPE as example.
28959 * doc/tm.texi: Regenerate.
28961 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28963 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28964 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28965 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28966 * config/rs6000/rs6000.md (type): Remove "brinc".
28968 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28970 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28971 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28972 * config/rs6000/linuxspe.h: Delete file.
28973 * config/rs6000/rs6000.md: Don't include spe.md.
28974 * config/rs6000/spe.h: Delete file.
28975 * config/rs6000/spe.md: Delete file.
28976 * config/rs6000/t-rs6000: Remove spe.md.
28978 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28980 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28981 (reg_or_none500mem_operand): Delete.
28982 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28983 instead of reg_or_none500mem_operand.
28985 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28987 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28988 handling of SPE flags.
28989 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28991 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28993 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28995 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28996 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28997 paired_divv2sf3): Similar.
28998 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28999 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
29000 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
29002 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
29003 Rename the paired_* instruction patterns.
29004 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
29006 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29007 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29008 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29009 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29010 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29011 PAIRED_VECTOR_MODE.
29012 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29013 (spe_func_has_64bit_regs_p): Delete.
29014 (spe_expand_predicate_builtin): Delete.
29015 (spe_expand_evsel_builtin): Delete.
29016 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29017 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29018 (invalid_e500_subreg): Delete.
29019 (rs6000_legitimize_address): Always force_reg op2 as well, for
29020 paired single memory accesses.
29021 (rs6000_member_type_forces_blk): Delete.
29022 (rs6000_spe_function_arg): Delete.
29023 (rs6000_expand_unop_builtin): Delete SPE handling.
29024 (rs6000_expand_binop_builtin): Ditto.
29025 (spe_expand_stv_builtin): Delete.
29026 (bdesc_2arg_spe): Delete.
29027 (spe_expand_builtin): Delete.
29028 (spe_expand_predicate_builtin): Delete.
29029 (spe_expand_evsel_builtin): Delete.
29030 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29031 (spe_init_builtins): Delete.
29032 (spe_func_has_64bit_regs_p): Delete.
29033 (savres_routine_name): Delete "info" parameter. Adjust callers.
29034 (rs6000_emit_stack_reset): Ditto.
29035 (rs6000_dwarf_register_span): Delete.
29036 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29037 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29038 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29040 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29042 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29043 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29044 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29045 mulv2sf3, divv2sf3): Delete expanders.
29047 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29049 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29051 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29053 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29054 * config/rs6000/rs6000.c: Ditto.
29056 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29058 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29059 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29060 comparison_operator.
29062 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29064 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29065 * config/rs6000/rs6000.opt: Ditto.
29066 * config/rs6000/t-rtems: Ditto.
29068 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29070 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29071 TARGET_E500_SINGLE by 0, simplify.
29072 * config/rs6000/rs6000.c: Ditto.
29073 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29074 (spe_build_register_parallel): Delete.
29075 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29076 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29077 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29078 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29079 (E500_CONVERT): Delete.
29080 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29082 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29084 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29085 * config/rs6000/dfp.md: Ditto.
29086 (negdd2, *negdd2_fpr): Merge.
29087 (absdd2, *absdd2_fpr): Merge.
29088 (negtd2, *negtd2_fpr): Merge.
29089 (abstd2, *abstd2_fpr): Merge.
29090 * config/rs6000/e500.h: Delete file.
29091 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29092 TARGET_FPRS by 1 and simplify.
29093 * config/rs6000/rs6000-c.c: Ditto.
29094 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29095 TARGET_DF_SPE by 0.
29096 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29098 * config/rs6000/rs6000.md: Ditto.
29099 (floatdidf2, *floatdidf2_fpr): Merge.
29100 (move_from_CR_gt_bit): Delete.
29101 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29102 (E500_CR_IOR_COMPARE): Delete.
29103 (All patterns that require !TARGET_FPRS): Delete.
29104 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29106 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29108 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29110 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29112 * graphds.c (add_edge): Intitialize edge's attached data.
29113 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29114 pointer parameter. Call pointed function on each edge during
29115 graph traversing. Skip traversing the edge when the function
29117 (graphds_dfs, graphds_scc): Ditto.
29118 (for_each_edge): New parameter. Pass the new parameter to callback
29120 * graphds.h (skip_edge_callback): New function pointer type.
29121 (graphds_dfs, graphds_scc): New function pointer parameter.
29122 (graphds_edge_callback, for_each_edge): New parameter.
29124 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29126 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29127 out code checking if runtime alias check is possible to below ...
29128 Call the new function.
29129 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29130 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29132 2017-06-07 Marek Polacek <polacek@redhat.com>
29135 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29136 TYPE_OVERFLOW_WRAPS check.
29138 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29140 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29141 if versioning is required.
29142 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29143 peeling with the check for versioning.
29145 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29147 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29148 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29149 Set true to new parameter if new ssa variable is defined.
29150 (vect_gen_vector_loop_niters): Refactor. Set range information
29151 for the new vector loop bound variable.
29152 (vect_do_peeling): Ditto.
29154 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29156 * tree-affine.c (ssa.h): Include header file.
29157 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29158 has wrapping overflow behavior.
29160 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29162 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29164 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29166 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29167 (tree_to_aff_combination): ... here.
29169 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29171 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29172 reg_pressure model function.
29173 (ivopts_global_cost_for_size): Delete.
29174 (determine_set_costs, iv_ca_recount_cost): Call new model function
29175 ivopts_estimate_reg_pressure.
29177 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29179 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29180 expensive than udiv. Remove floating point cases from mod.
29182 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29184 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29185 Increase idiv cost.
29187 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29189 * config/aarch64/aarch64.md
29190 (copysignsf3): Fix mask generation.
29192 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29194 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29196 (class dump_manager): Add register_dumps method.
29197 * dumpfile.c: Include langhooks.h.
29198 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29199 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29200 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29201 (dump_manager::dump_register): Start with 512 entries instead of 32.
29202 (dump_manager::register_dumps): New method.
29203 * toplev.c (general_init): Instead of invoking register_dumps
29204 langhook, invoke register_dumps method on the dump manager.
29205 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29208 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29210 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29211 Say that # requires an associated define_split to exist, and that
29212 the define_split must be suitable for use after register allocation.
29214 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29216 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29217 (compute_outgoing_frequencies): Also initialize zero counts.
29218 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29219 around loops; preserve more of profile when nothing changes.
29221 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29223 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29225 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29226 * config/arm/arm-cpu-cdata.h: Regenerate.
29227 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29228 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29229 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29230 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29231 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29233 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29235 * config/arm/t-rmprofile: Likewise.
29236 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29238 2017-06-06 David S. Miller <davem@davemloft.net>
29241 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29242 blockage if function uses alloca.
29244 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29246 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29247 New "uid" fields to hold pretty-print IDs of group and ref.
29248 Memory references are now identified as <group_id>:<ref_id>
29249 instead of using [random] addresses.
29250 (dump_mem_details): Simplify, no functional change.
29251 (dump_mem_ref): Simplify and make output more concise.
29252 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29253 (find_or_create_group): Initialize group uid.
29254 (record_ref): Initialize ref uid. Improve debug output.
29255 (prune_group_by_reuse, should_issue_prefetch_p,)
29256 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29257 (mark_nontemporal_store, determine_loop_nest_reuse):
29258 Improve debug output.
29260 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29262 * dbgcnt.def (prefetch): New debug counter.
29263 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29264 (schedule_prefetches): Stop issueing prefetches if debug counter
29267 2017-06-06 Tom de Vries <tom@codesourcery.com>
29269 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29270 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29272 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29274 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29275 Use aarch64_reg_or_zero predicate for operand 4.
29276 (aarch64_compare_and_swap<mode> define_insn_and_split):
29277 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29278 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29280 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29282 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29283 (arm_compute_save_core_reg_mask): This.
29284 (thumb1_compute_save_reg_mask): Rename into ...
29285 (thumb1_compute_save_core_reg_mask): This.
29286 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29287 (arm_compute_frame_layout): Likewise.
29289 2017-06-06 Richard Biener <rguenther@suse.de>
29291 PR tree-optimization/80974
29292 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29293 keep or clear leaders SSA info.
29295 2017-06-06 Tom de Vries <tom@codesourcery.com>
29297 * config/nvptx/nvptx.c (split_mode_p): New function.
29298 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29300 2017-06-06 Tom de Vries <tom@codesourcery.com>
29302 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29304 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29307 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29310 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29312 * shrink-wrap.c (handle_simple_exit): Update profile.
29313 (try_shrink_wrapping): Upate profile.
29315 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29317 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29318 (tree_guess_outgoing_edge_probabilities): New.
29319 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29320 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29322 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29324 * ipa-split.c (split_function): Initialize return bb profile.
29326 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29328 * profile.c (compute_branch_probabilities): Also initialize
29329 EXIT_BLOCK profile.
29331 2017-06-06 Richard Biener <rguenther@suse.de>
29333 PR tree-optimization/80928
29334 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29335 (vect_analyze_loop_operations): Properly guard analysis for
29337 (vect_transform_loop): Likewise.
29338 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29339 (vect_model_induction_cost): Do not cost for pure SLP.
29340 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29341 of induction in inner loop vectorization.
29342 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29343 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29344 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29346 (vect_analyze_slp_cost_1): Cost induction.
29347 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29348 (vect_get_slp_vect_defs): Likewise.
29349 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29350 (vect_transform_stmt): Handle SLP reductions.
29351 * tree-vectorizer.h (vectorizable_induction): Adjust.
29353 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29355 * config/rs6000/rs6000.c (make_resolver_func): Update
29356 init_lowered_empty_function call.
29358 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29360 * doc/invoke.texi: Document the -fprofile-abs-path option.
29361 * common.opt (fprofile-abs-path): New option.
29362 * gcov-io.h (gcov_write_filename): Declare.
29363 * gcov-io.c (gcov_write_filename): New function.
29364 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29365 * profile.c (output_location): Likewise.
29367 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29369 * shring-wrap.c: Revert accidental commit.
29371 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29373 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29375 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29377 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29379 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29380 profile in callgraph edge.
29381 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29382 (apply_scale): Likewise.
29383 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29384 Also scale profile when inlining function with zero profile.
29385 (initialize_cfun): Update exit block profile even when it is zero.
29386 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29387 when profile is read.
29389 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29391 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29392 (CLONE_*): New constants to define the processors we can generate
29393 code for with the target_clone attribute.
29394 (rs6000_clone_map): New array to identify which clone processors
29395 the current program is running on.
29396 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29397 target_clone attribute.
29398 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29399 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29400 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29401 (cpu_expand_builtin): Add support for target_clone attribute.
29402 (rs6000_valid_attribute_p): Allow "default" attribute.
29403 (get_decl_name): New debug function to simplify printing the
29404 current function name in debugging statements.
29405 (rs6000_clone_priority): New functions to support the target_clone
29406 attribute, and be able to generate code to switch between ISA 2.05
29407 through ISA 3.0 (power6 through power9).
29408 (rs6000_compare_version_priority): Likewise.
29409 (rs6000_get_function_versions_dispatcher): Likewise.
29410 (make_resolver_func): Likewise.
29411 (add_condition_to_bb): Likewise.
29412 (dispatch_function_versions): Likewise.
29413 (rs6000_generate_version_dispatcher_body): Likewise.
29414 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29415 (fusion_gpr_load_p): Fix a spacing issue.
29416 * doc/extend.texi (Common Function Attributes): Document that the
29417 PowerPC supports the target_clone attribute.
29419 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29421 * config/arm/arm.h: explain F symbol found in description of ARM
29422 register allocation in its legend.
29424 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29426 * config/mips/frame-header-opt.c: Include profile-count.h.
29427 * config/riscv/riscv.c: Include profile-count.h
29429 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29431 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29433 (sm_set_flag_if_changed): Add bbs field.
29434 (execute_sm_if_changed_flag_set): Pass BBS.
29435 (execute_sm): Update.
29437 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29439 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29442 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29444 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29445 (peephole2): New peephole2 to emit the above.
29446 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29448 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29450 * config/aarch64/aarch64.c (define_peephole2 above
29451 *sub_<shift>_<mode>): New peephole.
29453 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29455 * config/i386/i386.c (make_resolver_func): Update.
29456 * Makefile.in: Add profile-count.h and profile-count.o
29457 * auto-profile.c (afdo_indirect_call): Update to new API.
29458 (afdo_set_bb_count): Update.
29459 (afdo_propagate_edge): Update.
29460 (afdo_propagate_circuit): Update.
29461 (afdo_calculate_branch_prob): Update.
29462 (afdo_annotate_cfg): Update.
29463 * basic-block.h: Include profile-count.h
29464 (struct edge_def): Turn count to profile_count.
29465 (struct basic_block_def): Likewie.
29466 (REG_BR_PROB_BASE): Move to profile-count.h
29467 (RDIV): Move to profile-count.h
29468 * bb-reorder.c (max_entry_count): Turn to profile_count.
29469 (find_traces): Update.
29470 (rotate_loop):Update.
29471 (connect_traces):Update.
29472 (sanitize_hot_paths):Update.
29473 * bt-load.c (migrate_btr_defs): Update.
29474 * cfg.c (RDIV): Remove.
29475 (init_flow): Use alloc_block.
29476 (alloc_block): Uninitialize count.
29477 (unchecked_make_edge): Uninitialize count.
29478 (check_bb_profile): Update.
29479 (dump_edge_info): Update.
29480 (dump_bb_info): Update.
29481 (update_bb_profile_for_threading): Update.
29482 (scale_bbs_frequencies_int): Update.
29483 (scale_bbs_frequencies_gcov_type): Update.
29484 (scale_bbs_frequencies_profile_count): New.
29485 * cfg.h (update_bb_profile_for_threading): Update.
29486 (scale_bbs_frequencies_profile_count): Declare.
29487 * cfgbuild.c (compute_outgoing_frequencies): Update.
29488 (find_many_sub_basic_blocks): Update.
29489 * cfgcleanup.c (try_forward_edges): Update.
29490 (try_crossjump_to_edge): Update.
29491 * cfgexpand.c (expand_gimple_tailcall): Update.
29492 (construct_exit_block): Update.
29493 * cfghooks.c (verify_flow_info): Update.
29494 (dump_bb_for_graph): Update.
29495 (split_edge): Update.
29496 (make_forwarder_block): Update.
29497 (duplicate_block): Update.
29498 (account_profile_record): Update.
29499 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29500 (get_estimated_loop_iterations): Update.
29501 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29502 (single_likely_exit): Update.
29503 * cfgloopmanip.c (scale_loop_profile): Update.
29505 (set_zero_probability): Update.
29506 (lv_adjust_loop_entry_edge): Update.
29507 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29508 (purge_dead_edges): Update.
29509 (rtl_account_profile_record): Update.
29510 * cgraph.c (cgraph_node::create): Uninitialize count.
29511 (symbol_table::create_edge): Uninitialize count.
29512 (cgraph_update_edges_for_call_stmt_node): Update.
29513 (cgraph_edge::dump_edge_flags): Update.
29514 (cgraph_node::dump): Update.
29515 (cgraph_edge::maybe_hot_p): Update.
29516 * cgraph.h: Include profile-count.h
29517 (create_clone), create_edge, create_indirect_edge): Update.
29518 (cgraph_node): Turn count to profile_count.
29519 (cgraph_edge0: Likewise.
29520 (make_speculative, clone): Update.
29521 (create_edge): Update.
29522 (init_lowered_empty_function): Update.
29523 * cgraphclones.c (cgraph_edge::clone): Update.
29524 (duplicate_thunk_for_node): Update.
29525 (cgraph_node::create_clone): Update.
29526 * cgraphunit.c (cgraph_node::analyze): Update.
29527 (cgraph_node::expand_thunk): Update.
29528 * final.c (dump_basic_block_info): Update.
29529 * gimple-streamer-in.c (input_bb): Update.
29530 * gimple-streamer-out.c (output_bb): Update.
29531 * graphite.c (print_global_statistics): Update.
29532 (print_graphite_scop_statistics): Update.
29533 * hsa-brig.c: Include basic-block.h.
29534 * hsa-dump.c: Include basic-block.h.
29535 * hsa-gen.c (T sum_slice): Update.
29536 (convert_switch_statements):Update.
29537 * hsa-regalloc.c: Include basic-block.h.
29538 * ipa-chkp.c (chkp_produce_thunks): Update.
29539 * ipa-cp.c (struct caller_statistics): Update.
29540 (init_caller_stats): Update.
29541 (gather_caller_stats): Update.
29542 (ipcp_cloning_candidate_p): Update.
29543 (good_cloning_opportunity_p): Update.
29544 (get_info_about_necessary_edges): Update.
29545 (dump_profile_updates): Update.
29546 (update_profiling_info): Update.
29547 (update_specialized_profile): Update.
29548 (perhaps_add_new_callers): Update.
29549 (decide_about_value): Update.
29550 (ipa_cp_c_finalize): Update.
29551 * ipa-devirt.c (struct odr_type_warn_count): Update.
29552 (struct decl_warn_count): Update.
29553 (struct final_warning_record): Update.
29554 (possible_polymorphic_call_targets): Update.
29555 (ipa_devirt): Update.
29556 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29557 * ipa-icf.c (sem_function::merge): Update.
29558 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29559 * ipa-inline.c (compute_uninlined_call_time): Update.
29560 (compute_inlined_call_time): Update.
29561 (want_inline_small_function_p): Update.
29562 (want_inline_self_recursive_call_p): Update.
29563 (edge_badness): Update.
29564 (lookup_recursive_calls): Update.
29565 (recursive_inlining): Update.
29566 (inline_small_functions): Update.
29567 (dump_overall_stats): Update.
29568 (dump_inline_stats): Update.
29569 * ipa-profile.c (ipa_profile_generate_summary): Update.
29570 (ipa_propagate_frequency): Update.
29571 (ipa_profile): Update.
29572 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29573 * ipa-utils.c (ipa_merge_profiles): Update.
29574 * loop-doloop.c (doloop_modify): Update.
29575 * loop-unroll.c (report_unroll): Update.
29576 (unroll_loop_runtime_iterations): Update.
29577 * lto-cgraph.c (lto_output_edge): Update.
29578 (lto_output_node): Update.
29579 (input_node): Update.
29580 (input_edge): Update.
29581 (merge_profile_summaries): Update.
29582 * lto-streamer-in.c (input_cfg): Update.
29583 * lto-streamer-out.c (output_cfg): Update.
29584 * mcf.c (create_fixup_graph): Update.
29585 (adjust_cfg_counts): Update.
29586 (sum_edge_counts): Update.
29587 * modulo-sched.c (sms_schedule): Update.
29588 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29589 * predict.c (maybe_hot_count_p): Update.
29590 (probably_never_executed): Update.
29591 (dump_prediction): Update.
29592 (combine_predictions_for_bb): Update.
29593 (propagate_freq): Update.
29594 (handle_missing_profiles): Update.
29595 (counts_to_freqs): Update.
29596 (rebuild_frequencies): Update.
29597 (force_edge_cold): Update.
29598 * predict.h: Include profile-count.h
29599 (maybe_hot_count_p, counts_to_freqs): UPdate.
29600 * print-rtl-function.c: Do not include cfg.h
29601 * print-rtl.c: Include basic-block.h
29602 * profile-count.c: New file.
29603 * profile-count.h: New file.
29604 * profile.c (is_edge_inconsistent): Update.
29605 (correct_negative_edge_counts): Update.
29606 (is_inconsistent): Update.
29607 (set_bb_counts): Update.
29608 (read_profile_edge_counts): Update.
29609 (compute_frequency_overlap): Update.
29610 (compute_branch_probabilities): Update; Initialize and deinitialize
29612 (branch_prob): Update.
29613 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29614 (edge_gcov_count): New.
29615 (bb_gcov_count): New.
29616 * shrink-wrap.c (try_shrink_wrapping): Update.
29617 * tracer.c (better_p): Update.
29618 * trans-mem.c (expand_transaction): Update.
29619 (ipa_tm_insert_irr_call): Update.
29620 (ipa_tm_insert_gettmclone_call): Update.
29621 * tree-call-cdce.c: Update.
29622 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29623 (gimple_duplicate_sese_tail): Update.
29624 (gimple_account_profile_record): Update.
29625 (execute_fixup_cfg): Update.
29626 * tree-inline.c (copy_bb): Update.
29627 (copy_edges_for_bb): Update.
29628 (initialize_cfun): Update.
29629 (freqs_to_counts): Update.
29630 (copy_cfg_body): Update.
29631 (expand_call_inline): Update.
29632 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29633 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29634 (try_unroll_loop_completely): Update.
29635 (try_peel_loop): Update.
29636 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29637 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29638 * tree-ssa-loop-split.c (connect_loops): Update.
29639 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29640 * tree-ssa-reassoc.c (branch_fixup): Update.
29641 * tree-ssa-tail-merge.c (replace_block_by): Update.
29642 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29643 (compute_path_counts): Update.
29644 (update_profile): Update.
29645 (recompute_probabilities): Update.
29646 (update_joiner_offpath_counts): Update.
29647 (estimated_freqs_path): Update.
29648 (freqs_to_counts_path): Update.
29649 (clear_counts_path): Update.
29650 (ssa_fix_duplicate_block_edges): Update.
29651 (duplicate_thread_path): Update.
29652 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29653 (struct switch_conv_info): Update.
29654 * tree-tailcall.c (decrease_profile): Update.
29655 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29656 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29657 * value-prof.c (check_counter): Update.
29658 (gimple_divmod_fixed_value): Update.
29659 (gimple_mod_pow2): Update.
29660 (gimple_mod_subtract): Update.
29661 (gimple_ic_transform): Update.
29662 (gimple_stringop_fixed_value): Update.
29663 * value-prof.h (gimple_ic): Update.
29665 2017-06-02 Carl Love <cel@us.ibm.com>
29667 * config/rs6000/rs6000-c: Add support for built-in functions
29668 vector double vec_doublee (vector signed int);
29669 vector double vec_doublee (vector unsigned int);
29670 vector double vec_doublee (vector float);
29671 vector double vec_doubleh (vector signed int);
29672 vector double vec_doubleh (vector unsigned int);
29673 vector double vec_doubleh (vector float);
29674 vector double vec_doublel (vector signed int);
29675 vector double vec_doublel (vector unsigned int);
29676 vector double vec_doublel (vector float);
29677 vector double vec_doubleo (vector signed int);
29678 vector double vec_doubleo (vector unsigned int);
29679 vector double vec_doubleo (vector float);.
29680 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29681 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29683 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29684 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29685 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29687 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29688 vec_doublel, vec_doubleh.
29689 * doc/extend.texi: Update the built-in documentation file for the
29690 new built-in functions.
29692 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29695 * ipa-inline-analysis.c (free_growth_caches): Set
29696 edge_removal_hook_holder to NULL after removing it.
29698 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29700 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29701 comparision with zero.
29703 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29704 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29705 for early expansion of vec_min and vec_max builtins.
29706 (builtin_function_type): Add min/max unsigned variants to those
29707 identified as having unsigned arguments.
29709 2017-06-02 Olivier Hainque <hainque@adacore.com>
29711 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29713 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29715 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29716 Use VALL_F16 iterator rather than VALL.
29718 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29720 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29721 Emit CBNZ inside loop when doing a strong exchange and comparing
29722 against zero. Generate the CC flags after the loop.
29724 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29726 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29727 (dl_section_ref): New.
29728 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29729 On AIX, append an expression to subtract the size of the
29730 section length to dl_section_ref.
29732 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29734 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29735 for early expansion of vector absolute builtins.
29737 2017-06-02 Richard Biener <rguenther@suse.de>
29739 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29740 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29742 2017-06-02 Richard Biener <rguenther@suse.de>
29744 PR tree-optimization/80948
29745 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29746 stmt order as well.
29748 2017-06-02 Richard Biener <rguenther@suse.de>
29750 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29752 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29753 for inductions as relevant.
29755 2017-06-02 Richard Biener <rguenther@suse.de>
29757 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29758 (vectorizable_induction): ... this. Remove dead code.
29760 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29762 * builtins. (expand_builtin_alloca): Remove second parameter and
29763 infer its value from the first parameter instead.
29764 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29766 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29768 PR rtl-optimization/80903
29769 * loop-doloop.c (add_test): Unshare sequence.
29771 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29773 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29775 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29777 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29779 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29780 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29781 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29782 (xlogue_layout::get_stub_rtx): Make static.
29783 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29784 (xlogue_layout::compute_stub_managed_regs): Rename to...
29785 (xlogue_layout::count_stub_managed_regs): ...this.
29786 (xlogue_layout::is_stub_managed_reg): New function.
29787 (xlogue_layout::m_stub_names): Rename to...
29788 (xlogue_layout::s_stub_names): ...this, make static.
29789 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29790 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29791 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29792 xlogue_layout::s_stub_names): Instantiate statics.
29793 (stub_managed_regs): Remove.
29794 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29795 (disable_call_ms2sysv_xlogues): Rename to...
29796 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29797 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29799 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29800 change after reload_completed.
29801 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29803 (ix86_expand_prologue): Likewise.
29804 (ix86_expand_epilogue): Likewise.
29805 (ix86_expand_split_stack_prologue): Likewise.
29806 (ix86_compute_frame_layout): Remove frame parameter ...
29807 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29808 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29810 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29811 (ix86_frame): Move from here ...
29812 * config/i386/i386.h (ix86_frame): ... to here.
29813 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29814 complete ix86_frame data structure instead. Remove some_ld_name.
29816 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29818 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29819 symbols that hold a DECL_VALUE_EXPR.
29821 2017-06-01 Martin Jambor <mjambor@suse.cz>
29823 PR tree-optimization/80898
29824 * tree-sra.c (process_subtree_disqualification): Removed.
29825 (disqualify_candidate): Do not acll
29826 process_subtree_disqualification.
29827 (subtree_mark_written_and_enqueue): New function.
29828 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29829 RHS has been disqualified and re-queue LHS if necessary. Apart
29830 from that, ignore disqualified RHS.
29832 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29834 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29835 address fetch for z10 or later.
29837 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29839 * config/arc/arc.md (tst_movb): Add guard when splitting.
29841 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29843 * config/arc/arc.c (arc_can_eliminate): Test against
29844 arc_frame_pointer_needed.
29846 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29848 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29849 to prevent store reordering.
29850 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29851 (type): Add block type.
29852 (stack_tie): Define special instruction to be used in
29855 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29857 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29858 constraint. It is not valid for the pattern.
29859 (noncommutative_binary_comparison): Likewise.
29861 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29863 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29866 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29868 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29869 be used by the reg-alloc.
29871 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29873 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29874 reg-alloc when having mul64 or mul32x16 instructions.
29875 (mulsidi3): Likewise.
29876 (umulsidi3): Likewise.
29877 (mulsi32x16): New pattern.
29878 (mulsi64): Likewise.
29879 (mulsidi64): Likewise.
29880 (umulsidi64): Likewise.
29881 (MUL32x16_REG): Define.
29882 (mul64_600): Use MUL32x16_REG.
29883 (mac64_600): Likewise.
29884 (umul64_600): Likewise.
29885 (umac64_600): Likewise.
29887 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29889 * config/arc/arc.md (mulsi3_700): Make it commutative.
29891 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29893 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29895 (*sign_extendsidi2_insn): Likewise for movstosw.
29897 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29899 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29900 the type of the input discriminant value. Convert the
29901 discriminant value of signedness vary.
29903 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29905 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29906 Add to -Wall section.
29908 2017-06-01 Richard Biener <rguenther@suse.de>
29910 PR middle-end/66313
29911 * fold-const.c (fold_plusminus_mult_expr): If the factored
29912 factor may be zero use a wrapping type for the inner operation.
29913 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29914 and handle moved defs.
29915 (process_assignment): Properly guard the unary op case. Return a
29916 tri-state indicating that moving the stmt before the call may allow
29917 to continue. Pass through to_move.
29918 (find_tail_calls): Handle moving unrelated defs before
29921 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29924 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29925 splitter result in the canonical way.
29927 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29929 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29930 also for 32bit target. Update insn attributes.
29931 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29933 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29935 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29936 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29937 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29938 (_mm_maskz_min_ss): New intrinsics.
29940 2017-05-31 Martin Liska <mliska@suse.cz>
29942 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29943 Change comment style to one we normally use.
29944 (vectorizable_reduction): Likewise.
29945 (vectorizable_induction): Likewise.
29946 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29947 (vectorizable_call): Likewise.
29948 (vectorizable_simd_clone_call): Likewise.
29949 (vectorizable_conversion): Likewise.
29950 (vectorizable_assignment): Likewise.
29951 (vectorizable_shift): Likewise.
29952 (vectorizable_operation): Likewise.
29953 (vectorizable_store): Likewise.
29954 (vectorizable_load): Likewise.
29955 * tree-vectorizer.h: Likewise.
29957 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29959 * passes.c (emergency_dump_function): New.
29960 * tree-pass.h (emergency_dump_function): Declare.
29961 * plugin.c (plugins_internal_error_function): Remove.
29962 * plugin.h (plugins_internal_error_function): Remove declaration.
29963 * toplev.c (internal_error_function): New static function. Use it...
29964 (general_init): ...here.
29966 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29968 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29969 (arc_rtx_costs): Add costs for new patterns.
29970 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29971 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29973 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29975 * tree-ssa-strlen.c (get_next_strinfo): New function.
29976 (get_stridx_plus_constant): Use it.
29977 (zero_length_string): Likewise.
29978 (adjust_related_strinfos): Likewise.
29979 (adjust_last_stmt): Likewise.
29981 2017-05-31 Richard Biener <rguenther@suse.de>
29984 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29985 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29987 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29989 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29990 loop_vinfo argument and use of dependence distance vectors.
29991 Check instead whether the two references differ only in their
29992 initial value and assume that they have the same alignment if the
29993 difference is a multiple of the vector alignment.
29994 (vect_analyze_data_refs_alignment): Update call accordingly.
29996 2017-05-31 Martin Liska <mliska@suse.cz>
29999 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
30001 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30003 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
30004 (create_intersect_range_checks): Move from ...
30005 * tree-data-ref.c (create_intersect_range_checks_index)
30006 (create_intersect_range_checks): ... to here.
30007 (create_runtime_alias_checks): New function factored from ...
30008 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30009 here. Call above function.
30010 * tree-data-ref.h (create_runtime_alias_checks): New function.
30012 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30014 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30015 segment length for dr_b and compute it in wide_int.
30017 2017-05-31 Richard Biener <rguenther@suse.de>
30019 PR tree-optimization/80906
30020 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30021 and pass through iv_map.
30022 (copy_bb_and_scalar_dependences): Adjust.
30023 (translate_pending_phi_nodes): Likewise.
30024 (copy_loop_close_phi_args): Handle code-generating IVs instead
30027 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30029 * diagnostic-color.c (color_dict): Add "type-diff".
30030 (parse_gcc_colors): Update comment.
30031 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30032 -fdiagnostics-show-template-tree and -fno-elide-type.
30033 (GCC_COLORS): Add type-diff to example.
30035 (-fdiagnostics-show-template-tree): New.
30036 (-fno-elide-type): New.
30037 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30038 the pp_format_decoder callback. Call any m_format_postprocessor's
30040 (pretty_printer::pretty_printer): Initialize
30041 m_format_postprocessor.
30042 (pretty_printer::~pretty_printer): Delete any
30043 m_format_postprocessor.
30044 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30045 (class format_postprocessor): New class.
30046 (struct pretty_printer::format_decoder): Document the new parameters.
30047 (struct pretty_printer::m_format_postprocessor): New field.
30048 * tree-diagnostic.c (default_tree_printer): Update for new
30049 bool and const char ** params.
30050 * tree-diagnostic.h (default_tree_printer): Likewise.
30052 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30054 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30055 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30056 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30057 rs6000_gen_cell_microcode code.
30058 (rs6000_final_prescan_insn): Delete.
30059 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30060 "warn-cell-microcode" entries.
30061 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30062 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30063 throughout. Change cc_reg_not_micro_cr0_operand to
30064 cc_reg_not_cr0_operand throughout.
30065 (*extendhi<mode>2_noload): Delete.
30066 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30067 (mwarn-cell-microcode): Delete.
30068 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30069 -mgen-cell-microcode and -mwarn-cell-microcode.
30071 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30074 * config/i386/constraints.md (Yd): New constraint.
30076 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30077 and (?Yd, r) alternatives. Update insn attributes.
30078 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30079 and (?*Yd, r) alternatives. Update insn attributes.
30080 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30082 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30084 * gimplify.c (gimplify_modify_expr): Don't create a
30085 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30088 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30090 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30092 2017-05-30 Richard Biener <rguenther@suse.de>
30094 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30095 and reduc_def fields.
30096 (STMT_VINFO_REDUC_TYPE): New define.
30097 (STMT_VINFO_REDUC_DEF): Likewise.
30098 (vect_force_simple_reduction): Adjust prototype.
30099 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30100 (vect_is_simple_reduction): Remove check_reduction argument.
30101 (vect_force_simple_reduction): Adjust and set
30102 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30103 (vectorizable_reduction): Do not re-do reduction analysis
30104 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30105 * tree-parloops.c (gather_scalar_reductions): Adjust.
30107 2017-05-30 Richard Biener <rguenther@suse.de>
30109 PR middle-end/80901
30110 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30113 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30115 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30116 Introduce unknown_misalignment parameter and remove vf.
30117 (vect_peeling_hash_get_lowest_cost):
30118 Pass unknown_misalignment parameter.
30119 (vect_enhance_data_refs_alignment):
30120 Fix unsupportable data ref treatment.
30122 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30124 * tree-vect-data-refs.c (vect_get_data_access_cost):
30125 Workaround for SLP handling.
30126 (vect_enhance_data_refs_alignment):
30127 Compute costs for doing no peeling at all, compare to the best
30128 peeling costs so far and avoid peeling if cheaper.
30130 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30132 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30133 Return peeling info and set costs to zero for unlimited cost
30135 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30136 with unknown misalignment. Compute and costs for unknown
30137 misalignment, compare them to the costs for known misalignment
30138 and choose the cheapest for peeling.
30140 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30142 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30143 (vect_get_peeling_costs_all_drs): Create function.
30144 (vect_peeling_hash_get_lowest_cost):
30145 Use vect_get_peeling_costs_all_drs.
30146 (vect_peeling_supportable): Create function.
30147 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30149 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30151 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30152 DR_HAS_NEGATIVE_STEP.
30153 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30154 (vect_enhance_data_refs_alignment): Use.
30155 (vect_duplicate_ssa_name_ptr_info): Use.
30156 * tree-vectorizer.h (dr_misalignment): Use.
30157 (known_alignment_for_access_p): Use.
30159 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30162 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30164 (has_section_name): New function.
30166 2017-05-30 Martin Liska <mliska@suse.cz>
30169 * auto-profile.c (get_function_decl_from_block): Fix
30172 2017-05-30 Richard Biener <rguenther@suse.de>
30174 PR middle-end/80876
30175 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30177 2017-05-30 Martin Liska <mliska@suse.cz>
30179 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30180 * dumpfile.h (struct dump_file_info): Remove ctors.
30182 2017-05-30 Martin Liska <mliska@suse.cz>
30184 * predict.def: Fix GNU coding style.
30186 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30188 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30189 Mark 'to' argument with ATTRIBUTE_UNUSED.
30191 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30193 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30194 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30195 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30198 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30200 * doc/install.texi (Options specification): Restore entry of
30201 --enable-sjlj-exceptions.
30203 2017-05-27 Michael Eager <eager@eagercon.com>
30206 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30208 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30210 * config/microblaze/microblaze.h
30211 (FIXED_REGISTERS): Update in macro.
30212 (CALL_USED_REGISTERS): Update in macro.
30214 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30216 * doc/install.texi: Add links to macOS binary distributions.
30218 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30222 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30224 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30226 2017-05-26 Martin Liska <mliska@suse.cz>
30228 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30230 2017-05-26 Martin Liska <mliska@suse.cz>
30232 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30233 always leading ';; '.
30234 (dump_bb_info): Likewise.
30235 (brief_dump_cfg): Likewise.
30236 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30237 * dumpfile.c: Remove usage of TDF_VERBOSE.
30238 * dumpfile.h (enum dump_kind): Likewise.
30239 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30240 * print-tree.c (debug_verbose): Remove.
30241 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30242 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30243 * tree-diagnostic.c (default_tree_printer): Replace
30244 TDF_DIAGNOSTIC with TDF_SLIM.
30246 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30248 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30249 in parameter loop, rather than loop_vinfo.
30250 (create_intersect_range_checks): Ditto.
30251 (vect_create_cond_for_alias_checks): Update call to above functions.
30253 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30255 PR tree-optimization/80815
30256 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30257 for merging runtime alias checks. Handle negative DR_STEPs.
30259 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30261 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30263 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30264 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30265 out code pruning runtime alias checks.
30266 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30267 factored out from above.
30268 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30270 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30272 (prune_runtime_alias_test_list): New decalaration.
30274 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30276 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30277 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30278 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30279 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30280 (operator==, comp_dr_with_seg_len_pair): Ditto.
30281 (vect_prune_runtime_alias_test_list): Ditto.
30283 2017-05-26 Martin Liska <mliska@suse.cz>
30286 * params.def: Bound partial-inlining-entry-probability param.
30288 2017-05-26 Marek Polacek <polacek@redhat.com>
30291 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30294 2017-05-26 Richard Biener <rguenther@suse.de>
30296 PR tree-optimization/80842
30297 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30300 2017-05-26 Richard Biener <rguenther@suse.de>
30302 PR tree-optimization/80844
30303 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30305 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30307 * doc/md.texi (Machine Constraints): Update x86 family
30308 machine constraints section to match 'config/i386/constraints.md'.
30310 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30312 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30314 2017-05-25 Nathan Sidwell <nathan@acm.org>
30316 * doc/invoke.texi (--enable-languages): Update documentation.
30318 2017-05-25 Martin Liska <mliska@suse.cz>
30320 * dumpfile.c: Add TDF_FOLDING.
30321 * dumpfile.h (enum dump_kind): Likewise.
30322 * genmatch.c (dt_simplify::gen_1): Use it.
30324 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30326 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30328 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30330 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30331 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30333 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30335 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30336 * match.pd (X == C): Rewrite it here.
30337 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30338 with_certain_nonzero_bits2): New predicates.
30339 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30341 2017-05-24 Nathan Sidwell <nathan@acm.org>
30343 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30346 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30349 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30351 * config/powerpcspe: New port. Files are copied from the rs6000
30352 port, with "rs6000" in filenames replaced by "powerpcspe".
30354 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30356 PR rtl-optimization/80754
30357 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30359 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30361 * config/sparc/sparc.md (length): Return the correct value for -mflat
30362 sibcalls to match output_sibcall.
30364 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30368 * config/rs6000/rs6000.c (struct machine_function): Add new field
30370 (rs6000_get_separate_components): Init that field, use it.
30371 (rs6000_components_for_bb): Use the field.
30373 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30375 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30377 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30379 PR middle-end/80823
30380 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30382 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30385 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30386 address against address_operand predicate.
30387 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30389 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30391 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30392 parameters passed indirectly.
30394 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30396 * config/i386/i386.md (*movdi_internal): Remove SSE4
30397 alternative 18 (?r, *v). Update insn attributes.
30398 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30399 Update insn attributes.
30400 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30401 Update insn attributes.
30402 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30403 alternative 1 (r, v). Remove isa attribute.
30404 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30405 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30406 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30408 2017-05-23 Tom de Vries <tom@codesourcery.com>
30410 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30413 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30415 * cgraphunit.c (symbol_table::process_new_functions): Update.
30416 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30417 (inline_generate_summary): Rename to ...
30418 (ipa_fn_summary_generate): ... this one.
30419 (inline_read_summary): Rename to ...
30420 (ipa_fn_summary_read): ... this one.
30421 (inline_write_summary): Rename to ...
30422 (ipa_fn_summary_write): ... this one.
30423 (inline_free_summary): Rename to ...
30424 (ipa_free_fn_summary): ... this one.
30425 (pass_data_local_fn_summary, pass_local_fn_summary,
30426 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30427 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30428 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30429 make_pass_ipa_fn_summary): New.
30430 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30431 inline_write_summary, inline_free_summary): Remove.
30432 (ipa_free_fn_summary) : New.
30433 * ipa-inline.c (ipa_inline): Update.
30434 (pass_ipa_inline): Do not generate summaries.
30435 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30437 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30438 and add pass_ipa_fn_summary.
30439 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30441 (make_pass_inline_parameters): Remove.
30443 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30445 * omp-low.c (struct omp_context): Remove "default_kind" member.
30448 * omp-offload.c (execute_oacc_device_lower): Remove the
30449 parallelism dimensions function attributes for unparallelized
30450 OpenACC kernels constructs.
30452 2017-05-23 Martin Liska <mliska@suse.cz>
30454 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30456 (cgraph_edge::make_speculative): Likewise.
30457 (cgraph_edge::resolve_speculation): Likewise.
30458 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30459 (cgraph_node::dump): Likewise.
30460 * cgraph.h: Likewise.
30461 * cgraphunit.c (analyze_functions): Likewise.
30462 (symbol_table::compile): Likewise.
30463 * ipa-cp.c (print_all_lattices): Likewise.
30464 (determine_versionability): Likewise.
30465 (initialize_node_lattices): Likewise.
30466 (ipcp_verify_propagated_values): Likewise.
30467 (estimate_local_effects): Likewise.
30468 (update_profiling_info): Likewise.
30469 (create_specialized_node): Likewise.
30470 (perhaps_add_new_callers): Likewise.
30471 (decide_about_value): Likewise.
30472 (decide_whether_version_node): Likewise.
30473 (identify_dead_nodes): Likewise.
30474 (ipcp_store_bits_results): Likewise.
30475 * ipa-devirt.c (dump_targets): Likewise.
30476 (ipa_devirt): Likewise.
30477 * ipa-icf.c (sem_item::dump): Likewise.
30478 (sem_function::equals): Likewise.
30479 (sem_variable::equals): Likewise.
30480 (sem_item_optimizer::read_section): Likewise.
30481 (sem_item_optimizer::execute): Likewise.
30482 (congruence_class::dump): Likewise.
30483 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30484 (dump_inline_summary): Likewise.
30485 (estimate_node_size_and_time): Likewise.
30486 (inline_analyze_function): Likewise.
30487 * ipa-inline-transform.c (inline_call): Likewise.
30488 * ipa-inline.c (report_inline_failed_reason): Likewise.
30489 (want_early_inline_function_p): Likewise.
30490 (edge_badness): Likewise.
30491 (update_edge_key): Likewise.
30492 (inline_small_functions): Likewise.
30493 * ipa-profile.c (ipa_profile): Likewise.
30494 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30495 (ipa_make_edge_direct_to_target): Likewise.
30496 (remove_described_reference): Likewise.
30497 (ipa_impossible_devirt_target): Likewise.
30498 (propagate_controlled_uses): Likewise.
30499 (ipa_print_node_params): Likewise.
30500 (ipcp_transform_function): Likewise.
30501 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30502 (propagate_pure_const): Likewise.
30503 * ipa-reference.c (generate_summary): Likewise.
30504 (read_write_all_from_decl): Likewise.
30505 (propagate): Likewise.
30506 (ipa_reference_read_optimization_summary): Likewise.
30507 * ipa-utils.c (ipa_merge_profiles): Likewise.
30508 * ipa.c (walk_polymorphic_call_targets): Likewise.
30509 (symbol_table::remove_unreachable_nodes): Likewise.
30510 (ipa_single_use): Likewise.
30511 * passes.c (execute_todo): Likewise.
30512 * predict.c (drop_profile): Likewise.
30513 * symtab.c (symtab_node::get_dump_name): New function.
30514 (symtab_node::dump_name): Likewise.
30515 (symtab_node::dump_asm_name): Likewise.
30516 (symtab_node::dump_references): Likewise.
30517 (symtab_node::dump_referring): Likewise.
30518 (symtab_node::dump_base): Likewise.
30519 (symtab_node::debug_symtab): Likewise.
30520 * tree-sra.c (convert_callers_for_node): Likewise.
30521 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30522 * value-prof.c (init_node_map): Likewise.
30524 2017-05-23 Martin Liska <mliska@suse.cz>
30526 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30527 and symtab_node::debug_symtab to symbol_table::debug.
30528 * cgraphunit.c (analyze_functions): Use the renamed function.
30529 (symbol_table::compile): Likewise.
30530 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30531 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30532 * passes.c (execute_todo): Likewise.
30533 * symtab.c (symbol_table::dump): New function.
30534 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30536 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30538 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30539 that nonconst implies exec.
30541 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30543 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30544 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30545 (inline_edge_summary_vec): Turn into ...
30546 (ipa_call_summaries): ... this one.
30547 (redirect_to_unreachable, edge_set_predicate,
30548 evaluate_properties_for_edge, inline_summary_alloc,
30549 reset_ipa_call_summary, reset_inline_summary,
30550 inline_summary_t::duplicate): Update.
30551 (inline_edge_duplication_hook): Turn to ...
30552 (ipa_call_summary_t::duplicate): ... this one.
30553 (inline_edge_removal_hook): Turn to ...
30554 (ipa_call_summary_t::remove): ... this one.
30555 (dump_inline_edge_summary): Turn to ...
30556 (dump_ipa_call_summary): ... this one.
30557 (estimate_function_body_sizes): Update.
30558 (inline_update_callee_summaries): Update.
30559 (remap_edge_change_prob): Update.
30560 (remap_edge_summaries): Update.
30561 (inline_merge_summary): Update.
30562 (do_estimate_edge_time): Update.
30563 (inline_generate_summary): Update.
30564 (inline_read_section): Update.
30565 (inline_read_summary): Update.
30566 (inline_free_summary): Update.
30567 * ipa-inline.c (can_inline_edge_p): Update.
30568 (compute_inlined_call_time): Update.
30569 (want_inline_small_function_p): Update.
30570 (edge_badness): Update.
30571 (early_inliner): Update.
30572 * ipa-inline.h (inline_edge_summary): Turn to ...
30573 (ipa_call_summary): ... this one.
30574 (ipa_call_summary_t): New class.
30575 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30576 (ipa_call_summaries): New.
30577 (inline_edge_summary): Remove.
30578 (estimate_edge_growth): Update.
30579 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30580 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30581 * ipa-split.c (execute_split_functions): Update.
30582 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30584 2017-05-23 Tom de Vries <tom@codesourcery.com>
30586 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30587 attributes): Document rdrand effective target.
30589 2017-05-23 Tom de Vries <tom@codesourcery.com>
30591 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30592 attributes): Sort alphabetically.
30594 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30596 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30598 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30601 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30602 V2DF/V2DI splat into two separate patterns, one that handles
30603 registers, and the other that only handles memory. Drop support
30604 for splatting from a GPR on ISA 2.07 and then splitting the
30605 splat into direct move and splat.
30606 (vsx_splat_<mode>_reg): Likewise.
30607 (vsx_splat_<mode>_mem): Likewise.
30609 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30611 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30613 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30615 PR middle-end/80809
30616 * omp-low.c (finish_taskreg_remap): New function.
30617 (finish_taskreg_scan): If unit size of ctx->record_type
30618 is non-constant, unshare the size expression and replace
30619 decls in it with possible outer var refs.
30621 PR middle-end/80809
30622 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30623 GOVD_SHARED rather than GOVD_PRIVATE with it.
30624 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30625 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30627 PR middle-end/80853
30628 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30629 as last argument to build_outer_var_ref for pointer bases of array
30630 section reductions.
30632 2017-05-19 Martin Sebor <msebor@redhat.com>
30634 * print-tree.c (print_node): Print DECL_READ_P flag.
30636 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30638 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30639 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30640 * cgraph.c: Likewise.
30641 * cgraphunit.c: Likewise.
30642 * gengtype.c: Likewise.
30643 * ipa-cp.c: Likewise.
30644 * ipa-devirt.c: Likewise.
30645 * ipa-icf.c: Likewise.
30646 * ipa-predicate.c: Likewise.
30647 * ipa-profile.c: Likewise.
30648 * ipa-prop.c: Likewise.
30649 * ipa-split.c: Likewise.
30651 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30652 edge_predicate_pool, dump_inline_hints,
30653 inline_summary::account_size_time, redirect_to_unreachable,
30654 edge_set_predicate, set_hint_predicate,
30655 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30656 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30657 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30658 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30659 ipa_call_summary_t::remove, initialize_growth_caches,
30660 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30661 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30662 mark_modified, unmodified_parm_1, unmodified_parm,
30663 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30664 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30665 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30666 will_be_nonconstant_predicate, record_modified_bb_info,
30667 get_minimal_bb, record_modified, param_change_prob,
30668 phi_result_unknown_predicate, predicate_for_phi_result,
30669 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30670 estimate_function_body_sizes, compute_inline_parameters,
30671 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30672 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30673 inline_update_callee_summaries, remap_edge_change_prob,
30674 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30675 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30676 inline_analyze_function, inline_summary_t::insert,
30677 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30678 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30679 inline_free_summary): Move to ipa-fnsummary.h
30680 (predicate_t): Remove.
30681 * ipa-fnsummary.c: New file.
30682 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30683 (enum inline_hints_vals, inline_hints, agg_position_info,
30684 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30685 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30686 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30687 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30688 inline_read_summary, inline_write_summary, inline_free_summary,
30689 inline_analyze_function, initialize_inline_failed,
30690 inline_merge_summary, inline_update_overall_summary,
30691 compute_inline_parameters): Move to ipa-fnsummary.h
30692 * ipa-fnsummary.h: New file.
30693 * ipa-inline-transform.h: Include ipa-inline.h.
30694 * ipa-inline.c: LIkewise.
30696 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30698 * ipa-inline.c (edge_badness): Use inlined_time instead of
30699 inline_summaries->get.
30701 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30703 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30705 2017-05-22 Nathan Sidwell <nathan@acm.org>
30707 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30708 (fdump-lang): Document 'raw' option.
30709 * dumpfile.h (TDI_tu): Delete.
30710 * dumpfile.c (dump_files): Remove translation-unit.
30711 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30713 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30715 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30716 command option from $(AWK) call.
30717 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30718 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30719 [FORMAT]: Remove handling of variable.
30720 * config/avr/t-multilib: Regenerate.
30722 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30724 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30726 (dump_inline_summary): Do not print self_time.
30727 (estimate_function_body_sizes): Do not set self_time.
30728 (compute_inline_parameters): Likewise.
30729 (inline_read_section, inline_write_summary): Do not stream self_time.
30730 * ipa-inline.h (inline_summary): Drop self_time.
30732 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30734 * ipa-inline-analysis.c (account_size_time): Rename to ...
30735 (inline_summary::account_size_time): ... this one.
30736 (reset_ipa_call_summary): Turn to ...
30737 (ipa_call_summary::reset): ... this one.
30738 (reset_inline_summary): Turn to ...
30739 (inline_summary::reset): ... this one.
30740 (inline_summary_t::remove): Update.
30741 (inline_summary_t::duplicate): Update.
30742 (ipa_call_summary_t::remove): Update.
30743 (dump_inline_summary): Update.
30744 (estimate_function_body_sizes): Update.
30745 (compute_inline_parameters): Update.
30746 (estimate_node_size_and_time): Update.
30747 (inline_merge_summary): Update.
30748 (inline_update_overall_summary): Update.
30749 (inline_read_section): Update.
30750 (inline_write_summary): Update.
30751 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30752 add account_size_time and reset member functions.
30753 (ipa_call_summary): Add reset function.
30754 * ipa-predicate.h (predicate::operator &): Constify.
30756 2017-05-22 Richard Biener <rguenther@suse.de>
30758 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30760 2017-05-19 Jason Merrill <jason@redhat.com>
30762 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30764 2017-05-19 Marek Polacek <polacek@redhat.com>
30767 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30768 TYPE_OVERFLOW_WRAPS checks.
30770 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30772 * tree-core.h (enum omp_clause_default_kind): Add
30773 "OMP_CLAUSE_DEFAULT_PRESENT".
30774 * tree-pretty-print.c (dump_omp_clause): Handle it.
30775 * gimplify.c (enum gimplify_omp_var_data): Add
30776 "GOVD_MAP_FORCE_PRESENT".
30777 (gimplify_adjust_omp_clauses_1): Map it to
30778 "GOMP_MAP_FORCE_PRESENT".
30779 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30781 * gimplify.c (oacc_default_clause): Clarify.
30783 2017-05-19 Nathan Sidwell <nathan@acm.org>
30785 LANG_HOOK_REGISTER_DUMPS
30786 * toplev.c (general_init): Call register dump lang hook.
30787 * doc/invoke.texi: Document -fdump-lang option family.
30788 * dumpfile.c (dump_files): Remove class dump here.
30789 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30790 * dumpfile.h (tree_dump_index): Remove TDI_class.
30791 * langhooks-def.h (lhd_register_dumps): Declare.
30792 (LANG_HOOKS_REGISTER_DUMPS): Define.
30793 (LANG_HOOKS_INITIALIZER): Add it.
30794 * langhooks.c (lhd_register_dumps): Define.
30795 * langhooks.h (struct lang_hooks): Add register_dumps.
30797 2017-05-19 Nathan Sidwell <nathan@acm.org>
30799 * context.h (context::set_passes): New.
30800 * context.c (context::context): Do not create pass manager.
30801 * toplev.c (general_init): Create pass manager here.
30803 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30805 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30806 use this splitter if two add or or instructions would also work for
30807 the constant we want to generate.
30809 2017-05-19 Richard Biener <rguenther@suse.de>
30812 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30813 predicate evaluation.
30815 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30817 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30819 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30822 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30824 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30825 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30826 (inline_edge_summary_vec): Turn into ...
30827 (ipa_call_summaries): ... this one.
30828 (redirect_to_unreachable, edge_set_predicate,
30829 evaluate_properties_for_edge, inline_summary_alloc,
30830 reset_ipa_call_summary, reset_inline_summary,
30831 inline_summary_t::duplicate): Update.
30832 (inline_edge_duplication_hook): Turn to ...
30833 (ipa_call_summary_t::duplicate): ... this one.
30834 (inline_edge_removal_hook): Turn to ...
30835 (ipa_call_summary_t::remove): ... this one.
30836 (dump_inline_edge_summary): Turn to ...
30837 (dump_ipa_call_summary): ... this one.
30838 (estimate_function_body_sizes): Update.
30839 (inline_update_callee_summaries): Update.
30840 (remap_edge_change_prob): Update.
30841 (remap_edge_summaries): Update.
30842 (inline_merge_summary): Update.
30843 (do_estimate_edge_time): Update.
30844 (inline_generate_summary): Update.
30845 (inline_read_section): Update.
30846 (inline_read_summary): Update.
30847 (inline_free_summary): Update.
30848 * ipa-inline.c (can_inline_edge_p): Update.
30849 (compute_inlined_call_time): Update.
30850 (want_inline_small_function_p): Update.
30851 (edge_badness): Update.
30852 (early_inliner): Update.
30853 * ipa-inline.h (inline_edge_summary): Turn to ...
30854 (ipa_call_summary): ... this one.
30855 (ipa_call_summary_t): New class.
30856 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30857 (ipa_call_summaries): New.
30858 (inline_edge_summary): Remove.
30859 (estimate_edge_growth): Update.
30860 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30861 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30862 * ipa-split.c (execute_split_functions): Update.
30863 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30865 2017-05-19 Richard Biener <rguenther@suse.de>
30867 PR middle-end/80764
30868 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30870 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30872 * config/rs6000/rs6000.c (struct machine_function): Add field
30873 fpr_is_wrapped_separately.
30874 (rs6000_get_separate_components): Use 64 components. Handle the
30875 new FPR components.
30876 (rs6000_components_for_bb): Handle the FPR components.
30877 (rs6000_emit_prologue_components): Handle the FPR components.
30878 (rs6000_emit_epilogue_components): Handle the FPR components.
30879 (rs6000_set_handled_components): Handle the FPR components.
30880 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30881 that are already separately shrink-wrapped.
30882 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30883 that are already separately shrink-wrapped.
30885 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30888 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30891 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30892 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30893 cases where the register allocator uses a move and an offsettable
30894 memory operation to/from a FPR register on ISA 2.06/2.07.
30895 (define_peephole2 for Altivec d-form store): Likewise.
30897 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30900 * config/i386/mmx.md (*mov<mode>_internal): Enable
30901 alternatives 11, 12, 13 and 14 also for 32bit targets.
30902 Remove alternatives 15, 16, 17 and 18.
30903 * config/i386/sse.md (vec_concatv2di): Change
30904 alternative (!x, *y) to (x, ?!*Yn).
30906 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30908 * dumpfile.h (enum dump_kind): Remove stray comma.
30910 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30912 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30913 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30914 predicate::num_conditions
30915 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30916 (CHANGED): turn into predicate::changed.
30917 (agg_position_info): Move to ipa-predicate.h
30918 (add_condition, predicate::add_clause, predicate::operator &=,
30919 predicate::or_with, predicate::evaluate, predicate::probability,
30920 dump_condition, dump_clause, predicate::dump,
30921 predicate::remap_after_duplication, predicate::remap_after_inlining,
30922 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30923 (evaluate_conditions_for_known_args): Update.
30924 (set_cond_stmt_execution_predicate): Update.
30925 * ipa-inline.h: Include ipa-predicate.h
30926 (condition, inline_param_summary, conditions, agg_position_info,
30927 predicate): Move to ipa-predicate.h
30928 * ipa-predicate.c: New file.
30929 * ipa-predicate.h: New file.
30931 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30933 * final.c (leaf_function_p): Check we are not in a sequence.
30935 2017-05-18 Martin Liska <mliska@suse.cz>
30937 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30938 * dumpfile.c (dump_register): Use new enum dump_kind.
30939 (get_dump_file_name): Likewise.
30940 (dump_enable_all): Likewise.
30941 (dump_switch_p_1): Likewise.
30942 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30943 * dumpfile.h (enum dump_kind): New enum type.
30944 (struct dump_file_info): Create constructor and
30945 format fields and comments.
30946 * passes.c (pass_manager::register_one_dump_file):
30948 * statistics.c (statistics_early_init): Likewise.
30949 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30950 TDF_TREE with TDF_SLIM.
30951 (gather_memory_references_ref): Likewise.
30953 2017-05-18 Martin Liska <mliska@suse.cz>
30955 * vec.h (struct vnull): Use it.
30957 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30959 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30960 (true_predicate, false_predicate, true_predicate_p,
30961 false_predicate_p): Remove.
30962 (single_cond_predicate, not_inlined_predicate): Turn to member function
30964 (add_condition): Update.
30965 (add_clause): Turn to...
30966 (predicate::add_clause): ... this one; update; allow passing NULL
30968 (and_predicates): Turn to ...
30969 (predicate::operator &=): ... this one.
30970 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30971 (or_predicates): Turn to ...
30972 (predicate::or_with): ... this one.
30973 (evaluate_predicate): Turn to ...
30974 (predicate::evaluate): ... this one.
30975 (predicate_probability): Turn to ...
30976 (predicate::probability): ... this one.
30977 (dump_condition): Update.
30978 (dump_predicate): Turn to ...
30979 (predicate::dump): ... this one.
30980 (account_size_time): Update.
30981 (edge_set_predicate): Update.
30982 (set_hint_predicate): UPdate.
30983 (evaluate_conditions_for_known_args): Update.
30984 (evaluate_properties_for_edge): Update.
30985 (remap_predicate_after_duplication): Turn to...
30986 (predicate::remap_after_duplication): ... this one.
30987 (remap_hint_predicate_after_duplication): Update.
30988 (inline_summary_t::duplicate): UPdate.
30989 (dump_inline_edge_summary): Update.
30990 (dump_inline_summary): Update.
30991 (set_cond_stmt_execution_predicate): Update.
30992 (set_switch_stmt_execution_predicate): Update.
30993 (compute_bb_predicates): Update.
30994 (will_be_nonconstant_expr_predicate): Update.
30995 (will_be_nonconstant_predicate): Update.
30996 (phi_result_unknown_predicate): Update.
30997 (predicate_for_phi_result): Update.
30998 (array_index_predicate): Update.
30999 (estimate_function_body_sizes): Update.
31000 (estimate_node_size_and_time): Update.
31001 (estimate_ipcp_clone_size_and_time): Update.
31002 (remap_predicate): Rename to ...
31003 (predicate::remap_after_inlining): ... this one.
31004 (remap_hint_predicate): Update.
31005 (inline_merge_summary): Update.
31006 (inline_update_overall_summary): Update.
31007 (estimate_size_after_inlining): Update.
31008 (read_predicate): Rename to ...
31009 (predicate::stream_in): ... this one.
31010 (read_inline_edge_summary): Update.
31011 (write_predicate): Rename to ...
31012 (predicate::stream_out): ... this one.
31013 (write_inline_edge_summary): Update.
31014 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31015 (clause_t): Turn to uint32_t
31016 (predicate): Turn to class; implement constructor and operators
31018 (size_time_entry): Update.
31019 (inline_summary): Update.
31020 (inline_edge_summary): Update.
31022 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31024 * fold-const.c (fold_binary_loc): Move transformation...
31025 * match.pd (C - X CMP X): ... here.
31027 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31029 * config/sparc/sparc.c (sparc_option_override): Set function
31030 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31031 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31033 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31035 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31037 2017-05-18 Marek Polacek <polacek@redhat.com>
31040 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31041 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31042 gimple_assign_load_p.
31044 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31046 PR middle-end/80692
31047 * real.c (do_compare): Give decimal_do_compare preference over
31048 comparing just the signs.
31050 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31052 * doc/md.texi (Canonicalization of Instructions): Describe the
31053 canonical form of instructions that inherently set a condition
31056 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31058 PR middle-end/80775
31059 * tree-cfg.c: Move deletion of unreachable case statements to after
31060 the merging of consecutive case labels.
31062 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31064 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31065 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31066 restoring of callee-saved registers.
31068 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31070 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31071 * config/visium/visium.c (single_set_and_flags): Likewise.
31072 * config/visium/visium.md (Substitutions): Likewise.
31074 2017-05-17 Martin Liska <mliska@suse.cz>
31076 * cfg.c: Introduce dump_flags_t type and
31077 use it instead of int type.
31079 * cfghooks.c: Likewise.
31080 * cfghooks.h (struct cfg_hooks): Likewise.
31081 * cfgrtl.c: Likewise.
31082 * cfgrtl.h: Likewise.
31083 * cgraph.c (cgraph_node::get_body): Likewise.
31084 * coretypes.h: Likewise.
31085 * domwalk.c: Likewise.
31086 * domwalk.h: Likewise.
31087 * dumpfile.c (struct dump_option_value_info): Likewise.
31088 (dump_enable_all): Likewise.
31089 (dump_switch_p_1): Likewise.
31090 (opt_info_switch_p): Likewise.
31091 * dumpfile.h (enum tree_dump_index): Likewise.
31092 (struct dump_file_info): Likewise.
31093 * genemit.c: Likewise.
31094 * generic-match-head.c: Likewise.
31095 * gengtype.c (open_base_files): Likewise.
31096 * gimple-pretty-print.c: Likewise.
31097 * gimple-pretty-print.h: Likewise.
31098 * graph.c (print_graph_cfg): Likewise.
31099 * graphite-scop-detection.c (dot_all_sese): Likewise.
31100 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31101 * loop-unroll.c (report_unroll): Likewise.
31102 * passes.c (pass_manager::register_one_dump_file): Likewise.
31103 * print-tree.c: Likewise.
31104 * statistics.c: Likewise.
31105 * tree-cfg.c: Likewise.
31106 * tree-cfg.h: Likewise.
31107 * tree-dfa.c: Likewise.
31108 * tree-dfa.h: Likewise.
31109 * tree-dump.c (dump_function): Likewise.
31110 * tree-dump.h (struct dump_info): Likewise.
31111 * tree-pretty-print.c: Likewise.
31112 * tree-pretty-print.h: Likewise.
31113 * tree-ssa-live.c: Likewise.
31114 * tree-ssa-live.h: Likewise.
31115 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31116 * tree-vect-loop.c: Likewise.
31117 * tree-vect-slp.c: Likewise.
31119 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31120 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31122 PR tree-optimization/80457
31123 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31124 of all arguments to a statement as scalar_to_vec operations.
31125 (vectorizable_call): Adjust call to vect_model_simple_cost for
31127 (vectorizable_conversion): Likewise.
31128 (vectorizable_assignment): Likewise.
31129 (vectorizable_shift): Likewise.
31130 (vectorizable_operation): Likewise.
31131 (vectorizable_comparison): Likewise.
31132 (vect_is_simple_cond): Record the def types for operands.
31133 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31134 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31135 for statement argument count.
31137 2017-05-16 Carl Love <cel@us.ibm.com>
31139 * config/rs6000/rs6000-c: Add support for built-in functions
31140 vector unsigned long long vec_bperm (vector unsigned long long,
31141 vector unsigned char)
31142 vector signed long long vec_mule (vector signed int,
31144 vector unsigned long long vec_mule (vector unsigned int,
31145 vector unsigned int)
31146 vector signed long long vec_mulo (vector signed int,
31148 vector unsigned long long vec_mulo (vector unsigned int,
31149 vector unsigned int)
31150 vector signed char vec_sldw (vector signed char,
31151 vector signed char,
31153 vector unsigned char vec_sldw (vector unsigned char,
31154 vector unsigned char,
31156 vector signed short vec_sldw (vector signed short,
31157 vector signed short,
31159 vector unsigned short vec_sldw (vector unsigned short,
31160 vector unsigned short,
31162 vector signed int vec_sldw (vector signed int,
31165 vector unsigned int vec_sldw (vector unsigned int,
31166 vector unsigned int,
31168 vector signed long long vec_sldw (vector signed long long,
31169 vector signed long long,
31171 vector unsigned long long vec_sldw (vector unsigned long long,
31172 vector unsigned long long,
31174 * config/rs6000/rs6000-c: Add support for built-in functions
31175 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31176 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31177 * doc/extend.texi: Update the built-in documentation for the
31178 new built-in functions.
31180 2017-05-16 Marek Polacek <polacek@redhat.com>
31184 * tree.c (save_expr): Don't fold the expression.
31186 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31188 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31189 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31190 and (?*y,m). Update insn attributes.
31192 2017-05-16 Martin Liska <mliska@suse.cz>
31194 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31195 flags argument of print_gimple_stmt, print_gimple_expr,
31196 print_generic_stmt and print_generic_expr.
31197 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31198 * coretypes.h: Likewise.
31199 * except.c (dump_eh_tree): Likewise.
31200 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31201 * gimple-pretty-print.h: Likewise.
31202 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31203 (backprop::push_to_worklist): Likewise.
31204 (backprop::pop_from_worklist): Likewise.
31205 (backprop::process_use): Likewise.
31206 (backprop::intersect_uses): Likewise.
31207 (note_replacement): Likewise.
31208 * gimple-ssa-store-merging.c
31209 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31210 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31211 (pass_store_merging::execute): Likewise.
31212 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31213 (ssa_base_cand_dump_callback): Likewise.
31214 (dump_incr_vec): Likewise.
31215 (replace_refs): Likewise.
31216 (replace_mult_candidate): Likewise.
31217 (create_add_on_incoming_edge): Likewise.
31218 (create_phi_basis): Likewise.
31219 (insert_initializers): Likewise.
31220 (all_phi_incrs_profitable): Likewise.
31221 (introduce_cast_before_cand): Likewise.
31222 (replace_one_candidate): Likewise.
31223 * gimplify.c (gimplify_expr): Likewise.
31224 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31225 (set_rename): Likewise.
31226 (rename_uses): Likewise.
31227 (copy_loop_phi_nodes): Likewise.
31228 (add_close_phis_to_merge_points): Likewise.
31229 (copy_loop_close_phi_args): Likewise.
31230 (copy_cond_phi_args): Likewise.
31231 (graphite_copy_stmts_from_block): Likewise.
31232 (translate_pending_phi_nodes): Likewise.
31233 * graphite-poly.c (print_pdr): Likewise.
31234 (dump_gbb_cases): Likewise.
31235 (dump_gbb_conditions): Likewise.
31236 (print_scop_params): Likewise.
31237 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31238 (build_cross_bb_scalars_use): Likewise.
31239 (gather_bbs::before_dom_children): Likewise.
31240 * hsa-dump.c (dump_hsa_immed): Likewise.
31241 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31242 (get_replacement_map): Likewise.
31243 * ipa-inline-analysis.c (dump_condition): Likewise.
31244 (estimate_function_body_sizes): Likewise.
31245 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31246 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31247 * ipa-prop.c (ipa_dump_param): Likewise.
31248 (ipa_print_node_jump_functions_for_edge): Likewise.
31249 (ipa_modify_call_arguments): Likewise.
31250 (ipa_modify_expr): Likewise.
31251 (ipa_dump_param_adjustments): Likewise.
31252 (ipa_dump_agg_replacement_values): Likewise.
31253 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31254 * ipa-pure-const.c (check_stmt): Likewise.
31255 (pass_nothrow::execute): Likewise.
31256 * ipa-split.c (execute_split_functions): Likewise.
31257 * omp-offload.c (dump_oacc_loop_part): Likewise.
31258 (dump_oacc_loop): Likewise.
31259 * trans-mem.c (tm_log_emit): Likewise.
31260 (tm_memopt_accumulate_memops): Likewise.
31261 (dump_tm_memopt_set): Likewise.
31262 (dump_tm_memopt_transform): Likewise.
31263 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31264 (print_loop): Likewise.
31265 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31266 (chkp_gather_checks_info): Likewise.
31267 (chkp_get_check_result): Likewise.
31268 (chkp_remove_check_if_pass): Likewise.
31269 (chkp_use_outer_bounds_if_possible): Likewise.
31270 (chkp_reduce_bounds_lifetime): Likewise.
31271 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31272 (chkp_mark_completed_bounds): Likewise.
31273 (chkp_register_incomplete_bounds): Likewise.
31274 (chkp_mark_invalid_bounds): Likewise.
31275 (chkp_maybe_copy_and_register_bounds): Likewise.
31276 (chkp_build_returned_bound): Likewise.
31277 (chkp_get_bound_for_parm): Likewise.
31278 (chkp_build_bndldx): Likewise.
31279 (chkp_get_bounds_by_definition): Likewise.
31280 (chkp_generate_extern_var_bounds): Likewise.
31281 (chkp_get_bounds_for_decl_addr): Likewise.
31282 * tree-chrec.c (chrec_apply): Likewise.
31283 * tree-data-ref.c (dump_data_reference): Likewise.
31284 (dump_subscript): Likewise.
31285 (dump_data_dependence_relation): Likewise.
31286 (analyze_overlapping_iterations): Likewise.
31287 * tree-inline.c (expand_call_inline): Likewise.
31288 (tree_function_versioning): Likewise.
31289 * tree-into-ssa.c (dump_defs_stack): Likewise.
31290 (dump_currdefs): Likewise.
31291 (dump_names_replaced_by): Likewise.
31292 (dump_update_ssa): Likewise.
31293 (update_ssa): Likewise.
31294 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31295 * tree-parloops.c (build_new_reduction): Likewise.
31296 (try_create_reduction_list): Likewise.
31297 (ref_conflicts_with_region): Likewise.
31298 (oacc_entry_exit_ok_1): Likewise.
31299 (oacc_entry_exit_single_gang): Likewise.
31300 * tree-pretty-print.h: Likewise.
31301 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31302 (get_scalar_evolution): Likewise.
31303 (add_to_evolution): Likewise.
31304 (get_loop_exit_condition): Likewise.
31305 (analyze_evolution_in_loop): Likewise.
31306 (analyze_initial_condition): Likewise.
31307 (analyze_scalar_evolution): Likewise.
31308 (instantiate_scev): Likewise.
31309 (number_of_latch_executions): Likewise.
31310 (gather_chrec_stats): Likewise.
31311 (final_value_replacement_loop): Likewise.
31312 (scev_const_prop): Likewise.
31313 * tree-sra.c (dump_access): Likewise.
31314 (disqualify_candidate): Likewise.
31315 (create_access): Likewise.
31316 (reject): Likewise.
31317 (maybe_add_sra_candidate): Likewise.
31318 (create_access_replacement): Likewise.
31319 (analyze_access_subtree): Likewise.
31320 (analyze_all_variable_accesses): Likewise.
31321 (sra_modify_assign): Likewise.
31322 (initialize_constant_pool_replacements): Likewise.
31323 (find_param_candidates): Likewise.
31324 (decide_one_param_reduction): Likewise.
31325 (replace_removed_params_ssa_names): Likewise.
31326 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31327 * tree-ssa-copy.c (dump_copy_of): Likewise.
31328 (copy_prop_visit_cond_stmt): Likewise.
31329 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31330 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31331 (record_equivalences_from_stmt): Likewise.
31332 * tree-ssa-dse.c (compute_trims): Likewise.
31333 (delete_dead_call): Likewise.
31334 (delete_dead_assignment): Likewise.
31335 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31336 (forward_propagate_into_cond): Likewise.
31337 (pass_forwprop::execute): Likewise.
31338 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31339 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31341 (move_computations_worker): Likewise.
31342 (execute_sm): Likewise.
31343 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31344 (remove_exits_and_undefined_stmts): Likewise.
31345 (remove_redundant_iv_tests): Likewise.
31346 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31347 (adjust_iv_update_pos): Likewise.
31348 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31349 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31350 (value_replacement): Likewise.
31351 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31352 * tree-ssa-pre.c (print_pre_expr): Likewise.
31353 (get_representative_for): Likewise.
31354 (create_expression_by_pieces): Likewise.
31355 (insert_into_preds_of_block): Likewise.
31356 (eliminate_insert): Likewise.
31357 (eliminate_dom_walker::before_dom_children): Likewise.
31358 (eliminate): Likewise.
31359 (remove_dead_inserted_code): Likewise.
31360 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31361 * tree-ssa-reassoc.c (get_rank): Likewise.
31362 (eliminate_duplicate_pair): Likewise.
31363 (eliminate_plus_minus_pair): Likewise.
31364 (eliminate_not_pairs): Likewise.
31365 (undistribute_ops_list): Likewise.
31366 (eliminate_redundant_comparison): Likewise.
31367 (update_range_test): Likewise.
31368 (optimize_range_tests_var_bound): Likewise.
31369 (optimize_vec_cond_expr): Likewise.
31370 (rewrite_expr_tree): Likewise.
31371 (rewrite_expr_tree_parallel): Likewise.
31372 (linearize_expr): Likewise.
31373 (break_up_subtract): Likewise.
31374 (linearize_expr_tree): Likewise.
31375 (attempt_builtin_powi): Likewise.
31376 (attempt_builtin_copysign): Likewise.
31377 (transform_stmt_to_copy): Likewise.
31378 (transform_stmt_to_multiply): Likewise.
31379 (dump_ops_vector): Likewise.
31380 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31381 (print_scc): Likewise.
31382 (set_ssa_val_to): Likewise.
31383 (visit_reference_op_store): Likewise.
31384 (visit_use): Likewise.
31385 (sccvn_dom_walker::before_dom_children): Likewise.
31386 (run_scc_vn): Likewise.
31387 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31389 (expr_hash_elt::print): Likewise.
31390 (const_and_copies::pop_to_marker): Likewise.
31391 (const_and_copies::record_const_or_copy_raw): Likewise.
31392 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31393 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31394 (dump_predicates): Likewise.
31395 (find_uninit_use): Likewise.
31396 (warn_uninitialized_phi): Likewise.
31397 (pass_late_warn_uninitialized::execute): Likewise.
31398 * tree-ssa.c (verify_vssa): Likewise.
31399 (verify_ssa): Likewise.
31400 (maybe_optimize_var): Likewise.
31401 * tree-vrp.c (dump_value_range): Likewise.
31402 (dump_all_value_ranges): Likewise.
31403 (dump_asserts_for): Likewise.
31404 (register_edge_assert_for_2): Likewise.
31405 (vrp_visit_cond_stmt): Likewise.
31406 (vrp_visit_switch_stmt): Likewise.
31407 (vrp_visit_stmt): Likewise.
31408 (vrp_visit_phi_node): Likewise.
31409 (simplify_cond_using_ranges_1): Likewise.
31410 (fold_predicate_in): Likewise.
31411 (evrp_dom_walker::before_dom_children): Likewise.
31412 (evrp_dom_walker::push_value_range): Likewise.
31413 (evrp_dom_walker::pop_value_range): Likewise.
31414 (execute_early_vrp): Likewise.
31416 2017-05-16 Richard Biener <rguenther@suse.de>
31418 * dwarf2out.c (loc_list_from_tree_1): Do not create
31419 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31421 2017-05-16 Richard Biener <rguenther@suse.de>
31423 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31425 (note_variable_value_in_expr): If we resolved the decl ref
31426 do not push to the stack.
31428 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31430 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31431 operations in fast-math mode.
31432 (vaddq_f16): Likewise.
31433 (vmul_f16): Likewise.
31434 (vmulq_f16): Likewise.
31435 (vsub_f16): Likewise.
31436 (vsubq_f16): Likewise.
31437 * config/arm/neon.md (add<mode>3): New.
31439 (fma:<VH:mode>3): New. Also remove outdated comment.
31442 2017-05-16 Martin Liska <mliska@suse.cz>
31446 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31447 (odr_types_equivalent_p): Likewise.
31449 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31451 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31453 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31456 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31457 non-interunit SSE move alternatives with '?'.
31458 (zero-extendsidi peephole2): New peephole to skip intermediate
31459 general register in SSE zero-extend sequence.
31461 2017-05-15 Jeff Law <law@redhat.com>
31463 * reorg.c (relax_delay_slots): Create a new variable to hold
31464 the temporary target rather than clobbering TARGET_LABEL.
31466 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31467 missing argument to extract_bit_field call.
31468 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31470 2017-05-15 Martin Liska <mliska@suse.cz>
31473 * gcc.c (process_command): Do not allow empty argument of -o option.
31475 2017-05-15 Renlin Li <renlin.li@arm.com>
31477 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31478 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31479 * config/aarch64/constraints.md (Usf): Add long call check.
31480 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31481 (call_value): Likewise.
31482 (sibcall): Likewise.
31483 (sibcall_value): Likewise.
31485 (call_value_insn): New.
31486 (sibcall_insn): Update rtx pattern.
31487 (sibcall_value_insn): Likewise.
31488 (call_internal): Remove.
31489 (call_value_internal): Likewise.
31490 (sibcall_internal): Likewise.
31491 (sibcall_value_internal): Likewise.
31492 (call_reg): Likewise.
31493 (call_symbol): Likewise.
31494 (call_value_reg): Likewise.
31495 (call_value_symbol): Likewise.
31497 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31500 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31502 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31504 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31505 compatible with CCGOCmode and with CCZmode.
31507 2017-05-14 Martin Sebor <msebor@redhat.com>
31509 PR middle-end/77671
31510 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31511 (gimple_fold_builtin_snprintf): Same.
31512 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31513 (gimple_fold_builtin_snprintf): Same.
31514 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31515 of character types.
31516 (is_call_safe): New function.
31517 (try_substitute_return_value): Call it.
31518 (try_simplify_call): New function.
31519 (pass_sprintf_length::handle_gimple_call): Call it.
31521 2017-05-14 Martin Sebor <msebor@redhat.com>
31523 PR middle-end/80669
31524 * builtins.c (expand_builtin_stpncpy): Simplify.
31526 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31528 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31529 * config/i386/i386.h
31530 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31531 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31532 (struct machine_function): Add new members call_ms2sysv,
31533 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31534 (struct machine_frame_state): New fields sp_realigned and
31535 sp_realigned_offset.
31536 * config/i386/i386.c
31537 (enum xlogue_stub): New enum.
31538 (enum xlogue_stub_sets): New enum.
31539 (class xlogue_layout): New class.
31540 (struct ix86_frame): New fields stack_realign_allocate_offset,
31541 stack_realign_offset and outlined_save_offset. Modify comments to
31542 detail stack layout when using out-of-line stubs.
31543 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31544 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31545 -mcall-ms2sysv-xlogues.
31546 (stub_managed_regs): New static variable.
31547 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31548 registers managed by out-of-line stub.
31549 (disable_call_ms2sysv_xlogues): New function.
31550 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31551 m->call_ms2sysv when appropriate and compute frame layout for
31553 (sp_valid_at, fp_valid_at): New inline functions.
31554 (choose_basereg): New function.
31555 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31557 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31558 Use align parameter of choose_baseaddr to generated aligned SSE movs
31560 (pro_epilogue_adjust_stack): Modify to track
31561 machine_frame_state::sp_realigned.
31562 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31563 (ix86_nsaved_sseregs): Likewise.
31564 (ix86_emit_save_regs): Likewise.
31565 (ix86_emit_save_regs_using_mov): Likewise.
31566 (ix86_emit_save_sse_regs_using_mov): Likewise.
31567 (get_scratch_register_on_entry): Likewise.
31568 (gen_frame_set): New function.
31569 (gen_frame_load): Likewise.
31570 (gen_frame_store): Likewise.
31571 (emit_outlined_ms2sysv_save): Likewise.
31572 (emit_outlined_ms2sysv_restore): Likewise.
31573 (ix86_expand_prologue): Modify stack re-alignment code and call
31574 emit_outlined_ms2sysv_save when appropriate.
31575 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31576 parameter rtx_insn *insn, which allows the function to be used to only
31577 generate the notes.
31578 (ix86_expand_epilogue): Modify validity checks of frame and stack
31579 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31580 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31581 * config/i386/predicates.md
31582 (save_multiple): New predicate.
31583 (restore_multiple): Likewise.
31584 * config/i386/sse.md
31585 (save_multiple<mode>): New pattern.
31586 (save_multiple_realign<mode>): Likewise.
31587 (restore_multiple<mode>): Likewise.
31588 (restore_multiple_and_return<mode>): Likewise.
31589 (restore_multiple_leave_return<mode>): Likewise.
31590 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31592 2017-05-14 Julia Koval <julia.koval@intel.com>
31594 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31595 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31596 (__builtin_ia32_xsetbv): New builtins.
31597 * config/i386/i386.c (ix86_expand_special_args_builtin):
31599 (ix86_expand_builtin): Special expand for new intrinsics.
31600 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31601 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31602 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31604 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31606 * cfganal.c (inverted_post_order_compute): Change argument type
31608 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31609 * df-core.c (rest_of_handle_df_initialize): Adjust.
31610 (rest_of_handle_df_finish): Likewise.
31611 (df_analyze_1): Likewise.
31612 (df_analyze): Likewise.
31613 (loop_inverted_post_order_compute): Change argument to be a vec *.
31614 (df_analyze_loop): Adjust.
31615 (df_get_n_blocks): Likewise.
31616 (df_get_postorder): Likewise.
31617 * df.h (struct df_d): Change field to be a vec.
31618 * lcm.c (compute_laterin): Adjust.
31619 (compute_available): Likewise.
31620 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31621 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31622 * tree-ssa-pre.c (compute_antic): Likewise.
31624 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31626 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31627 (depth_first_search::depth_first_search): Change structure init
31628 function to this constructor.
31629 (depth_first_search::add_bb): Rename function to this member.
31630 (depth_first_search::execute): Likewise.
31631 (flow_dfs_compute_reverse_finish): Adjust.
31633 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31635 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31636 (longest_simple_path): Likewise.
31637 * shrink-wrap.c (spread_components): Likewise.
31638 (disqualify_problematic_components): Likewise.
31639 (emit_common_heads_for_components): Likewise.
31640 (emit_common_tails_for_components): Likewise.
31641 (insert_prologue_epilogue_for_components): Likewise.
31643 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31645 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31648 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31650 * df-core.c (df_set_blocks): Start using auto_bitmap.
31651 (df_compact_blocks): Likewise.
31652 * df-problems.c (df_rd_confluence_n): Likewise.
31653 * df-scan.c (df_insn_rescan_all): Likewise.
31654 (df_process_deferred_rescans): Likewise.
31655 (df_update_entry_block_defs): Likewise.
31656 (df_update_exit_block_uses): Likewise.
31657 (df_entry_block_bitmap_verify): Likewise.
31658 (df_exit_block_bitmap_verify): Likewise.
31659 (df_scan_verify): Likewise.
31660 * lra-constraints.c (lra_constraints): Likewise.
31661 (undo_optional_reloads): Likewise.
31662 (lra_undo_inheritance): Likewise.
31663 * lra-remat.c (calculate_gen_cands): Likewise.
31664 (do_remat): Likewise.
31665 * lra-spills.c (assign_spill_hard_regs): Likewise.
31666 (spill_pseudos): Likewise.
31667 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31668 (bitmap_set_subtract_values): Likewise.
31670 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31672 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31673 management with auto_bitmap.
31674 (fix_inter_tick): Likewise.
31675 (fix_recovery_deps): Likewise.
31676 * ira.c (add_store_equivs): Likewise.
31677 (find_moveable_pseudos): Likewise.
31678 (split_live_ranges_for_shrink_wrap): Likewise.
31679 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31680 (rtx_reuse_manager::seen_def_p): Likewise.
31681 (rtx_reuse_manager::set_seen_def): Likewise.
31682 * print-rtl.h (class rtx_reuse_manager): Likewise.
31684 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31686 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31688 (migrate_btr_def): Likewise.
31689 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31690 * df-core.c (loop_post_order_compute): Likewise.
31691 (loop_inverted_post_order_compute): Likewise.
31692 * hsa-common.h: Likewise.
31693 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31694 * init-regs.c (initialize_uninitialized_regs): Likewise.
31695 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31696 (inline_small_functions): Likewise.
31697 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31698 * ira.c (combine_and_move_insns): Likewise.
31699 (build_insn_chain): Likewise.
31700 * loop-invariant.c (find_invariants): Likewise.
31701 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31702 * predict.c (tree_predict_by_opcode): Likewise.
31703 (predict_paths_leading_to): Likewise.
31704 (predict_paths_leading_to_edge): Likewise.
31705 (estimate_loops_at_level): Likewise.
31706 (estimate_loops): Likewise.
31707 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31708 (spread_components): Likewise.
31709 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31710 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31711 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31712 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31713 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31714 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31715 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31716 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31717 (mark_threaded_blocks): Likewise.
31718 (thread_through_all_blocks): Likewise.
31719 * tree-ssa.c (verify_ssa): Likewise.
31720 (execute_update_addresses_taken): Likewise.
31721 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31723 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31725 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31727 (post_order_compute): Likewise.
31728 (inverted_post_order_compute): Likewise.
31729 (pre_and_rev_post_order_compute_fn): Likewise.
31731 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31733 * genrecog.c (int_set::int_set): Explicitly construct our
31734 auto_vec base class.
31735 * vec.h (auto_vec::auto_vec): New constructor.
31737 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31739 * bitmap.h (class auto_bitmap): New constructor taking
31740 bitmap_obstack * argument.
31742 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31744 * bitmap.h (class auto_bitmap): Change type of m_bits to
31745 bitmap_head, and adjust ctor / dtor and member operators.
31747 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31749 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31750 when returned register mode doesn't match original mode.
31752 2017-05-12 Jeff Law <law@redhat.com>
31753 Jakub Jelinek <jakub@redhat.com>
31755 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31756 we look for cc setter after the compare-elim changes.
31757 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31758 within the vector to match what compare-elim now expects.
31759 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31760 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31762 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31763 after the compare-elim changes.
31764 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31765 the vector to match what compare-elim now expects.
31766 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31767 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31768 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31769 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31770 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31772 * config/visium/visium.c (single_set_and_flags): Fix where
31773 we look for cc setter after the compare-elim changes.
31774 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31775 with the vector to match what compare-elim now expects.
31776 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31777 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31778 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31779 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31780 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31781 (neg<mode>2_insn_set_overflow): Likewise.
31783 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31785 PR middle-end/79794
31786 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31787 maybe_expand_insn call, set ops[0].target. If still set after call,
31788 set alt_rtl. Add extra arg to recursive calls.
31789 (extract_bit_field): Add alt_rtl argument. Pass to
31791 * expmed.h (extract_bit_field): Fix prototype.
31792 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31793 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31794 to extract_bit_field_calls.
31795 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31796 Pass alt_rtl to extract_bit_field calls.
31797 * calls.c (store_unaligned_arguments_into_psuedos)
31798 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31799 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31801 * optabs.h (struct expand_operand): Add target bitfield.
31803 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31805 * compare-elim.c (try_eliminate_compare): Canonicalize
31806 operation with embedded compare to
31807 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31808 (set (reg) (operation)].
31810 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31812 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31815 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31816 cost of adding a carry flag for ADC instruction.
31817 [case MINUS]: Ignore the cost of subtracting a carry flag
31818 for SBB instruction.
31820 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31822 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31824 * config/rs6000/bmiintrin.h: New file.
31825 * config/rs6000/bmi2intrin.h: New file.
31826 * config/rs6000/x86intrin.h: New file.
31828 2017-05-12 Jeff Law <law@redhat.com>
31830 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31833 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31835 PR middle-end/80707
31836 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31838 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31840 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31841 early expansion of vector divide builtins.
31842 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31843 builtins identified as having unsigned arguments.
31845 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31847 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31848 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31849 expansion of vector logical operations (and, andc, or, xor,
31852 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31854 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31855 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31857 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31859 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31860 early GIMPLE expansion of vector multiplies.
31862 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31864 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31865 TARGET_HAVE_MOVT conditional.
31866 (movt splitter): Likewise.
31868 2017-05-12 Richard Biener <rguenther@suse.de>
31870 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31871 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31872 Fold all stmts not inplace.
31874 2017-05-12 Richard Biener <rguenther@suse.de>
31876 PR tree-optimization/80713
31877 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31878 inserted_exprs bit for not removed stmts.
31880 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31882 PR middle-end/69921
31883 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31884 parallelized" attribute for parallelized OpenACC kernels.
31885 * omp-offload.c (execute_oacc_device_lower): Use it.
31887 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31888 Set "oacc kernels" attribute.
31889 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31890 parameter. Adjust all users.
31891 (oacc_fn_attrib_kernels_p): Remove function.
31892 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31893 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31894 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31895 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31896 assert "oacc kernels" attribute is set.
31898 2017-05-11 Carl Love <cel@us.ibm.com>
31900 * config/rs6000/rs6000-c: Add support for built-in functions
31901 vector unsigned char vec_popcnt (vector signed char)
31902 vector unsigned char vec_popcnt (vector unsigned char)
31903 vector unsigned short vec_popcnt (vector signed short)
31904 vector unsigned short vec_popcnt (vector unsigned short)
31905 vector unsigned int vec_popcnt (vector signed int)
31906 vector unsigned int vec_popcnt (vector unsigned int)
31907 vector unsigned long long vec_popcnt (vector signed long long)
31908 vector unsigned long long vec_popcnt (vector unsigned long long)
31909 vector signed long long vec_slo (vector signed long long,
31910 vector signed char)
31911 vector signed long long vec_slo (vector signed long long,
31912 vector unsigned char)
31913 vector unsigned long long vec_slo (vector unsigned long long,
31914 vector signed char)
31915 vector unsigned long long vec_slo (vector unsigned long long,
31916 vector unsigned char)
31917 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31918 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31919 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31920 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31921 * doc/extend.texi: Update the built-in documentation file for the
31922 new built-in functions.
31924 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31926 * attribs.h (sorted_attr_string): Move machine independent
31927 functions for target clone support from the i386 port to common
31928 code. Rename ix86_function_versions to common_function_versions.
31929 Rename make_name to make_unique_name.
31930 (common_function_versions): Likewise.
31931 (make_unique_name): Likewise.
31932 (make_dispatcher_decl): Likewise.
31933 (is_function_default_version): Likewise.
31934 * attribs.c (attr_strcmp): Likewise.
31935 (sorted_attr_string): Likewise.
31936 (common_function_versions): Likewise.
31937 (make_unique_name): Likewise.
31938 (make_dispatcher_decl): Likewise.
31939 (is_function_default_version): Likewise.
31940 * config/i386/i386.c (attr_strcmp): Likewise.
31941 (sorted_attr_string): Likewise.
31942 (ix86_function_versions): Likewise.
31943 (make_name): Likewise.
31944 (make_dispatcher_decl): Likewise.
31945 (is_function_default_version): Likewise.
31946 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31948 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31951 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31952 Account for direct move costs for vec_construct of integer
31955 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31958 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31959 (UNSPEC_STX_ATOMIC): Ditto.
31960 (loaddi_via_sse): New insn.
31961 (storedi_via_sse): Ditto.
31962 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31963 Update corresponding peephole2 patterns.
31964 (atomic_storedi_fpu): Ditto.
31966 2017-05-11 Julia Koval <julia.koval@intel.com>
31968 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31969 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31971 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31972 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31973 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31975 2017-05-11 Nathan Sidwell <nathan@acm.org>
31977 * graphite-poly.c: Include dumpfile.h.
31979 * dumpfle.h (dump_function): Declare here ...
31980 * tree-dump.h (dump_function): ... not here.
31981 * dumpfile.c: #include tree-cfg.h.
31982 (dump_function): Move here from ...
31983 * tree-dump.c (dump_function): ... here.
31984 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31985 * graphite-poly.c: Don't include tree-dump.h.
31986 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31987 * print-tree.c: Likewise.
31988 * stor-layout.c: Likewise.
31989 * tree-nested.c: Likewise.
31991 * dumpfile.c (dump_start): Use TDF_FLAGS.
31992 (dump_enable_all): Fix TDF_KIND check thinko.
31994 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31996 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31997 array entries to represent two legal parameterizations of the
31998 overloaded __builtin_cmpb function, as represented by the
31999 P6_OV_BUILTIN_CMPB constant.
32000 (altivec_resolve_overloaded_builtin): Add special case handling
32001 for the __builtin_cmpb function, as represented by the
32002 P6_OV_BUILTIN_CMPB constant.
32003 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
32004 (BU_P6_64BIT_2): New macro.
32005 (BU_P6_OVERLOAD_2): New macro
32006 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32007 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32008 (CMPB): Add overload support to represent both 32-bit and 64-bit
32009 compare-bytes function.
32010 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32011 support for TARGET_CMPB.
32012 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32013 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32014 documentation of the __builtin_cmpb overloaded built-in function.
32016 2017-05-11 Richard Biener <rguenther@suse.de>
32018 PR tree-optimization/80705
32019 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32020 bases are not vectorizable.
32022 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32024 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32025 when counting register pressure.
32027 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32029 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32030 (iv_ca_more_deps): Renamed to ...
32031 (iv_ca_compare_deps): ... this.
32032 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32034 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32036 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32038 (determine_group_iv_costs): ... here.
32039 (find_inv_vars_cb): Record inv var if it's not recorded before.
32041 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32043 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32044 (get_shiftadd_cost): Ditto.
32046 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32048 * tree-ssa-address.c: Include header file.
32049 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32051 (add_to_parts): Refactor.
32052 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32053 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32056 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32058 PR tree-optimization/53090
32059 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32061 (extract_cond_operands): Detect condition with IV on both sides
32062 and return COMP_IV_EXPR_2.
32063 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32064 (rewrite_use_compare): Simplify by removing call to function
32065 extract_cond_operands.
32067 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32069 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32070 (extract_cond_operands): Detect condition comparing against non-
32071 invariant bound and return appropriate enum value.
32072 (find_interesting_uses_cond): Update use of extract_cond_operands.
32073 Handle its return value accordingly.
32074 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32076 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32078 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32079 nonlinear iv_use computation in loop invariant sensitive way.
32081 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32083 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32084 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32086 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32088 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32089 (dump_cand): Support iv_cand.inv_exprs.
32090 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32092 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32095 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32097 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32099 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32100 as local function. Include necessary header files.
32101 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32103 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32105 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32107 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32109 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32110 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32111 RSHIFT_EXPR and BIT_NOT_EXPR.
32113 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32115 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32116 (adjust_setup_cost): New parameter supporting round up adjustment.
32117 (struct address_cost_data): Delete.
32118 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32119 (split_address_cost, ptr_difference_cost): Delete.
32120 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32121 (struct ainc_cost_data): New struct.
32122 (get_address_cost_ainc): New function.
32123 (get_address_cost, get_computation_cost): Reimplement.
32124 (determine_group_iv_cost_address): Record inv_expr for all uses of
32126 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32127 (iv_ca_has_deps): Reimplemented to ...
32128 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32130 (iv_ca_extend): Call iv_ca_more_deps.
32132 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32134 * tree-ssa-address.c (struct mem_address): Move to header file.
32135 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32136 * tree-ssa-address.h (struct mem_address): Move from C file.
32137 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32139 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32141 * tree-affine.h (aff_combination_type): New interface.
32142 (aff_combination_zero_p): Remove static.
32143 (aff_combination_const_p): New interface.
32144 (aff_combination_singleton_var_p): New interfaces.
32146 2017-05-11 Richard Biener <rguenther@suse.de>
32148 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32149 Skip unreachable blocks and destinations.
32150 (eliminate): Move stmt removal and fixup ...
32151 (fini_eliminate): ... here. Skip inserted exprs.
32152 (pass_pre::execute): Move fini_pre after fini_eliminate.
32153 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32154 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32155 PRE to get rid of dead code that has invalid SSA form and
32156 split critical edges again.
32158 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32160 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32162 2017-05-11 Richard Biener <rguenther@suse.de>
32164 * passes.c (execute_function_todo): Verify loops if they are
32165 said to be up-to-date.
32166 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32167 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32169 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32172 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32173 handle calling assemble_external ourself.
32176 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32177 modes with zero size. Enhance comment.
32179 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32181 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32182 built-ins for vec_xl and vec_xst with short and char pointer
32185 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32187 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32188 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32189 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32190 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32191 (_mm_maskz_min_round_ss): New intrinsics.
32192 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32193 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32194 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32195 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32196 (__builtin_ia32_minss_mask_round): New builtins.
32197 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32198 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32199 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32201 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32202 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32204 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32207 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32209 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32210 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32211 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32212 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32213 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32214 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32215 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32216 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32217 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32218 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32219 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32220 (__builtin_ia32_mulss_mask_round): New builtins.
32221 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32222 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32223 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32225 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32226 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32228 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32231 2017-05-10 Julia Koval <julia.koval@intel.com>
32233 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32234 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32235 (_mm256_setr_m128i): New intrinsics.
32237 2017-05-10 Julia Koval <julia.koval@intel.com>
32239 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32240 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32241 (_mm_maskz_rcp14_ss): New intrinsics.
32242 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32243 (__builtin_ia32_rcp14ss_mask): New builtins.
32244 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32246 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32248 PR tree-optimization/51513
32249 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32250 (assert_unreachable_fallthru_edge_p): Use it.
32251 (group_case_labels_stmt): Likewise.
32252 * tree-cfg.h: Prototype it.
32253 * stmt.c: Include cfghooks.h and tree-cfg.h.
32254 (emit_case_dispatch_table) <gap_label>: New local variable.
32255 Use it to fill dispatch table gaps.
32256 Test for default_label before updating probabilities.
32257 (expand_case) <default_label>: Remove unneeded initialization.
32258 Test for unreachable default case statement and remove its edge.
32259 Set default_label accordingly.
32260 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32262 2017-05-10 Carl Love <cel@us.ibm.com>
32264 * config/rs6000/rs6000-c: Add support for built-in functions
32265 vector signed char vec_neg (vector signed char)
32266 vector signed short int vec_neg (vector short int)
32267 vector signed int vec_neg (vector signed int)
32268 vector signed long long vec_neg (vector signed long long)
32269 vector float vec_neg (vector float)
32270 vector double vec_neg (vector double)
32271 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32273 * config/rs6000/altivec.h: Add define for vec_neg
32274 * doc/extend.texi: Update the built-in documentation for the
32275 new built-in functions.
32277 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32279 PR tree-optimization/77644
32280 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32282 2017-05-10 Nathan Sidwell <nathan@acm.org>
32284 * dumpfile.h (TDI_lang_all): New.
32285 (TDF_KIND): New. Renumber others
32286 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32288 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32290 (get_dump_file_name): Adjust suffix generation.
32291 (dump_enable_all): Use TDF_KIND.
32292 * doc/invoke.texi (-fdump-lang-all): Document.
32294 * dumpfile.h: Tabify.
32296 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32299 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32300 Move member access before delete.
32302 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32304 * tree-inline.c (expand_call_inline): Split block at stmt
32307 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32310 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32311 are now unused after splitting mov{sf,sd}_hardfloat.
32312 (f32_lr2): Likewise.
32313 (f32_lm): Likewise.
32314 (f32_lm2): Likewise.
32315 (f32_li): Likewise.
32316 (f32_li2): Likewise.
32317 (f32_lv): Likewise.
32318 (f32_sr): Likewise.
32319 (f32_sr2): Likewise.
32320 (f32_sm): Likewise.
32321 (f32_sm2): Likewise.
32322 (f32_si): Likewise.
32323 (f32_si2): Likewise.
32324 (f32_sv): Likewise.
32325 (f32_dm): Likewise.
32326 (f32_vsx): Likewise.
32327 (f32_av): Likewise.
32328 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32329 For movsf, order stores so the VSX stores occur before the GPR
32330 store which encourages the register allocator to use a traditional
32331 FPR instead of a GPR. For movsd, order the stores so that the GPR
32332 store comes before the VSX stores to allow the power6 to work.
32333 This is due to the power6 not having a 32-bit integer store
32334 instruction from a FPR.
32335 (movsf_hardfloat): Likewise.
32336 (movsd_hardfloat): Likewise.
32338 2017-05-09 Martin Sebor <msebor@redhat.com>
32340 PR translation/80280
32341 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32344 PR translation/80280
32345 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32346 data member added in r247778.
32347 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32349 2017-05-09 Nathan Sidwell <nathan@acm.org>
32351 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32353 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32356 2017-05-09 Marek Polacek <polacek@redhat.com>
32358 * doc/invoke.texi: Fix typo.
32360 2017-05-09 Richard Biener <rguenther@suse.de>
32362 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32363 (vrp_val_is_min): Likewise.
32364 (set_value_range_to_value): Likewise.
32365 (set_value_range_to_nonnegative): Likewise.
32366 (gimple_assign_nonzero_p): Likewise.
32367 (gimple_stmt_nonzero_p): Likewise.
32368 (vrp_int_const_binop): Likewise. Remove unreachable case.
32369 (adjust_range_with_scev): Adjust comments.
32370 (compare_range_with_value): Likewise.
32371 (extract_range_from_phi_node): Likewise.
32372 (test_for_singularity): Likewise.
32374 2017-05-09 Richard Biener <rguenther@suse.de>
32376 * tree-vrp.c (get_single_symbol): Add assert that we don't
32377 get overflowed constants as invariant part.
32378 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32379 checks. Use wi::cmp instead of recursing for integer constants.
32380 (compare_values): Just ignore whether we assumed undefined
32381 overflow instead of failing the compare.
32382 (extract_range_for_var_from_comparison_expr): Add comment before the
32383 TREE_NO_WARNING sets.
32384 (test_for_singularity): Likewise.
32385 (extract_range_from_comparison): Do not disable optimization
32386 when we assumed undefined overflow.
32387 (extract_range_basic): Remove init of unused var.
32389 2017-05-09 Richard Biener <rguenther@suse.de>
32391 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32392 (extract_range_from_multiplicative_op_1): Adjust.
32393 (extract_range_from_binary_expr_1): Use int_const_binop.
32395 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32398 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32399 rs6000_store_data_bypass_p in seven define_bypass directives and
32400 in several comments.
32401 * config/rs6000/rs6000-protos.h: Add prototype for
32402 rs6000_store_data_bypass_p function.
32403 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32404 function implements slightly different (rs6000-specific) semantics
32405 than store_data_bypass_p, returning false rather than aborting
32406 with assertion error when arguments do not satisfy the
32407 requirements of store data bypass.
32408 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32409 rs6000_store_data_bypass_p.
32411 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32413 * config/xtensa/xtensa-protos.h
32414 (xtensa_initial_elimination_offset): New declaration.
32415 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32416 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32417 macro definition, add case for FRAME_POINTER_REGNUM when
32418 FRAME_GROWS_DOWNWARD.
32419 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32420 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32421 xtensa_initial_elimination_offset.
32423 2017-05-08 Nathan Sidwell <nathan@acm.org>
32425 * doc/invoke.texi: Alphabetize -fdump options.
32427 2017-05-08 Martin Sebor <msebor@redhat.com>
32429 PR translation/80280
32430 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32432 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32434 * target.def (compute_frame_layout): New optional target hook.
32435 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32436 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32437 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32439 * reload1.c (verify_initial_elim_offsets): Likewise.
32440 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32441 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32442 (arm_get_frame_offsets): Split up into this ...
32443 (arm_compute_frame_layout): ... and this function.
32445 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32447 * config/aarch64/constraints.md (Usa): New constraint.
32448 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32450 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32452 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32453 with_multilib_list after it has been checked.
32455 2017-05-08 Richard Biener <rguenther@suse.de>
32457 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32458 (bitmap_set_subtract_values): Likewise.
32460 2017-05-08 Richard Biener <rguenther@suse.de>
32462 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32463 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32465 (gimple_stmt_nonzero_warnv_p): Rename to ...
32466 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32468 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32469 (extract_range_basic): Adjust, do not disable propagation on
32470 strict overflow sensitive simplification.
32471 (vrp_visit_cond_stmt): Likewise.
32473 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32475 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32476 body size unconditionally.
32478 2017-05-07 Jeff Law <law@redhat.com>
32481 2017-05-06 Jeff Law <law@redhat.com>
32482 PR tree-optimization/78496
32483 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32486 PR tree-optimization/78496
32487 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32488 (simplify_stmt_using_ranges): Call it.
32489 (vrp_dom_walker::before_dom_children): Extract equivalences
32490 from an ASSERT_EXPR with an equality comparison against a
32493 2017-05-06 Jeff Law <law@redhat.com>
32495 PR tree-optimization/78496
32496 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32499 PR tree-optimization/78496
32500 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32501 (simplify_stmt_using_ranges): Call it.
32502 (vrp_dom_walker::before_dom_children): Extract equivalences
32503 from an ASSERT_EXPR with an equality comparison against a
32506 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32508 * lra-constraints.c (lra_copy_reg_equiv): New function.
32509 (split_reg): Use it to copy equivalence information from the
32510 original register to the spill register.
32512 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32514 PR rtl-optimization/75964
32515 * simplify-rtx.c (simplify_const_relational_operation): Remove
32516 invalid handling of comparisons of integer ABS.
32518 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32520 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32521 initialize to zero.
32522 (init_regs): Remove declaration.
32523 (function_arg_advance_32): Initialize error_p as boolean variable.
32525 2017-05-05 Nathan Sidwell <nathan@acm.org>
32527 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32528 lines. Use for (;;).
32530 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32532 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32533 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32534 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32535 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32536 VF=2 that require versioning.
32538 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32540 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32543 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32545 * diagnostic.h (diagnostic_override_option_index): Convert from
32546 macro to inline function.
32548 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32550 * diagnostic.c (last_module_changed_p): New function.
32551 (set_last_module): New function.
32552 (diagnostic_report_current_module): Convert macro usage to
32553 the above functions.
32554 * diagnostic.h (diagnostic_context::last_module): Strengthen
32555 from const line_map * to const line_map_ordinary *.
32556 (diagnostic_last_module_changed): Delete macro.
32557 (diagnostic_set_last_module): Delete macro.
32559 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32561 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32562 with diagnostic_report_diagnostic.
32563 (diagnostic_n_impl_richloc): Likewise.
32564 * diagnostic.h (report_diagnostic): Delete macro.
32565 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32566 with diagnostic_report_diagnostic.
32567 * substring-locations.c (format_warning_va): Likewise.
32569 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32571 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32572 save/restor of format_spec. Move option-printing code to...
32573 (print_option_information): ...this new function, and
32574 reimplement by simply printing to the pretty_printer,
32575 rather than appending to the format string.
32577 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32579 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32580 handling logic into...
32581 (update_effective_level_from_pragmas): ...this new function.
32583 2017-05-04 Andrew Waterman <andrew@sifive.com>
32585 * config/riscv/riscv.opt (mstrict-align): New option.
32586 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32587 (SLOW_UNALIGNED_ACCESS): Define.
32588 (riscv_slow_unaligned_access): Declare.
32589 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32591 (riscv_slow_unaligned_access): New variable.
32592 (rocket_tune_info): Set slow_unaligned_access to true.
32593 (optimize_size_tune_info): Set slow_unaligned_access to false.
32594 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32595 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32596 (riscv_option_override): Set riscv_slow_unaligned_access.
32597 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32599 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32601 * config/riscv/riscv.md: Unify indentation.
32603 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32608 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32610 (extendsi<mode>2): Add support for doing sign extension via
32611 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32612 don't have ISA 3.0 instructions.
32613 (extendsi<mode>2 splitter): Likewise.
32614 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32615 generate the normal insns since SImode can now go in vector
32616 registers. Disallow the special UNSPECs needed for previous
32617 machines to hide SImode being used. Add new insns
32618 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32619 (fix_trunc<mode>si2_stfiwx): Likewise.
32620 (fix_trunc<mode>si2_internal): Likewise.
32621 (fixuns_trunc<mode>si2): Likewise.
32622 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32623 (fctiw<u>z_<mode>_smallint): Likewise.
32624 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32625 of floating point to 32-bit integer from doing a direct move to
32626 the GPR registers to do a store.
32627 (fctiwz_<mode>): Break long line.
32629 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32631 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32632 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32633 (addr_list, addr_offset_valid_p): New.
32634 (split_address_groups): Check offset validity with above function.
32635 (gt-tree-ssa-loop-ivopts.h): Include header file.
32637 2017-05-05 Nathan Sidwell <nathan@acm.org>
32639 * config.gcc (arm*-*-*): Add missing 'fi'.
32641 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32643 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32644 included in -fopt-info does not matter.
32645 * doc/optinfo.texi (-fopt-info): Fix description of default
32646 behavour. Explicitly say order of options included in -fopt-info
32649 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32651 * config.gcc: Allow combinations of aprofile and rmprofile values for
32652 --with-multilib-list.
32653 * config/arm/t-multilib: New file.
32654 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32655 variables. Remove setting of ISA and floating-point ABI in
32656 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32657 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32658 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32659 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32661 * config/arm/t-rmprofile: Likewise except for the matches changes.
32662 * doc/install.texi (--with-multilib-list): Document the combination of
32663 aprofile and rmprofile values and warn about pitfalls in doing that.
32665 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32667 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32668 (movdi_aarch64): Likewise.
32670 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32672 PR tree-optimization/80632
32673 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32675 (build_arrays): Initialize it for virtual phis.
32676 (fix_phi_nodes): Use it for virtual phis.
32678 PR tree-optimization/80558
32679 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32680 [x, y] op z into [x op, y op z] for op & or | if conditions
32683 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32684 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32687 * config/arm/arm.md (use_literal_pool): Remove.
32688 (64-bit immediate split): No longer takes cost into consideration
32689 if arm_disable_literal_pool is enabled.
32690 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32691 used when arm_disable_literal_pool is enabled.
32692 (arm_max_const_double_inline_cost): Remove use of
32693 arm_disable_literal_pool.
32694 (push_minipool_fix): Add assert.
32695 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32696 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32697 (no_literal_pool_sf_immediate): New.
32699 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32701 PR tree-optimization/80613
32702 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32703 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32705 2017-05-05 Richard Biener <rguenther@suse.de>
32707 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32709 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32711 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32712 of this flag from insn conditions due to removal from r247495.
32714 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32716 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32718 (arm_early_store_addr_dep_ptr): Likewise.
32719 * config/arm/aarch-common-protos.h
32720 (arm_early_load_addr_dep_ptr): Add prototype.
32721 (arm_early_store_addr_dep_ptr): Likewise.
32722 * config/arm/cortex-a53.md: Add new bypasses.
32724 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32726 * tree.c (next_type_uid): Change type to unsigned.
32727 (type_hash_canon): Decrement back next_type_uid if
32728 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32729 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32732 2017-05-04 Martin Sebor <msebor@redhat.com>
32734 * builtins.c: Fix a trivial typo in a comment.
32736 PR middle-end/79234
32737 * builtins.c (check_sizes): Adjust to handle reading past the end.
32738 Avoid printing excessive upper bound of ranges. Use %E to print
32739 tree nodes instead of converting them to %wu.
32740 (expand_builtin_memchr): New function.
32741 (compute_dest_size): Rename...
32742 (compute_objsize): ...to this.
32743 (expand_builtin_memcpy): Adjust.
32744 (expand_builtin_mempcpy): Adjust.
32745 (expand_builtin_strcat): Adjust.
32746 (expand_builtin_strcpy): Adjust.
32747 (check_strncat_sizes): Adjust.
32748 (expand_builtin_strncat): Adjust.
32749 (expand_builtin_strncpy): Adjust and simplify.
32750 (expand_builtin_memset): Adjust.
32751 (expand_builtin_bzero): Adjust.
32752 (expand_builtin_memcmp): Adjust.
32753 (expand_builtin): Handle memcmp.
32754 (maybe_emit_chk_warning): Check strncat just once.
32756 2017-05-04 Martin Sebor <msebor@redhat.com>
32758 PR preprocessor/79214
32759 PR middle-end/79222
32760 PR middle-end/79223
32761 * builtins.c (check_sizes): Add inlining context and issue
32762 warnings even when -Wno-system-headers is set.
32763 (check_strncat_sizes): Same.
32764 (expand_builtin_strncat): Same.
32765 (expand_builtin_memmove): New function.
32766 (expand_builtin_stpncpy): Same.
32767 (expand_builtin): Handle memmove and stpncpy.
32769 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32771 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32772 which is not used any more.
32774 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32776 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32778 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32780 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32781 (cortexa53_tunings): Likewise.
32782 (cortexa57_tunings): Likewise.
32783 (cortexa72_tunings): Likewise.
32784 (cortexa73_tunings): Likewise.
32786 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32788 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32789 Set loop alignment to 8.
32791 2017-05-04 Martin Sebor <msebor@redhat.com>
32793 PR translation/80280
32794 * builtins.c (expand_builtin_object_size): Add missing quoting to
32795 %D and like directives.
32796 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32797 (hsa_type_for_tree_type): Same.
32798 (verify_function_arguments): Same.
32799 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32800 * varasm.c (get_section): Same.
32803 2017-05-04 Martin Sebor <msebor@redhat.com>
32805 PR translation/80280
32806 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32808 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32810 * config/aarch64/aarch64.c (generic_addrcost_table):
32811 Change HI/TI mode setting.
32813 2017-05-04 Martin Jambor <mjambor@suse.cz>
32815 PR tree-optimization/80622
32816 * tree-sra.c (comes_initialized_p): New function.
32817 (build_accesses_from_assign): Only set write lazily when
32818 comes_initialized_p is false.
32819 (analyze_access_subtree): Use comes_initialized_p.
32820 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32821 instead of testing for PARM_DECL.
32823 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32825 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32826 constraint on operand 0 to allow more general addressing modes.
32827 Adjust output template.
32828 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32830 * config/aarch64/aarch64-protos.h
32831 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32832 * config/aarch64/constraints.md (Dp): New address constraint.
32833 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32836 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32838 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32839 update use of estimate_ipcp_clone_size_and_time.
32840 (estimate_local_effects): Update use of
32841 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32842 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32843 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32844 Return nonspecialized time.
32846 2017-05-04 Richard Biener <rguenther@suse.de>
32848 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32849 for the last VUSE which def dominates the PHI. Directly call
32851 (get_continuation_for_phi_1): Remove.
32853 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32855 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32856 to explain the use of truncating division. Cap the number of
32857 iterations to the maximum given by nb_iterations_upper_bound,
32860 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32862 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32863 * configure: Regenerate.
32864 * config.in: Regenerate.
32865 * config/i386/driver-mingw32.c: new file.
32866 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32867 * config.host: Link driver-mingw32.o on MinGW host.
32868 * doc/install.texi: Document new --enable-mingw-wildcard configure
32871 2017-05-04 Marek Polacek <polacek@redhat.com>
32873 PR tree-optimization/80612
32874 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32876 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32877 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32879 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32880 (movt splitter): Likewise.
32881 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32882 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32883 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32884 block for Thumb-1 with MOVT.
32885 (thumb2_legitimate_address_p): Move code block ...
32886 (can_avoid_literal_pool_for_label_p): ... into this new function.
32887 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32889 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32890 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32891 "M-profile targets with the MOVT instruction".
32893 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32895 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32896 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32897 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32899 2017-05-04 Martin Liska <mliska@suse.cz>
32901 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32902 variable cond_code.
32904 2017-05-04 Richard Biener <rguenther@suse.de>
32906 * tree.c (array_at_struct_end_p): Handle arrays at struct
32907 end with flexarrays more conservatively. Refactor and treat
32908 arrays of arrays or aggregates more strict. Fix
32909 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32910 * tree.c (array_at_struct_end_p): Adjust prototype.
32911 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32912 * gimple-fold.c (get_range_strlen): Likewise.
32913 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32915 2017-05-04 Richard Biener <rguenther@suse.de>
32917 PR tree-optimization/31130
32918 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32920 (supports_overflow_infinity): Likewise.
32921 (is_negative_overflow_infinity): Likewise.
32922 (is_positive_overflow_infinity): Likewise.
32923 (is_overflow_infinity): Likewise.
32924 (stmt_overflow_infinity): Likewise.
32925 (overflow_infinity_range_p): Likewise.
32926 (usable_range_p): Remove as always returning true.
32927 (make_overflow_infinity): Remove.
32928 (negative_overflow_infinity): Likewise.
32929 (positive_overflow_infinity): Likewise.
32930 (avoid_overflow_infinity): Likewise.
32931 (set_value_range): Adjust accordingly.
32932 (set_value_range_to_nonnegative): Likewise, remove now unused
32933 overflow_infinity arg.
32934 (vrp_operand_equal_p): Adjust.
32935 (update_value_range): Likewise.
32936 (range_int_cst_singleton_p): Likewise.
32937 (operand_less_p): Likewise.
32938 (compare_values_warnv): Likewise.
32939 (extract_range_for_var_from_comparison_expr): Likewise.
32940 (vrp_int_const_binop): Likewise.
32941 (zero_nonzero_bits_from_vr): Likewise.
32942 (extract_range_from_multiplicative_op_1): Likewise.
32943 (extract_range_from_binary_expr_1): Likewise.
32944 (extract_range_from_unary_expr): Likewise.
32945 (extract_range_from_comparison): Likewise.
32946 (extract_range_basic): Likewise.
32947 (adjust_range_with_scev): Likewise.
32948 (compare_ranges): Likewise.
32949 (compare_range_with_value): Likewise.
32950 (dump_value_range): Likewise.
32951 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32953 (simplify_cond_using_ranges): Adjust.
32955 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32957 * brig-builtins.def: Added a builtin for class_f64.
32958 * builtin-types.def: Added a builtin type needed by class_f64.
32960 2017-05-03 Jason Merrill <jason@redhat.com>
32962 * timevar.def: Add TV_CONSTEXPR.
32964 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32966 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32968 2017-05-03 Martin Jambor <mjambor@suse.cz>
32970 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32971 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32972 * ipa-cp.c (ipcp_propagate_stage): Do not call
32973 ipa_update_after_lto_read.
32974 * ipa-inline.c (ipa_inline): Likewise.
32976 2017-05-03 Martin Jambor <mjambor@suse.cz>
32978 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32979 tag. Added a default constructor and a destructor.
32980 (ipa_edge_args_sum_t): New class;
32981 (ipa_edge_args_sum): Declare.
32982 (ipa_edge_args_vector): Remove declaration.
32983 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32984 (ipa_free_edge_args_substructures): Remove declaration.
32985 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32986 (ipa_edge_args_info_available_for_edge_p): Likewise.
32987 * ipa-prop.c (ipa_edge_args_vector): Removed.
32988 (edge_removal_hook_holder): Likewise.
32989 (edge_duplication_hook_holder): Likewise.
32990 (ipa_edge_args_sum): New variable.
32991 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32992 ipa_edge_args_vector.
32993 (ipa_free_edge_args_substructures): Likewise.
32994 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32995 ipa_edge_args_vector.
32996 (ipa_edge_removal_hook): Turned into method
32997 ipa_edge_args_sum_t::remove.
32998 (ipa_edge_duplication_hook): Turned into method
32999 ipa_edge_args_sum_t::duplicate.
33000 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
33001 registering edge hooks.
33002 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
33003 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
33004 ipa_edge_args_sum instead of ipa_edge_args_vector.
33005 * ipa-profile.c (ipa_profile): Likewise.
33007 2017-05-03 Martin Jambor <mjambor@suse.cz>
33009 * symbol-summary.h (function_summary): New method exists.
33010 (function_summary::symtab_removal): Deallocate through release.
33011 (call_summary): New class.
33012 (gt_ggc_mx): New overload.
33013 (gt_pch_nx): Likewise.
33014 (gt_pch_nx): Likewise.
33016 2017-05-03 Jeff Law <law@redhat.com>
33018 PR tree-optimization/78496
33019 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33020 from simplify_cond_using_ranges. Split off code to walk
33021 backwards through casts into ...
33022 (simplify_cond_using_ranges_2): New function.
33023 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33024 (execute_vrp): After identifying jump threads, call
33025 simplify_cond_using_ranges_2.
33027 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33030 * ipa-inline.h (inline_summary): Add ctor.
33031 (create_ggc): Do not use ggc_cleared_alloc.
33033 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33034 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33036 * gcc.c (handle_braces): Support escaping in switch matching
33038 * doc/invoke.texi (Spec Files): Document it.
33039 Remove superfluous @code markup in items.
33041 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33043 * diagnostic-show-locus.c (struct column_range): New struct.
33044 (get_affected_columns): New function.
33045 (get_printed_columns): New function.
33046 (struct correction): New struct.
33047 (correction::ensure_capacity): New function.
33048 (correction::ensure_terminated): New function.
33049 (struct line_corrections): New struct.
33050 (line_corrections::~line_corrections): New dtor.
33051 (line_corrections::add_hint): New function.
33052 (layout::print_trailing_fixits): Reimplement in terms of the new
33054 (selftest::test_overlapped_fixit_printing): New function.
33055 (selftest::diagnostic_show_locus_c_tests): Call it.
33057 2017-05-03 Nathan Sidwell <nathan@acm.org>
33059 Canonicalize canonical type hashing
33060 * tree.h (type_hash_canon_hash): Declare.
33061 * tree.c (type_hash_list, attribute_hash_list): Move into
33062 type_hash_canon_hash.
33063 (build_type_attribute_qual_variant): Break out hash code calc into
33064 type_hash_canon_hash.
33065 (type_hash_canon_hash): New. Generic type hash computation.
33066 (build_range_type_1, build_array_type_1, build_function_type,
33067 build_method_type_directly, build_offset_type, build_complex_type,
33068 make_vector_type): Call it.
33070 2017-05-03 Richard Biener <rguenther@suse.de>
33072 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33073 When all DRs have unknown misaligned do not always peel
33074 when there is a store but apply the same costing model as if
33075 there were only loads.
33077 2017-05-03 Richard Biener <rguenther@suse.de>
33080 PR tree-optimization/80492
33081 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33082 compare_base_decls returning dont-know properly.
33084 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33086 * config/arm/iterators.md (CCSI): New mode iterator.
33087 (arch): New mode attribute.
33088 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33089 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33090 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33091 code iterator for success result mode.
33092 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33093 the corresponding new insn generators.
33095 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33098 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33099 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33101 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33103 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33104 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33105 (DDR_A): Wrap DDR argument in brackets.
33106 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33107 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33108 (DDR_REVERSED_P): Likewise.
33110 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33112 PR tree-optimization/79472
33113 * tree-switch-conversion.c (struct switch_conv_info): Add
33114 contiguous_range and default_case_nonstandard fields.
33115 (collect_switch_conv_info): Compute contiguous_range and
33116 default_case_nonstandard fields, don't clear final_bb if
33117 contiguous_range and only the default case doesn't have the required
33119 (check_all_empty_except_final): Set default_case_nonstandard instead
33120 of failing if contiguous_range and the default case doesn't have empty
33122 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33123 and only the default case doesn't have the required constants. Skip
33125 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33126 if default_case_nonstandard.
33127 (build_constructors): Build constant 1 just once. Assert that default
33128 values aren't inserted in between cases if contiguous_range. Skip
33130 (build_arrays): Skip virtual phis.
33131 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33132 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33133 Handle virtual phis.
33134 (gen_inbound_check): Handle default_case_nonstandard case.
33135 (process_switch): Adjust check_final_bb caller. Call
33136 gather_default_values with the first non-default case instead of
33137 default case if default_case_nonstandard.
33139 2017-05-02 Nathan Sidwell <nathan@acm.org>
33141 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33142 check. Fix formatting.
33144 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33146 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33147 errors when comparing specialized and unspecialized times.
33149 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33151 * diagnostic-show-locus.c
33152 (layout::should_print_annotation_line_p): Make private.
33153 (layout::print_annotation_line): Make private.
33154 (layout::annotation_line_showed_range_p): Make private.
33155 (layout::show_ruler): Make private.
33156 (layout::print_source_line): Make private. Pass in line and
33157 line_width, rather than calling location_get_source_line. Drop
33159 (layout::print_leading_fixits): New method.
33160 (layout::print_any_fixits): Rename to...
33161 (layout::print_trailing_fixits): ...this, and make private.
33162 Don't print newline fixits.
33163 (diagnostic_show_locus): Move logic for printing one row into...
33164 (layout::print_line): ...this new function. Move the
33165 location_get_source_line call and error-handling from
33166 print_source_line to here. Call print_leading_fixits, and rename
33167 print_any_fixits to print_trailing_fixits.
33168 (selftest::test_fixit_insert_containing_newline): Update now that
33169 newlines are partially supported.
33170 (selftest::test_fixit_insert_containing_newline_2): New test.
33171 (selftest::test_fixit_replace_containing_newline): Update comments.
33172 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33173 * edit-context.c (class added_line): New class.
33174 (class edited_line): Describe newline handling in comment.
33175 (edited_line::actually_edited_p): New method.
33176 (edited_line::print_content): Delete redundant decl.
33177 (edited_line::m_predecessors): New field.
33178 (edited_file::print_content): Call edited_line::print_content.
33179 (edited_file::print_diff): Update to support newlines.
33180 (edited_file::print_diff_hunk): Likewise.
33181 (edited_file::print_run_of_changed_lines): New function.
33182 (edited_file::print_diff_line): Convert to...
33183 (print_diff_line): ...this.
33184 (edited_file::get_effective_line_count): New function.
33185 (edited_line::edited_line): Initialize new field m_predecessors.
33186 (edited_line::~edited_line): Clean up m_predecessors.
33187 (edited_line::apply_fixit): Handle newlines.
33188 (edited_line::get_effective_line_count): New function.
33189 (edited_line::print_content): New function.
33190 (edited_line::print_diff_lines): New function.
33191 (selftest::test_applying_fixits_insert_containing_newline): New
33193 (selftest::test_applying_fixits_replace_containing_newline): New
33195 (selftest::insert_line): New function.
33196 (selftest::test_applying_fixits_multiple_lines): Add example of
33198 (selftest::edit_context_c_tests): Call the new tests.
33200 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33202 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33203 parameter cand. Update dump information.
33204 (get_computation_cost): Update uses.
33206 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33208 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33209 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33210 (get_computation_at, rewrite_use_address): Update use of
33211 get_computation_aff.
33213 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33215 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33216 (get_computation): Delete.
33217 (get_computation_cost): Implement like get_computation_cost_at.
33218 Use get_computation_at.
33219 (get_computation_cost_at): Delete.
33220 (rewrite_use_nonlinear_expr): Use get_computation_at.
33221 (rewrite_use_compare, remove_unused_ivs): Ditto.
33223 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33225 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33227 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33229 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33230 (ivopts_global_cost_for_size): Rename parameter and update uses.
33231 (iv_ca_recount_cost): Update uses.
33232 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33233 candidates seperately in n_invs and n_cands.
33234 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33236 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33238 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33239 (find_inv_vars_cb): New.
33240 (find_depends): Renamed to ...
33241 (find_inv_vars): ... this.
33242 (add_candidate_1, force_var_cost): Call find_inv_vars.
33243 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33245 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33247 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33248 inv_vars. Add inv_exprs.
33249 (struct iv_cand): Rename depends_on to inv_vars.
33250 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33251 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33252 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33253 (dump_cand): Dump inv_vars.
33254 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33255 (record_invariant, find_depends, add_candidate_1): Ditto.
33256 (set_group_iv_cost, force_var_cost): Ditto.
33257 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33258 (get_computation_cost_at, get_computation_cost): Ditto.
33259 (determine_group_iv_cost_generic): Ditto.
33260 (determine_group_iv_cost_address): Ditto.
33261 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33262 (determine_group_iv_costs): Ditto.
33263 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33264 (iv_ca_set_remove_invariants): Renamed to ...
33265 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33266 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33267 (iv_ca_set_add_invariants): Renamed to ...
33268 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33269 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33270 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33271 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33272 (create_new_ivs): Remove useless dump.
33274 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33276 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33278 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33279 (iv_ca_set_no_cp, create_new_iv): Ditto.
33281 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33283 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33285 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33287 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33288 function tree_check2.
33290 2017-05-02 Martin Liska <mliska@suse.cz>
33292 * doc/gcov.texi: Add missing preposition.
33293 * gcov.c (function_info::function_info): Properly fill up
33294 all member variables.
33296 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33298 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33300 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33302 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33304 2017-05-02 Martin Liska <mliska@suse.cz>
33307 * lto-streamer-in.c (lto_read_tree_1): Remove
33308 LTO_STREAMER_DEBUG.
33309 * lto-streamer.c (struct tree_hash_entry): Likewise.
33310 (struct tree_entry_hasher): Likewise.
33311 (tree_entry_hasher::hash): Likewise.
33312 (tree_entry_hasher::equal): Likewise.
33313 (lto_streamer_init): Likewise.
33314 (lto_orig_address_map): Likewise.
33315 (lto_orig_address_get): Likewise.
33316 (lto_orig_address_remove): Likewise.
33317 * lto-streamer.h: Likewise.
33318 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33319 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33321 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33323 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33324 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33325 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33326 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33327 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33328 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33329 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33330 (mm_maskz_sub_ss): New intrinsics.
33331 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33332 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33333 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33334 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33335 (__builtin_ia32_subss_mask_round): New builtins.
33336 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33337 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33338 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33340 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33341 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33343 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33346 2017-05-02 Martin Jambor <mjambor@suse.cz>
33348 PR tree-optimization/78687
33349 * tree-sra.c (access): New field parent.
33350 (process_subtree_disqualification): New function.
33351 (disqualify_candidate): Call it.
33352 (build_accesses_from_assign): Reset write flag if creating an
33354 (build_access_subtree): Fill in parent field and also prpagate
33355 down grp_write flag.
33356 (create_artificial_child_access): New parameter set_grp_write, set
33357 grp_write to its value.
33358 (propagate_subaccesses_across_link): Also propagate grp_write flag
33360 (propagate_all_subaccesses): Push the closest parent back to work
33361 queue if add_access_to_work_queue returned true.
33363 2017-05-02 Richard Biener <rguenther@suse.de>
33365 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33366 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33367 -fstrict-overflow documentation.
33368 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33369 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33370 flag_strict_overflow.
33371 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33372 * lto-opts.c (lto_write_options): Do not stream it.
33373 * lto-wrapper.c (merge_and_complain): Do not handle it.
33374 * opts.c (default_options_table): Do not set -fstrict-overflow.
33375 (finish_options): Likewise do not clear it when sanitizing.
33376 * simplify-rtx.c (simplify_const_relational_operation): Do not
33377 test flag_strict_overflow.
33379 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33381 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33382 using enabled attribute.
33383 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33384 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33385 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33386 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33387 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33388 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33389 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33390 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33391 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33392 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33394 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33396 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33398 2017-05-02 Richard Biener <rguenther@suse.de>
33400 PR tree-optimization/80591
33402 2017-04-10 Richard Biener <rguenther@suse.de>
33404 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33407 2017-05-02 Richard Biener <rguenther@suse.de>
33409 PR tree-optimization/80549
33410 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33411 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33412 headers if they do not have a preheader.
33414 2017-05-02 Martin Liska <mliska@suse.cz>
33417 * common.opt: Fix typo.
33418 * doc/invoke.texi: Likewise.
33420 2017-05-01 Jan Beulich <jbeulich@suse.com>
33422 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33423 swapping, add (x,x,m,x,n) alternative.
33425 2017-05-01 Nathan Sidwell <nathan@acm.org>
33427 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33428 unnecessary unadjusted_alignment check.
33430 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33433 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33435 * gimplify.c (gimplify_cilk_detach): New function.
33436 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33437 * tree-core.h: Document EXPR_CILK_SPAWN.
33438 * tree.h (EXPR_CILK_SPAWN): Define.
33440 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33442 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33443 to use new fixit_hint representation, using the "replace" logic.
33444 (get_line_span_for_fixit_hint): Likewise.
33445 (layout::print_any_fixits): Likewise.
33446 (selftest::test_one_liner_many_fixits): Rename to...
33447 (selftest::test_one_liner_many_fixits_1): ...this, and update
33448 comment and expected output to reflect that the multiple fix-it
33449 hints are now consolidated into one insertion.
33450 (selftest::test_one_liner_many_fixits_2): New test.
33451 (selftest::test_diagnostic_show_locus_one_liner): Update for
33453 (selftest::test_fixit_consolidation): Update for fix-it API
33455 * diagnostic.c (print_parseable_fixits): Likewise.
33456 * edit-context.c (edited_line::m_line_events): Convert from
33457 auto_vec <line_event *> to auto_vec <line_event>.
33458 (class line_event): Convert from abstract base class to a concrete
33459 class, taking over the role of replace_event.
33460 (class insert_event): Delete.
33461 (class replace_event): Rename to class line_event. Convert to
33463 (edit_context::add_fixits): Reimplement.
33464 (edit_context::apply_insert): Delete.
33465 (edit_context::apply_replace): Rename to...
33466 (edit_context::apply_fixit): ...this. Convert to half-open range.
33467 (edited_file::apply_insert): Delete.
33468 (edited_file::apply_replace): Rename to...
33469 (edited_file::apply_fixit): ...this.
33470 (edited_line::~edited_line): Drop deletion of events.
33471 (edited_line::apply_insert): Delete.
33472 (edited_line::apply_replace): Rename to...
33473 (edited_line::apply_fixit): ...this. Convert to half-open range.
33474 Update for change to type of m_line_events.
33475 * edit-context.h (edit_context::apply_insert): Delete.
33476 (edit_context::apply_replace): Rename to...
33477 (edit_context::apply_fixit): ...this.
33479 2017-05-01 Martin Sebor <msebor@redhat.com>
33481 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33484 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33487 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33488 __get_cpuid_max returns 0.
33489 (__get_cpuid_count): Ditto.
33491 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33493 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33494 replacement expression is another instance of one of its arguments.
33496 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33499 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33500 check for stack push/pop autoinc.
33501 * config/i386/i386.c (ix86_agi_dependent): Return false
33502 if the only reason why modified_in_p returned true is that
33503 addr is SP based and set_insn is a push or pop.
33505 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33507 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33510 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33513 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33514 (account_size_time): Use two predicates - exec_pred and
33516 (evaluate_conditions_for_known_args): Compute both clause and
33518 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33519 (inline_summary_t::duplicate): Update.
33520 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33522 (compute_inline_parameters): Likewise.
33523 (estimate_edge_size_and_time): Update caluclation of time.
33524 (estimate_node_size_and_time): Compute both time and nonspecialized
33526 (estimate_ipcp_clone_size_and_time): Update.
33527 (inline_merge_summary): Update.
33528 (do_estimate_edge_time): Update.
33529 (do_estimate_edge_size): Update.
33530 (do_estimate_edge_hints): Update.
33531 (inline_read_section, inline_write_summary): Stream both new predicates.
33532 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33534 (compute_inlined_call_time): Cleanup.
33535 (big_speedup_p): Update.
33536 (edge_badness): Update.
33537 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33538 (size_time_entry): Replace predicate by exec_predicate and
33539 nonconst_predicate.
33540 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33541 (estimate_edge_time): Return also nonspec_time.
33542 (reset_edge_growth_cache): Update.
33544 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33546 PR rtl-optimization/80491
33547 * ifcvt.c (noce_process_if_block): When looking for x setter
33548 with missing else_bb, don't check only the insn right before
33549 cond_earliest, but look for the last insn that x is modified in
33550 within the same bb.
33552 PR rtl-optimization/80491
33553 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33555 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33557 PR tree-optimization/80487
33558 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33560 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33562 PR tree-optimization/79697
33563 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33564 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33565 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33567 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33568 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33570 2017-04-28 Martin Sebor <msebor@redhat.com>
33572 PR tree-optimization/80523
33573 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33574 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33576 (maybe_warn, format_directive, parse_directive): Use new functions.
33577 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33579 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33581 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33583 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33585 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33586 target_header_dir): Set correctly.
33587 * configure: Regenerated.
33588 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33589 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33590 instead of SYSTEM_HEADER_DIR.
33592 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33594 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33595 (estimate_local_effects): Likewise.
33596 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33597 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33598 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33599 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33600 do_estimate_edge_time, estimate_edge_time): Likewise.
33601 * ipa-inline-analysis.c (estimate_node_size_and_time,
33602 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33603 (estimate_time_after_inlining): Remove.
33605 2017-04-28 Martin Liska <mliska@suse.cz>
33607 * doc/gcov.texi: Enhance documentation of gcov.
33609 2017-04-28 Martin Liska <mliska@suse.cz>
33611 * doc/gcov.texi: Sort options in alphabetic order.
33612 * doc/gcov-dump.texi: Likewise.
33613 * doc/gcov-tool.texi: Likewise.
33614 * gcov.c (print_usage): Likewise.
33615 * gcov-dump.c (print_usage): Likewise.
33616 * gcov-tool.c (print_merge_usage_message): Likewise.
33617 (print_rewrite_usage_message): Likewise.
33618 (print_overlap_usage_message): Likewise.
33620 2017-04-28 Martin Liska <mliska@suse.cz>
33622 PR gcov-profile/53915
33623 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33625 2017-04-28 Martin Liska <mliska@suse.cz>
33627 PR gcov-profile/79891
33628 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33629 is marked by compiler as living on a line.
33630 (get_cycles_count): Remove usage of the union.
33631 (output_intermediate_file): Likewise.
33632 (find_source): Fix GNU coding style.
33633 (accumulate_line_counts): Remove old non-all block mode.
33634 (output_lines): Remove usage of the union.
33635 * profile.c (output_location): Include all BBs, even if
33636 belonging to a same line (and file) as a previous BB.
33638 2017-04-28 Martin Liska <mliska@suse.cz>
33640 * gcov.c (process_args): Handle new argument 'w'.
33641 (read_graph_file): Assign ID to BBs.
33642 (output_branch_count): Display BB # if verbose flag is set.
33643 (output_lines): Likewise for arcs.
33644 (print_usage): Add '--verbose' option help.
33645 * doc/gcov.texi: Document --verbose (-w) option.
33647 2017-04-28 Martin Liska <mliska@suse.cz>
33649 * gcov.c (struct block_location_info): New struct.
33650 (process_file): Fill up the new structure.
33651 (read_graph_file): Replace usage of encoding by the newly added
33653 (add_line_counts): Likewise.
33654 (accumulate_line_counts): Remove usage of the union.
33655 (function_info::function_info): New function.
33656 (function_info::~function_info): Likewise.
33657 (process_file): Call delete instead of release_function.
33658 (release_function): Release the function.
33659 (release_structures): Call delete instead of release_function.
33660 (solve_flow_graph): Replace usage of num_blocks.
33661 (find_exception_blocks): Likewise.
33662 (output_lines): Fix GNU coding style.
33664 2017-04-28 Martin Liska <mliska@suse.cz>
33667 * coverage.c (coverage_remove_note_file): New function.
33668 * coverage.h: Declare the function.
33669 * toplev.c (finalize): Clean if an error has been seen.
33671 2017-04-28 Martin Liska <mliska@suse.cz>
33673 PR gcov-profile/80031
33674 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33675 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33676 * gcov.c (read_graph_file): Read just number of blocks.
33677 * profile.c (branch_prob): Do not stream 0 flags per a basic
33680 2017-04-28 Martin Liska <mliska@suse.cz>
33682 * gcov-dump.c (tag_*): Add new argument to declarations.
33683 (dump_gcov_file): Likewise.
33684 (tag_blocks): Add and use new argument depth.
33685 (tag_arcs): Likewise.
33686 (tag_lines): Likewise.
33687 (tag_counters): Likewise.
33688 (tag_summary): Likewise.
33689 (dump_working_sets): Use depth to do a proper indentation.
33691 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33694 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33695 * symtab.c (symtab_node::debug_symtab): Move definition here.
33697 2017-04-28 Richard Biener <rguenther@suse.de>
33699 * lto-streamer.h (LTO_major_version): Bump to 7.
33701 2017-04-28 Richard Biener <rguenther@suse.de>
33703 * tree-vrp.c (assert_info): New struct.
33704 (add_assert_info): New helper.
33705 (register_edge_assert_for_2): Refactor to add asserts to a vector
33707 (register_edge_assert_for_1): Likewise.
33708 (register_edge_assert_for): Likewise.
33709 (finish_register_edge_assert_for): New helper actually registering
33710 asserts where live on edge.
33711 (find_conditional_asserts): Adjust.
33712 (find_switch_asserts): Likewise.
33713 (evrp_dom_walker::try_find_new_range): Generalize.
33714 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33716 2017-04-27 Marek Polacek <polacek@redhat.com>
33719 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33720 arg10 and arg11 to itype.
33722 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33724 * doc/extend.texi (Object Size Checking): Improve grammar.
33726 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33729 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33730 that the logic for permitting reciprocal estimates matches that
33733 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33736 * tree.c (type_cache_hasher::equal): Only compare
33737 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33738 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33739 non-aggregate element types.
33740 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33741 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33743 2017-04-27 Richard Biener <rguenther@suse.de>
33745 PR middle-end/80533
33746 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33747 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33748 keeping a reference to a trailing array.
33750 2017-04-27 Richard Biener <rguenther@suse.de>
33752 PR middle-end/80539
33753 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33754 being in loop-closed SSA form conservatively.
33755 (chrec_fold_multiply_poly_poly): Likewise.
33757 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33759 PR middle-end/79665
33760 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33761 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33763 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33766 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33767 (aarch64_function_arg_alignment): Return unsigned int again, but still
33768 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33769 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33770 Don't emit -Wpsabi note.
33771 (aarch64_function_arg_boundary): Likewise.
33772 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33775 2017-04-26 Nathan Sidwell <nathan@acm.org>
33777 * tree.h (crc32_unsigned_n): Declare.
33778 (crc32_unsigned, crc32_unsigned): Make inline.
33779 * tree.c (crc32_unsigned_bits): Replace with ...
33780 (crc32_unsigned_n): ... this.
33781 (crc32_unsigned, crc32_byte): Remove.
33782 (crc32_string): Remove unnecessary braces.
33784 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33786 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33787 * ipa-inline-analysis.c (MAX_TIME): Remove.
33788 (account_size_time): Use sreal for time.
33789 (dump_inline_summary): Update.
33790 (estimate_function_body_sizes): Update.
33791 (estimate_edge_size_and_time): Update.
33792 (estimate_calls_size_and_time): Update.
33793 (estimate_node_size_and_time): Update.
33794 (inline_merge_summary): Update.
33795 (inline_update_overall_summary): Update.
33796 (estimate_time_after_inlining): Update.
33797 (inline_read_section): Update.
33798 (inline_write_summary): Update.
33799 * ipa-inline.c (compute_uninlined_call_time): Update.
33800 (compute_inlined_call_time): Update.
33801 (recursive_inlining): Update.
33802 (inline_small_functions): Update.
33803 (dump_overall_stats): Update.
33804 * ipa-inline.h: Include sreal.h.
33805 (size_time_entry): Turn time to sreal.
33806 (inline_summary): Turn self_time nad time to sreal.
33808 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33810 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33812 (sreal::stream_out, sreal::stream_in): New.
33813 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33815 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33817 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33820 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33823 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33825 (dimode_scalar_chain::compute_convert_gain): Ditto.
33826 (dimode_scalar_chain::make_vector_copies): Ditto.
33827 (dimode_scalar_chain::convert_reg): Ditto.
33828 (dimode_scalar_chain::convert_insn): Ditto.
33829 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33830 (VI248_AVX512BW_1): New mode iterator.
33831 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33832 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33835 2017-04-25 Martin Sebor <msebor@redhat.com>
33837 PR tree-optimization/80497
33838 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33839 constants are representable in HOST_WIDE_INT.
33840 (parse_directive): Ditto.
33842 2017-04-25 Martin Sebor <msebor@redhat.com>
33845 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33846 (new_zero_array): Adjust signature.
33847 (dom_info::dom_init): Used unsigned rather that size_t.
33848 (dom_info::dom_info): Same.
33850 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33851 Jakub Jelinek <jakub@redhat.com>
33854 * config/arm/arm.c: Include gimple.h.
33855 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33856 returns negative, increment ncrn only if it returned positive.
33857 (arm_needs_doubleword_align): Return int instead of bool,
33858 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33859 members, but if there is any such non-FIELD_DECL
33860 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33861 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33862 returns negative, increment nregs only if it returned positive.
33863 (arm_setup_incoming_varargs): Likewise.
33864 (arm_function_arg_boundary): Emit -Wpsabi note if
33865 arm_needs_doubleword_align returns negative, return
33866 DOUBLEWORD_ALIGNMENT only if it returned positive.
33868 2017-04-25 Marek Polacek <polacek@redhat.com>
33871 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33872 first argument to type.
33874 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33877 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33878 type checks to test for compatibility instead of equality.
33880 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33881 Jakub Jelinek <jakub@redhat.com>
33884 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33886 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33887 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33888 the alignment computation, but return their maximum in warn_alignment.
33889 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33890 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33892 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33893 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33896 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33898 * config/arc/simdext.md (dmpyh): Fix typo.
33900 2017-04-25 Richard Biener <rguenther@suse.de>
33902 PR tree-optimization/80492
33903 * alias.c (compare_base_decls): Handle registers with asm
33904 specification conservatively.
33905 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33906 compare_base_decls returning dont-know properly.
33908 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33910 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33911 (legitimate_offset_address_p): New function.
33912 (arc_legitimate_address_p): Use above function.
33914 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33916 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33918 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33920 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33921 ACCH registers whenever they are available.
33923 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33925 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33926 double regs fix when not used.
33928 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33930 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33932 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33933 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33935 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33937 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33938 when emitting short ADD instructions.
33940 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33942 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33944 (cmpsi_cc_c_insn): Likewise.
33945 (cbranchsi4_scratch): Compute proper instruction length using
33946 compact_hreg_operand.
33947 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33949 2017-04-25 Richard Biener <rguenther@suse.de>
33951 PR middle-end/80509
33952 * passes.c (pass_manager::pass_manager): Initialize
33953 m_name_to_pass_map.
33955 2017-04-25 Richard Biener <rguenther@suse.de>
33957 PR tree-optimization/79201
33958 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33960 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33963 * config/s390/vector.md: Split MEM->GPR vector moves for
33964 non-s_operand addresses.
33966 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33969 * config/s390/predicates.md (reload_const_wide_int_operand): New
33971 * config/s390/s390.md ("movti"): Remove d/P alternative.
33972 ("movti_bigconst"): New pattern definition.
33974 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33977 * s390-protos.h (s390_expand_cs_hqi): Removed.
33978 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33979 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33980 modes as well as CCZ1mode and CCZmode.
33981 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33982 signature of s390_emit_compare_and_swap.
33983 (s390_expand_cs_hqi): Likewise, make static.
33984 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33985 compare-and-swap, in some cases.
33986 (s390_expand_cs): Wrapper function.
33987 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33989 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33990 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33991 patterns for small and large integers. Forbid symref memory operands.
33992 Move expander to s390.c. Require cc register.
33993 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33994 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33995 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33996 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33997 symref memory operands. Remove CC mode and call s390_match_ccmode
33999 ("atomic_exchange<mode>"): Allow and implement all integer modes.
34001 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34003 * config/s390/s390.md (define_peephole2): New peephole to help
34004 combining the load-and-test pattern with volatile memory.
34006 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34008 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34009 with CCZmode for TARGET_Z196.
34011 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34013 PR rtl-optimization/80501
34014 * combine.c (make_compound_operation_int): Set subreg_code to SET
34015 even for AND with mask of the sign bit of mode.
34017 PR rtl-optimization/80500
34018 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34019 sum's initial value.
34021 2017-04-25 Julian Brown <julian@codesourcery.com>
34022 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34024 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34026 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34028 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34030 2017-04-25 Julian Brown <julian@codesourcery.com>
34031 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34033 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34034 (thunderx2t99_sha): New Reservation.
34036 2017-04-25 Julian Brown <julian@codesourcery.com>
34037 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34039 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34040 type for 1-element load.
34042 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34044 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34046 2017-04-24 Martin Jambor <mjambor@suse.cz>
34048 PR tree-optimization/80293
34049 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34050 char arrays not totally scalarizable if it is false.
34051 (analyze_all_variable_accesses): Pass correct value in the new
34052 parameter. Add a statistics counter.
34054 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34056 PR middle-end/79931
34057 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34059 2017-04-24 Richard Biener <rguenther@suse.de>
34061 PR tree-optimization/80494
34062 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34063 out for complex types.
34065 2017-04-24 Richard Biener <rguenther@suse.de>
34067 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34068 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34069 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34070 (DFS): Adjust and never fail.
34071 (sccvn_dom_walker::fail): Remove.
34072 (sccvn_dom_walker::before_dom_children): Adjust.
34073 (run_scc_vn): Likewise and never fail.
34074 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34075 (pass_fre::execute): Likewise.
34077 2017-04-24 Richard Biener <rguenther@suse.de>
34079 PR tree-optimization/79725
34080 * tree-ssa-sink.c (statement_sink_location): Return whether
34081 failure reason was zero uses. Move that check later.
34082 (sink_code_in_bb): Deal with zero uses by removing the stmt
34085 2017-04-24 Richard Biener <rguenther@suse.de>
34088 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34089 pointer-based references.
34091 2017-04-24 Richard Biener <rguenther@suse.de>
34094 * pass_manager.h (pass_manager::operator new): Remove.
34095 (pass_manager::operator delete): Likewise.
34096 * passes.c (pass_manager::operator new): Remove.
34097 (pass_manager::operator delete): Likewise.
34098 (pass_manager::pass_manager): Zero individual pass members.
34100 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34103 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34104 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34105 Check "XEXP (src, 1)" operand here.
34106 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34107 Check "XEXP (src, 1)" operand here.
34108 (dimode_scalar_chain::make_vector_copies): Detect count register
34109 of a shift instruction. Zero extend count register from QImode
34110 to DImode to satisfy vector shift pattern count operand predicate.
34111 Substitute vector shift count operand with a DImode copy.
34112 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34115 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34117 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34118 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34119 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34120 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34121 (UNSPEC_NOREX_MEM): Remove definition.
34123 2017-04-21 Richard Biener <rguenther@suse.de>
34125 PR tree-optimization/79547
34126 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34127 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34128 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34129 without any constraints.
34131 2017-04-21 Richard Biener <rguenther@suse.de>
34133 PR tree-optimization/78847
34134 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34136 2017-04-21 Richard Biener <rguenther@suse.de>
34138 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34139 (build_distinct_type_copy): Likewise.
34140 (build_variant_type_copy): Likewise.
34141 * tree.c (build_qualified_type): Pass down mem-stat info.
34142 (build_distinct_type_copy): Likewise.
34143 (build_variant_type_copy): Likewise.
34145 2017-04-21 Richard Biener <rguenther@suse.de>
34147 PR tree-optimization/80237
34148 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34150 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34151 for a simplified result.
34153 2016-04-21 Richard Biener <rguenther@suse.de>
34155 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34156 sth as strict as a simple_iv but a chrec without symbols and an
34157 operand defined in the loop we are peeling (and not some subloop).
34158 (propagate_constants_for_unrolling): Propagate all constants.
34160 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34163 * config/i386/i386.c (print_reg): Remove assert for disalowed
34164 regno values, call output_operand_lossage instead.
34166 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34169 * config/i386/constraints.md (Yc): New register constraint.
34170 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34171 Use Yc constraint for alternative 2 of operand 0. Remove
34172 preferred_for_speed attribute.
34174 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34176 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34177 lastprivate clauses in SIMT case.
34179 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34181 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34183 2017-04-20 Richard Biener <rguenther@suse.de>
34185 PR tree-optimization/57796
34186 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34187 as N scalar stores.
34188 (vect_model_load_cost): Cost gathers as N scalar loads.
34190 2017-04-20 Richard Biener <rguenther@suse.de>
34192 * ggc-page.c (ggc_allocated_p): Rename to ...
34193 (safe_lookup_page_table_entry): ... this and return the lookup
34195 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34197 2017-04-20 Richard Biener <rguenther@suse.de>
34199 PR tree-optimization/80453
34200 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34201 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34202 from the conditions.
34203 (vn_phi_eq): Pass them down.
34204 (vn_phi_lookup): Record them.
34205 (vn_phi_insert): Likewise.
34207 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34209 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34210 uninitialized variable warning to avoid buffer overrun.
34212 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34215 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34216 is suppressed for '{ 0 }' in C.
34218 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34220 * BASE-VER: Set to 8.0.0.
34222 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34224 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34225 priority .init_array and .fini_array section with SECTION_NOTYPE
34228 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34230 PR middle-end/80423
34231 * tree.h (build_array_type): Add typeless_storage default argument.
34232 * tree.c (type_cache_hasher::equal): Also compare
34233 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34234 (build_array_type): Add typeless_storage argument, set
34235 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34237 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34238 (build_array_type): Likewise. Add typeless_storage argument.
34240 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34241 Jakub Jelinek <jakub@redhat.com>
34243 PR tree-optimization/80426
34244 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34245 operation on symbolic operands, also compute the overflow for the
34246 invariant part when the operation degenerates into a negation.
34248 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34251 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34252 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34255 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34257 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34260 * config/avr/avr.c (tree.h): Include it.
34261 (cgraph.h): Include it.
34262 (avr_encode_section_info): Don't warn for uninitialized progmem
34263 variable if it's just an alias.
34265 2017-04-19 Richard Biener <rguenther@suse.de>
34268 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34269 when needed by AutoPGO.
34271 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34274 * doc/lto.texi: Remove an extra 'that'.
34276 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34278 PR rtl-optimization/80429
34279 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34280 are only used in debug insns.
34282 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34283 Vladimir Makarov <vmakarov@redhat.com>
34285 * config/sparc/predicates.md (input_operand): Add comment. Return
34286 true for any memory operand when LRA is in progress.
34287 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34289 2017-04-18 Jeff Law <law@redhat.com>
34292 * mips.md ({return,simple_return}_internal): Do not overwrite
34295 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34297 PR tree-optimization/80443
34298 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34299 instead of adding 1, subtract -1 and similarly instead of subtracting
34302 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34304 PR rtl-optimization/80357
34305 * haifa-sched.c (tmp_bitmap): New variable.
34306 (model_recompute): Handle duplicate use records.
34307 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34308 (free_global_sched_pressure_data): Free it.
34310 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34313 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34314 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34315 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34316 instead of SYSTEM_HEADER_DIR.
34318 2017-04-18 Jeff Law <law@redhat.com>
34320 PR middle-end/80422
34321 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34322 predecessors after walking up the insn chain.
34324 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34327 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34328 sizetype type into debug info.
34330 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34333 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34334 unneeded test for TARGET_UPPER_REGS_SF.
34335 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34337 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34340 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34341 instead of gsi_after_labels.
34343 2017-04-18 Jeff Law <law@redhat.com>
34345 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34349 2017-04-13 Jeff Law <law@redhat.com>
34350 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34351 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34353 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34356 * config/avr/avr.c (intl.h): Include it.
34357 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34359 2017-04-18 Martin Liska <mliska@suse.cz>
34361 PR gcov-profile/78783
34362 * gcov-tool.c (gcov_output_files): Validate that destination
34363 file is either removed by the tool or by a user.
34365 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34366 Guy Benyei <guybe@mellanox.com>
34368 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34369 block, and do not negate it, the stored id is already negative.
34371 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34373 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34375 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34378 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34379 masks of options that should be turned off if the VSX vector
34380 options are turned off.
34381 (OTHER_P8_VECTOR_MASKS): Likewise.
34382 (OTHER_VSX_VECTOR_MASKS): Likewise.
34383 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34384 rs6000_disable_incompatible_switches to validate no type switches
34386 (rs6000_incompatible_switch): New function to disallow turning on
34387 other vector options if -mno-vsx, -mno-power8-vector, or
34388 -mno-power9-vector are specified.
34390 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34392 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34394 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34396 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34397 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34398 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34399 (ARG_POINTER_CFA_OFFSET): Likewise.
34401 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34403 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34404 conditions to take advantage of various optimizations.
34406 2017-04-13 Jeff Law <law@redhat.com>
34408 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34409 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34410 (zero_extendsidi2_dext): Likewise.
34412 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34415 * fold-const.c (fold_ternary_loc): Revert
34416 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34419 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34421 PR rtl-optimization/80343
34422 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34423 new scratch pseudo.
34425 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34428 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34429 to ubsan_encode_value.
34431 2017-04-13 Jeff Law <law@redhat.com>
34433 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34434 appearing in DEBUG_INSNs.
34436 2017-04-13 Martin Liska <mliska@suse.cz>
34438 PR gcov-profile/80413
34439 * gcov-io.c (gcov_write_string): Copy to buffer just when
34440 allocated size is greater than zero.
34442 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34445 * dwarf2out.c (decls_for_scope): Ignore declarations of
34446 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34448 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34451 * ipa-visibility.c (non_local_p): Fix typos.
34452 (localize_node): When localizing symbol in same comdat group,
34453 dissolve the group only when we know external symbols are going
34455 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34457 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34459 PR tree-optimization/79390
34460 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34461 order does not result in usable sequence, retry with reversed operand
34467 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34468 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34469 op0 instead of fold_convert_loc (loc, type, arg0).
34471 2017-04-12 Jeff Law <law@redhat.com>
34473 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34474 has a delay slot in the generated code.
34476 * config/cris/cris.md (cris_preferred_reload_class): Return
34477 GENNONACR_REGS rather than GENERAL_REGS.
34479 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34482 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34483 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34484 signedness of the result type.
34486 2017-04-12 Richard Biener <rguenther@suse.de>
34487 Jeff Law <law@redhat.com>
34489 PR tree-optimization/80359
34490 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34491 trim stores to TARGET_MEM_REFs.
34493 2017-04-12 Richard Biener <rguenther@suse.de>
34495 PR tree-optimization/79390
34496 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34497 threading case even more.
34499 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34502 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34503 for quad_address_p for TImode, instead of just not indexed_address.
34505 2017-04-12 Richard Biener <rguenther@suse.de>
34506 Bernd Edlinger <bernd.edlinger@hotmail.de>
34508 PR middle-end/79671
34509 * alias.c (component_uses_parent_alias_set_from): Handle
34510 TYPE_TYPELESS_STORAGE.
34511 (get_alias_set): Likewise.
34512 * tree-core.h (tree_type_common): Add typeless_storage flag.
34513 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34514 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34515 for types containing members with TYPE_TYPELESS_STORAGE.
34516 (place_field): Likewise.
34517 (layout_type): Likewise for ARRAY_TYPE.
34518 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34519 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34520 TYPE_TYPELESS_STORAGE.
34521 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34523 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34526 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34527 first argument to type.
34529 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34533 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34534 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34535 (rs6000_expand_binop_builtin): Likewise.
34536 (rs6000_expand_ternop_builtin): Likewise; also add missing
34537 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34538 vshasigma built-ins.
34539 * doc/extend.texi: Document that vec_xxpermdi's third argument
34540 must be a constant.
34542 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34544 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34545 Use shift_const cost parameter when calculating gain of STV shifts.
34547 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34549 PR rtl-optimization/70478
34550 * lra-constraints.c (process_alt_operands): Check memory for
34551 disfavoring memory insn operand.
34553 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34555 PR middle-end/80100
34556 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34557 left shift in unsigned HOST_WIDE_INT type.
34559 PR rtl-optimization/80385
34560 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34561 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34564 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34565 if they have any depend clauses.
34567 2017-04-11 Martin Liska <mliska@suse.cz>
34570 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34571 * ipa-split.c (split_function): Create a local comdat symbol
34572 if caller is in a comdat group.
34574 2017-04-11 Martin Liska <mliska@suse.cz>
34577 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34580 2017-04-11 Martin Sebor <msebor@redhat.com>
34582 PR middle-end/80364
34583 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34584 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34586 (directive::set_width, directive::set_precision, format_character):
34588 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34591 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34594 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34595 conflict, set target->arch_name instead of target->cpu_name.
34597 2017-04-11 Richard Biener <rguenther@suse.de>
34599 PR tree-optimization/80374
34600 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34601 build_zero_cst, remove fold_convertible_p check again.
34603 2017-04-11 Martin Liska <mliska@suse.cz>
34606 * ubsan.c (instrument_object_size): Do not instrument register
34609 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34612 * config/i386/i386-builtin-types.def
34613 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34614 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34615 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34616 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34617 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34618 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34619 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34620 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34621 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34622 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34623 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34624 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34625 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34626 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34627 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34628 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34629 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34630 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34631 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34632 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34633 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34634 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34635 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34636 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34637 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34638 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34639 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34640 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34641 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34642 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34643 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34644 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34645 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34646 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34647 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34648 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34649 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34650 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34651 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34652 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34653 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34654 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34655 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34656 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34657 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34658 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34660 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34661 flag to second_arg_count, handle 4 argument function type _COUNT
34662 aliases, handle second_arg_count on second argument rather than last.
34664 2017-04-10 Jeff Law <law@redhat.com>
34666 PR tree-optimization/80374
34667 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34668 record anything if we can not convert integer_zero_node to the
34671 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34674 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34675 Enhance special handling given to the TARGET_P9_MINMAX option in
34676 relation to certain other options.
34678 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34680 PR tree-optimization/80153
34681 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34682 remove POINTER_PLUS_EXPR's base part directly, rather than through
34685 2017-04-10 Richard Biener <rguenther@suse.de>
34686 Bin Cheng <bin.cheng@arm.com>
34688 PR tree-optimization/80153
34689 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34690 the first element of pointer type aff_tree. Build result expr in
34692 (add_elt_to_tree): Convert to type unconditionally. Remove other
34693 fold_convert calls.
34694 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34695 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34697 2017-04-10 Richard Biener <rguenther@suse.de>
34699 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34702 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34704 PR rtl-optimization/70478
34705 * lra-constraints.c (curr_small_class_check): New.
34706 (update_and_check_small_class_inputs): New.
34707 (process_alt_operands): Update curr_small_class_check. Disfavor
34708 alternative insn memory operands. Check available regs for small
34711 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34714 * config/mips/mips.opt (-mvirt): Update description.
34715 * doc/invoke.texi (-mvirt): Likewise.
34717 2017-04-10 Richard Biener <rguenther@suse.de>
34719 PR middle-end/80362
34720 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34721 looking for NEGATE_EXPR in -A / -B to A / B folding.
34723 2017-04-10 Martin Liska <mliska@suse.cz>
34725 PR gcov-profile/80224
34726 * gcov.c (print_usage): Fix usage string.
34727 (get_gcov_intermediate_filename): Remove.
34728 (output_gcov_file): Use both for normal and intermediate format.
34729 (generate_results): Do not initialize special file for
34730 intermediate format.
34732 2017-04-10 Richard Biener <rguenther@suse.de>
34734 PR tree-optimization/80304
34735 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34738 2017-04-10 Nathan Sidwell <nathan@acm.org>
34741 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34742 (rs6000_init_builtins): Use it.
34744 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34746 * config/arm/arm.md (<mrc>): Add mode to SET source.
34747 (<mrrc>): Likewise.
34749 2017-04-10 Richard Biener <rguenther@suse.de>
34751 PR middle-end/80344
34752 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34754 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34757 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34758 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34759 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34760 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34761 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34762 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34763 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34764 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34765 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34766 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34767 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34768 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34769 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34770 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34771 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34772 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34773 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34774 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34775 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34776 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34777 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34778 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34779 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34781 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34783 PR rtl-optimization/70478
34784 * lra-constraints.c: Reverse the last patch.
34786 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34788 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34789 Add comment for WCHAR_T.
34791 2017-04-08 Martin Liska <mliska@suse.cz>
34794 2017-04-07 Martin Liska <mliska@suse.cz>
34797 * ipa-split.c (split_function): Add function part to a same comdat
34800 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34803 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34805 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34807 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34808 (vec_pairing): ...to this.
34809 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34810 (rs6000_sched_init): Adjust for name change.
34811 (struct rs6000_sched_context): Likewise.
34812 (rs6000_init_sched_context): Likewise.
34813 (rs6000_set_sched_context): Likewise.
34815 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34821 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34823 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34824 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34825 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34827 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34829 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34831 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34833 PR rtl-optimization/70703
34834 * ira-color.c (update_conflict_hard_regno_costs): Use
34835 int64_t instead of HOST_WIDE_INT.
34837 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34839 PR rtl-optimization/70478
34840 * lra-constraints.c (process_alt_operands): Disfavor alternative
34841 insn memory operands.
34843 2017-04-07 Jeff Law <law@redhat.com>
34845 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34846 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34848 2017-04-07 Martin Liska <mliska@suse.cz>
34851 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34852 Show error message instead of an ICE.
34854 2017-04-07 Martin Liska <mliska@suse.cz>
34857 * ipa-split.c (split_function): Add function part to a same comdat
34860 2017-04-07 Richard Biener <rguenther@suse.de>
34862 PR middle-end/80341
34863 * tree.c (get_unwidened): Also handle ! for_type case for
34865 * convert.c (do_narrow): Split out from ...
34866 (convert_to_integer_1): ... here. Do not pass final truncation
34867 type to get_unwidened for TRUNC_DIV_EXPR.
34869 2017-04-07 Richard Biener <rguenther@suse.de>
34871 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34873 (aff_combination_const): Adjust.
34874 (aff_combination_scale): Likewise.
34875 (aff_combination_add_elt): Likewise.
34876 (aff_combination_add_cst): Likewise.
34877 (aff_combination_convert): Likewise.
34878 (add_elt_to_tree): Likewise. Remove unused argument.
34879 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34881 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34883 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34885 * config/arm/arm.c (arm_default_short_enums): Use
34886 ARM_DEFAULT_SHORT_ENUMS.
34887 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34889 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34892 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34893 members with redundant out-of-class redeclaration.
34895 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34898 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34899 * config/i386/i386.md (*zero_extendsidi2):
34900 Add (?*x,*x) and (?*v,*v) alternatives.
34902 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34905 * config/i386/i386.c (ix86_expand_builtin)
34906 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34907 mode from insn data. Convert operands to insn operand mode.
34908 Copy operands that don't satisfy insn predicate to a register.
34910 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34912 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34915 2017-04-06 Richard Biener <rguenther@suse.de>
34917 PR tree-optimization/80334
34918 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34919 preserve alignment of accesses.
34921 2017-04-06 Richard Biener <rguenther@suse.de>
34923 PR tree-optimization/80262
34924 * tree-sra.c (build_ref_for_offset): Preserve address-space
34926 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34927 Drop useless address-space information on MEM_REF offsets.
34929 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34931 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34933 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34935 PR rtl-optimization/70703
34936 * ira-color.c (update_conflict_hard_regno_costs): Use
34937 HOST_WIDE_INT instead of long.
34939 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34942 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34943 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34945 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34946 for x86_64 target. Handle -m3dnowa option.
34948 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34950 PR rtl-optimization/70703
34951 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34952 (update_conflict_hard_regno_costs): Use long instead of unsigned
34953 arithmetic for cost calculation.
34955 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34956 Bernd Edlinger <bernd.edlinger@hotmail.de>
34959 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34962 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34965 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34966 ptr_mode with Pmode throughout.
34967 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34968 into probe_stack_range and use DImode.
34970 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34973 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34974 call_eh_return is true.
34976 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34978 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34979 Initialize last_match_fntype_index.
34981 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34984 * tree-nvr.c: Include internal-fn.h.
34985 (pass_return_slot::execute): Ignore internal calls without
34988 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34989 Richard Biener <rguenther@suse.de>
34992 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34993 captures used multiple times, except for the last use.
34994 * generic-match-head.c: Include gimplify.h.
34996 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34998 PR tree-optimization/79390
34999 * target.h (struct noce_if_info): Declare.
35000 * targhooks.h (default_noce_conversion_profitable_p): Declare.
35001 * target.def (noce_conversion_profitable_p): New target hook.
35002 * ifcvt.h (struct noce_if_info): New type, moved from ...
35003 * ifcvt.c (struct noce_if_info): ... here.
35004 (noce_conversion_profitable_p): Renamed to ...
35005 (default_noce_conversion_profitable_p): ... this. No longer
35007 (noce_try_store_flag_constants, noce_try_addcc,
35008 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35009 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35010 instead of noce_conversion_profitable_p.
35011 * config/i386/i386.c: Include ifcvt.h.
35012 (ix86_option_override_internal): Don't override
35013 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35014 (ix86_noce_conversion_profitable_p): New function.
35015 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35016 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35017 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35018 * doc/tm.texi: Regenerated.
35020 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35022 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35025 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35028 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35029 instructions for small multiply cores.
35031 2017-04-04 Jeff Law <law@redhat.com>
35033 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35035 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35036 that are not set by the loop over the elements.
35038 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35041 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35042 int mode, convert_modes it to mode as unsigned, otherwise use
35043 lowpart_subreg to mode rather than SImode.
35044 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35045 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35046 Use DImode instead of SImode for the shift count operand.
35047 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35050 2017-04-04 Richard Biener <rguenther@suse.de>
35052 PR middle-end/80281
35053 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35054 arithmetic done for the negate or the plus. Simplify.
35055 (A - (-B) -> A + B): Likewise.
35056 * fold-const.c (split_tree): Make sure to not negate pointers.
35058 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35060 PR rtl-optimization/60818
35061 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35062 a compare of comparisons with the thing compared if this results
35063 in a different machine mode.
35065 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35067 * alias.c (base_alias_check): Fix typo in comment.
35068 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35069 * cgraphunit.c (symbol_table::compile): Likewise.
35070 * collect2.c (maybe_run_lto_and_relink): Likewise.
35071 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35072 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35073 * config/avr/avr.c (avr_map_op_t): Likewise.
35074 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35075 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35076 * config/epiphany/epiphany.md (movcc): Likewise.
35077 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35078 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35080 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35081 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35082 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35083 * config/sh/sh.c (sh_rtx_costs): Likewise.
35084 * fold-const.c (fold_truth_andor): Likewise.
35085 * genautomata.c (collapse_flag): Likewise.
35086 * gengtype.h (struct type::u::s): Likewise.
35087 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35088 * input.c (FORMAT_AMOUNT): Likewise.
35089 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35090 (known_aggs_to_agg_replacement_list): Likewise.
35091 * ipa-inline-analysis.c: Likewise.
35092 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35093 * ipa-polymorphic-call.c
35094 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35095 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35096 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35098 * modulo-sched.c (apply_reg_moves): Likewise.
35099 * omp-expand.c (build_omp_regions_1): Likewise.
35100 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35101 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35102 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35103 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35104 * value-prof.c: Likewise.
35105 * var-tracking.c (val_reset): Likewise.
35107 2017-04-03 Richard Biener <rguenther@suse.de>
35109 PR tree-optimization/80275
35110 * fold-const.c (split_address_to_core_and_offset): Handle
35113 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35115 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35116 descriptors is at least equal to that of functions.
35118 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35120 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35122 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35125 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35126 (mov<IMOD4:mode>): New expander.
35127 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35129 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35131 PR rtl-optimization/79405
35132 * fwprop.c (propagations_left): New variable.
35133 (forward_propagate_into): Decrement it.
35134 (fwprop_init): Initialize it.
35135 (fw_prop): If the variable has reached zero, stop propagating.
35136 (fwprop_addr): Ditto.
35138 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35141 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35142 a FUNCTION_DECL, pass it as decl instead of origin to
35145 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35147 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35150 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35153 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35154 TARGET_VSX_SMALL_INTEGER.
35156 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35158 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35159 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35161 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35163 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35164 extraction from odd-numbered MSA register.
35166 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35168 PR middle-end/80173
35169 * expmed.c (store_bit_field_1): Don't attempt to create
35170 a word subreg out of hard registers wider than word if they
35171 have HARD_REGNO_NREGS of 1 for their mode.
35173 PR middle-end/80163
35174 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35175 conversions to integer types wider than word and pointer.
35178 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35179 (rtx_equal_for_cselib_p): Pass 0 to it.
35180 * cselib.c (cselib_hasher::equal): Likewise.
35181 (rtx_equal_for_cselib_1): Add depth argument. If depth
35182 is 128, don't look up VALUE locs and punt. Increment
35183 depth in recursive calls when walking VALUE locs.
35185 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35187 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35188 (make_gcov_file_name): Use the canonical path name for generating
35190 (read_line): Fix handling of files with ascii null bytes.
35192 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35194 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35195 to initialise a vector register instead
35196 of using a const_int.
35198 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35200 PR translation/80189
35201 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35202 diagnostic messages.
35204 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35207 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35208 (dfp_diex_<mode>): Update mode of operand 1.
35209 * doc/extend.texi (dxex, dxexq): Document change to return type.
35210 (diex, diexq): Document change to argument type.
35212 2017-03-30 Martin Jambor <mjambor@suse.cz>
35215 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35216 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35217 it reflects the signature changes performed at the callee side.
35218 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35219 to cgraph_build_function_type_skip_args.
35220 (build_function_decl_skip_args): Adjust call to the above function.
35222 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35225 * config/i386/sse.md
35226 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35227 register as dest whenever it is a MEM not rtx_equal_p to the
35228 corresponding dup operand, and when forcing into reg move the
35229 reg into the memory afterwards.
35230 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35231 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35232 for the force_reg mode.
35233 (avx512vl_vextractf128<mode>): Use register as dest either
35234 always when a MEM, or when it is a MEM not rtx_equal_p to the
35235 corresponding dup operand, or even not when it is a CONST_VECTOR
35236 depending on the mode and lo vs. hi.
35237 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35239 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35240 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35241 Likewise. Require that operands[2] is even.
35242 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35243 Remove extraneous parens. Require that operands[2] is a multiple
35245 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35246 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35247 disallow memory then.
35249 2017-03-30 Richard Biener <rguenther@suse.de>
35251 PR tree-optimization/77498
35252 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35253 to non-constants over backedges.
35255 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35257 PR rtl-optimization/80233
35258 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35259 as last_combined_insn. Do not test for BARRIER_P separately.
35261 2017-03-29 Andreas Schwab <schwab@suse.de>
35264 * calls.c (prepare_call_address): Convert funexp to Pmode before
35265 copying to temp reg.
35267 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35269 PR tree-optimization/80158
35270 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35271 Handle possible future case of more than one alternate
35273 (replace_rhs_if_not_dup): Likewise.
35274 (replace_one_candidate): Likewise.
35276 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35278 PR rtl-optimization/80193
35279 * ira.c (ira): Do not check allocation for LRA.
35281 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35283 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35284 (nvptx_output_simt_exit): Declare.
35285 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35286 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35287 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35288 (nvptx_declare_function_name): ...here. Emit declaration of local
35289 memory space buffer for omp_simt_enter insn.
35290 (nvptx_output_unisimt_switch): New.
35291 (nvptx_output_softstack_switch): New.
35292 (nvptx_output_simt_enter): New.
35293 (nvptx_output_simt_exit): New.
35294 * config/nvptx/nvptx.h (struct machine_function): New fields
35295 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35296 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35297 (UNSPECV_SIMT_EXIT): Ditto.
35298 (omp_simt_enter_insn): New insn.
35299 (omp_simt_enter): New expansion.
35300 (omp_simt_exit): New insn.
35301 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35303 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35304 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35305 (expand_GOMP_SIMT_EXIT): New.
35306 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35307 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35308 (GOMP_SIMT_EXIT): Ditto.
35309 * target-insns.def (omp_simt_enter): New insn.
35310 (omp_simt_exit): Ditto.
35311 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35313 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35314 (lower_rec_input_clauses): Likewise.
35315 (lower_lastprivate_clauses): Handle SIMT privatization.
35317 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35318 (ompdevlow_adjust_simt_enter): New.
35319 (find_simtpriv_var_op): New.
35320 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35321 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35323 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35324 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35325 (copy_decl_for_dup_finish): Ditto.
35327 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35329 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35332 * config/i386/i386.c (ix86_option_override_internal): Always
35333 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35335 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35337 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35339 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35341 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35342 mark new edge's irreducible flag accordign to it.
35343 (vect_do_peeling): Check loop preheader edge's irreducible flag
35344 and pass it to function slpeel_add_loop_guard.
35346 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35348 PR tree-optimization/80218
35349 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35350 Update block frequencies and counts.
35352 2017-03-28 Richard Biener <rguenther@suse.de>
35354 PR tree-optimization/78644
35355 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35356 of a simplification result we may not use it at all.
35358 2017-03-28 Richard Biener <rguenther@suse.de>
35361 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35362 without arguments, generate default definition of a SSA name.
35364 2017-03-28 Richard Biener <rguenther@suse.de>
35366 PR middle-end/80222
35367 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35368 TYPE_REF_CAN_ALIAS_ALL references.
35369 * fold-const.c (fold_indirect_ref_1): Likewise.
35371 2017-03-28 Martin Liska <mliska@suse.cz>
35374 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35375 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35377 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35378 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35380 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35381 (EXTRA_SPECS): Define.
35382 (SUBTARGET_EXTRA_SPECS): Likewise.
35383 (SUBTARGET_CPP_SPEC): Likewise.
35384 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35385 SUBTARGET_EXTRA_SPECS.
35386 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35388 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35390 * config/arc/simdext.md (vst64_insn): Update pattern.
35391 (vld32wh_insn): Likewise.
35392 (vld32wl_insn): Likewise.
35393 (vld64_insn): Likewise.
35394 (vld32_insn): Likewise.
35396 2017-03-28 Marek Polacek <polacek@redhat.com>
35399 * fold-const.c (fold_comparison): Use protected_set_expr_location
35400 instead of SET_EXPR_LOCATION.
35402 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35404 * tree.c (add_expr): Avoid name lookup warning.
35406 2017-03-27 Jeff Law <law@redhat.com>
35408 PR tree-optimization/80216
35409 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35410 function name. Limit recursion depth.
35411 (record_temporary_equivalences): Corresponding changes.
35413 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35415 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35418 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35421 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35423 * cfgcleanup.c (reg_note_cfa_p): New array.
35424 (insns_have_identical_cfa_notes): New function.
35425 (old_insns_match_p): Don't cross-jump in between /f
35426 and non-/f instructions. If both i1 and i2 are frame related,
35427 verify all CFA notes, their order and content.
35429 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35432 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35433 HImode and SImode with zero extend to DImode to one insn.
35434 (bswap<mode>2_extenddi): Likewise.
35435 (bswapsi2_extenddi): Likewise.
35436 (bswaphi2_extendsi): Likewise.
35437 (bswaphi2): Combine bswap HImode and SImode into one insn.
35438 Separate memory insns from swapping register.
35439 (bswapsi2): Likewise.
35440 (bswap<mode>2): Likewise.
35441 (bswaphi2_internal): Delete, no longer used.
35442 (bswapsi2_internal): Likewise.
35443 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35444 store, and gpr<-gpr swap insns.
35445 (bswap<mode>2_store): Likewise.
35446 (bswaphi2_reg): Register only splitter, combine with the splitter.
35447 (bswaphi2 splitter): Likewise.
35448 (bswapsi2_reg): Likewise.
35449 (bswapsi2 splitter): Likewise.
35450 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35451 the insns into load, store, and register/register insns.
35452 (bswapdi2_ldbrx): Likewise.
35453 (bswapdi2_load): Likewise.
35454 (bswapdi2_store): Likewise.
35455 (bswapdi2_reg): Likewise.
35457 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35459 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35460 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35462 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35465 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35467 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35468 special handling for target option conflicts between dform
35469 options (-mpower9-dform, -mpower9-dform-vector,
35470 -mpower9-dform-scalar) and -mno-direct-move.
35472 2017-03-27 Richard Biener <rguenther@suse.de>
35474 PR tree-optimization/80181
35475 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35477 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35479 * config/arc/predicates.md (move_double_src_operand): Replace the
35480 call to move_double_src_operand with a call to address_operand.
35482 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35484 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35485 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35486 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35488 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35490 * config/arc/predicates.md (long_immediate_loadstore_operand):
35491 Consider scaled addresses cases.
35493 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35495 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35496 restored when in interrupt.
35497 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35498 doesn't have delay slot.
35500 2017-03-27 Richard Biener <rguenther@suse.de>
35503 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35504 inlined thunk clones.
35506 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35509 * asan.c (instrument_derefs): Copy over last operand from
35510 original COMPONENT_REF to the new COMPONENT_REF with
35511 DECL_BIT_FIELD_REPRESENTATIVE.
35512 * ubsan.c (instrument_object_size): Likewise.
35514 2017-03-27 Richard Biener <rguenther@suse.de>
35516 PR tree-optimization/80170
35517 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35518 sure DR/SCEV didnt fold in constants we do not see when looking
35519 at the reference base alignment.
35521 2017-03-27 Richard Biener <rguenther@suse.de>
35523 PR middle-end/80171
35524 * gimple-fold.c (fold_ctor_reference): Properly guard against
35525 NULL return value from canonicalize_constructor_val.
35527 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35530 * config/i386/i386.c (ix86_expand_builtin)
35531 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35532 flags reg setting and flags reg using instructions.
35533 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35534 clobbering instructions to zero extend op2.
35536 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35538 * doc/install.texi (Configuration) <--with-aix-soname>:
35539 Update link to AIX ld.
35541 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35543 PR rtl-optimization/80160
35544 PR rtl-optimization/80159
35545 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35546 reg_alternate_class into account.
35548 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35551 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35552 to consider in curr_insn_transform.
35554 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35556 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35557 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35558 and emit_mode_inner.
35560 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35562 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35563 argument to the overloaded builtin variants. Use the new flag to
35564 deprecate certain builtin variants.
35565 * config/s390/s390-builtin-types.def: Add new builtin types.
35566 * config/s390/s390-builtins.h: Support new flags field for
35567 overloaded builtins.
35568 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35569 (s390_macro_to_expand): Enable vector float data type.
35570 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35571 builtins by incrementing the __VEC__ version number.
35572 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35574 (s390_resolve_overloaded_builtin): Emit error messages depending
35575 on the builtin flags.
35576 * config/s390/s390.c (s390_expand_builtin): Support additional
35577 flags argument. Change error message to match the messages
35578 emitted in s390-c.c.
35579 * config/s390/s390.md: New UNSPEC_* constants.
35580 (op_type): Add new instruction types.
35581 * config/s390/vecintrin.h: Add new builtins and test data class
35583 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35584 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35585 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35586 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35587 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35588 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35590 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35591 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35592 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35593 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35595 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35596 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35597 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35598 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35600 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35601 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35602 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35603 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35604 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35605 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35606 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35608 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35609 ("vec_scatter_element<V_HW_4:mode>_DI")
35610 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35611 ("vec_fpint<mode>", "vflls")
35612 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35613 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35614 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35615 ("*vec_cmphe<mode>_cc"): ... these.
35617 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35618 mode constant instead of magic value.
35620 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35622 * config/s390/s390.c (s390_expand_vec_compare): Support other
35623 vector floating point modes than just V2DF.
35624 (s390_expand_vcond): Likewise.
35625 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35626 (s390_cannot_change_mode_class): Prevent mode changes between TF
35627 and V1TF in vector registers.
35628 * config/s390/s390.md (DF, SF): New mode attributes.
35629 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35630 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35631 SFmode support for VRs.
35632 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35634 (VFT, VF_HW): New mode iterators.
35635 (vw, sdx): New mode attributes.
35636 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35637 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35638 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35639 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35640 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35641 also the new vector floating point modes. Renaming to ...
35643 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35644 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35645 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35646 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35647 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35648 ("vec_unordered<mode>"): ... these.
35650 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35651 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35652 ("*vec_extendv2df"): New insn definitions.
35654 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35656 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35657 ("mulditi3_2", "*muldi3_sign"): New patterns.
35658 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35659 rename the pattern definition.
35661 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35663 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35665 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35667 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35669 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35670 instruction if possible.
35671 * config/s390/vector.md (vec_halfnumelts): New mode
35673 ("*vec_vllezlf<mode>"): New pattern.
35675 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35677 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35678 ("popcountv4si2", "popcountv2di2"): Rename to ...
35679 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35680 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35682 ("popcount<mode>2_vxe"): New pattern.
35684 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35686 * common/config/s390/s390-common.c (processor_flags_table): Add
35688 * config.gcc: Add arch12.
35689 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35690 Default to arch12 for unknown CPU model numbers.
35691 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35692 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35693 PROCESSOR_max sanity check.
35694 * config/s390/s390-opts.h (enum processor_type): Add
35696 * config/s390/s390.c (processor_table): Add arch12.
35697 (s390_expand_builtin): Add check for B_VXE flag.
35698 (s390_issue_rate): Add PROCESSOR_ARCH12.
35699 (s390_get_sched_attrmask): Likewise.
35700 (s390_get_unit_mask): Likewise.
35701 (s390_sched_score): Enable z13 scheduling for arch12.
35702 (s390_sched_reorder): Likewise.
35703 (s390_sched_variable_issue): Likewise.
35704 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35706 (s390_tune_attr): Use z13 scheduling also for arch12.
35707 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35708 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35709 (TARGET_VXE_P): New macros.
35710 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35711 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35712 * config/s390/s390.opt: Add arch12 as processor_type.
35714 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35716 * config/s390/s390.md
35717 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35718 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35719 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35721 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35722 Rename expanders to ...
35724 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35725 ("fixuns_truncdddi2_emu"): ... these.
35727 ("fixuns_trunc<mode>si2_emu"): New expander.
35729 ("*fixuns_truncdfdi2_z13"): Rename to ...
35730 ("*fixuns_truncdfdi2_vx"): ... this.
35732 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35734 * config/s390/2964.md: Remove the single element vector compare
35735 instructions which are no longer used.
35736 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35738 (s390_canonicalize_comparison): Remove handling of DFmode
35740 (s390_expand_vec_compare_scalar): Remove function.
35741 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35742 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35744 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35746 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35748 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35750 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35751 will anyway by matched by mov<mode>_64dfp.
35753 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35755 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35756 vlef/vstef. Add missing operand to vleif.
35758 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35760 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35761 pair for all vector types with 64 bit elements.
35762 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35763 * config/s390/vector.md (V_HW_64): ... here.
35764 (V_128_NOSINGLE): New mode iterator.
35765 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35766 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35767 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35768 ("*vec_load_pairv2di"): Change to ...
35769 ("*vec_load_pair<mode>"): ... this one.
35771 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35773 * config/s390/constraints.md: Add comments.
35774 (jKK): Reject element sizes > 8 bytes.
35775 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35777 * config/s390/s390.md: Add the s_operand checks formerly in
35778 s390_split_ok_p to various splitters where they are still
35780 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35781 for 128 bit vectors. Plus two splitters.
35783 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35785 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35788 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35791 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35792 error if the boundary argument is not constant.
35794 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35796 PR rtl-optimization/80112
35797 * loop-doloop.c (doloop_condition_get): Don't check condition
35798 if cmp isn't SET with IF_THEN_ELSE src.
35800 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35802 PR tree-optimization/80158
35803 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35804 replacing a candidate statement, also replace it for the
35805 candidate's alternate interpretation.
35806 (replace_rhs_if_not_dup): Likewise.
35807 (replace_one_candidate): Likewise.
35809 2017-03-24 Richard Biener <rguenther@suse.de>
35811 PR tree-optimization/80167
35812 * graphite-isl-ast-to-gimple.c
35813 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35815 (translate_isl_ast_to_gimple::get_rename): Likewise.
35817 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35819 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35820 handling of certain combinations of target options, including the
35821 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35822 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35824 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35827 * config/arm/arm.md (*load_multiple): Add reload_completed to
35828 matching condition.
35830 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35831 Richard Biener <rguenth@suse.de>
35833 PR tree-optimization/79908
35834 PR tree-optimization/80136
35835 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35836 been cast away, gimplify_and_add suffices.
35838 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35840 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35842 2017-03-23 Richard Biener <rguenther@suse.de>
35844 PR tree-optimization/80032
35845 * gimplify.c (gimple_push_cleanup): Forced unconditional
35846 cleanups still have to go to the conditional_cleanups
35849 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35851 PR tree-optimization/80072
35852 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35854 (next_operand_entry_id): Change type to unsigned int.
35855 (sort_by_operand_rank): Make sure to return the right return value
35856 even if unsigned fields are bigger than INT_MAX.
35857 (struct oecount): Change cnt and id type to unsigned int.
35858 (oecount_hasher::equal): Formatting fix.
35859 (oecount_cmp): Make sure to return the right return value
35860 even if unsigned fields are bigger than INT_MAX.
35861 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35864 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35865 TREE_READONLY on result if writing it more than once.
35868 * doc/invoke.texi (-fsanitize=thread): Document that with
35869 -fnon-call-exceptions atomics are not able to throw
35873 * tsan.c: Include tree-eh.h.
35874 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35875 maybe_clean_or_replace_eh_stmt where needed.
35876 (instrument_memory_accesses): Add cfg_changed argument.
35877 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35878 if it returned true.
35879 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35881 PR rtl-optimization/63191
35882 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35883 wrapper function, moved the whole old content into ...
35884 (ix86_delegitimize_address_1): ... this. New inline function.
35885 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35886 true as last argument instead of ix86_delegitimize_address.
35888 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35890 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35891 cortexa57_branch_cost.
35893 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35895 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35897 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35900 * doc/md.texi (Constraints): Document wA constraint.
35901 * config/rs6000/constraints.md (wA): New.
35902 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35903 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35904 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35905 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35907 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35910 * gimplify.c (is_oacc_declared): New function.
35911 (oacc_default_clause): Use it to set default flags for acc declared
35912 variables inside parallel regions.
35913 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35914 declared variables.
35915 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35916 declare attribute to any decl as necessary.
35918 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35921 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35922 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35923 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35924 (arm_arch_lpae): This.
35925 * config/arm/arm.c (arm_arch7ve): Rename into ...
35926 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35927 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35930 2017-03-22 Martin Liska <mliska@suse.cz>
35933 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35934 error message instead of an ICE.
35936 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35938 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35940 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35942 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35944 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35947 2017-03-21 Martin Sebor <msebor@redhat.com>
35949 * doc/extend.texi: Use "cannot" instead of "can't."
35950 * doc/hostconfig.texi: Same.
35951 * doc/install.texi: Same.
35952 * doc/invoke.texi: Same.
35953 * doc/loop.texi: Same.
35954 * doc/md.texi: Same.
35955 * doc/objc.texi: Same.
35956 * doc/rtl.texi: Same.
35957 * doc/tm.texi: Same.
35958 * doc/tm.texi.in: Same.
35959 * doc/trouble.texi: Same.
35961 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35964 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35965 (collect_checksum_attributes): Set it.
35966 (die_checksum_ordered): Use it.
35968 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35970 PR tree-optimization/79908
35971 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35972 change: For a VA_ARG whose LHS has been cast away, use
35973 force_gimple_operand to construct the side effects.
35975 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35977 PR translation/80001
35978 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35979 more amenable to translation.
35980 (oacc_loop_auto_partitions): Likewise.
35982 2017-03-21 Marek Polacek <polacek@redhat.com>
35983 Martin Sebor <msebor@redhat.com>
35985 PR tree-optimization/80109
35986 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35987 on INTEGRAL_TYPE_P.
35989 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35990 Segher Boessenkool <segher@kernel.crashing.org>
35993 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35994 check reg_used_between_p between insn and one of succ or succ2
35995 depending on if succ is artificial insn not inserted into insn
35998 2017-03-21 Martin Liska <mliska@suse.cz>
36000 PR gcov-profile/80081
36001 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
36002 * doc/gcc.texi: Include gcov-dump stuff.
36003 * doc/gcov-dump.texi: New file.
36005 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36007 PR rtl-optimization/79150
36008 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36009 conditional jump, if the jump is the last insn of the loop.
36011 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36012 Richard Biener <rguenth@suse.de>
36014 PR tree-optimization/79908
36015 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36016 been cast away, use force_gimple_operand to construct the side
36019 2017-03-21 Martin Liska <mliska@suse.cz>
36021 PR libfortran/79956
36022 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36025 2017-03-21 Brad Spengler <spender@grsecurity.net>
36028 * plugin.c (htab_hash_plugin): New function.
36029 (add_new_plugin): Use it and adjust.
36030 (parse_plugin_arg_opt): Adjust.
36031 (init_one_plugin): Likewise.
36033 2017-03-21 Richard Biener <rguenther@suse.de>
36035 PR tree-optimization/80032
36036 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36037 if set force the cleanup to happen unconditionally.
36038 (gimplify_target_expr): Push inserted clobbers with force_uncond
36039 to avoid them being removed by control-dependent DCE.
36041 2017-03-21 Richard Biener <rguenther@suse.de>
36043 PR tree-optimization/80122
36044 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36045 va_arg_pack_len when the inlined call stmt requires pack
36047 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36049 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36052 * tsan.c (instrument_builtin_call): If the memory model argument
36053 is not a constant, assume it is valid.
36056 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36059 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36061 PR rtl-optimization/79910
36062 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36063 if its dest is used by an insn before I2 (other than the combined
36064 insns themselves, which are properly handled already).
36066 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36069 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36071 * combine.c (record_used_regs): New static function.
36072 (try_combine): Handle situations where there is an additional
36073 instruction between I2 and I3 which needs to have a LOG_LINK
36077 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36079 * combine.c (try_combine): Delete redundant i1 test. Call
36080 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36082 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36085 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36086 alternatives 13/14.
36088 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36090 PR tree-optimization/80054
36091 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36092 the optimization if a PHI or any of its arguments is not dominated
36093 by the candidate's basis. Use gphi* rather than gimple* as
36095 (replace_profitable_candidates): Clean up a gimple* variable that
36096 should be a gphi* variable.
36098 2017-03-20 Martin Sebor <msebor@redhat.com>
36101 * doc/extend.texi (attribute constructor): Document present limitation.
36103 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36106 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36107 __POWER9_VECTOR__ #ifdef control, change template definition to
36108 use Power9-specific built-in function.
36109 (vec_any_eq): Likewise.
36110 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36111 to control outcomes from this test.
36112 (vector_ae_<mode>p): For VEC_F modes, likewise.
36114 2017-03-20 Ian Lance Taylor <iant@google.com>
36116 * config/i386/i386.c (ix86_function_regparm): Save an extra
36117 register for -fsplit-stack with DECL_STATIC_CHAIN.
36119 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36122 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36123 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36125 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36127 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36129 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36132 2017-03-20 Marek Polacek <polacek@redhat.com>
36135 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36137 2017-03-20 Richard Biener <rguenther@suse.de>
36139 PR tree-optimization/80113
36140 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36141 allocate extra SSA name for PHI def.
36142 (add_close_phis_to_outer_loops): Likewise.
36143 (add_close_phis_to_merge_points): Likewise.
36144 (copy_loop_close_phi_args): Likewise.
36145 (copy_cond_phi_nodes): Likewise.
36147 2017-03-20 Martin Liska <mliska@suse.cz>
36149 PR middle-end/79753
36150 * tree-chkp.c (chkp_build_returned_bound): Do not build
36151 returned bounds for a LHS that's not a BOUNDED_P type.
36153 2017-03-20 Martin Liska <mliska@suse.cz>
36157 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36158 COMPLEX_CST and VECTOR_CST.
36160 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36163 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36164 target operand. A new splitter adds the clobber statement in case
36165 the target operand is dead anyway.
36167 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36169 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36170 to age-old versions of binutils and glibc.
36172 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36174 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36176 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36178 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36180 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36182 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36183 requirement for binutils 2.13.
36185 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36187 * combine.c (try_combine): Delete redundant i1 test. Call
36188 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36190 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36192 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36193 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36195 <riscv64-*-elf>: Re-arrange section
36196 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36197 <riscv32-*-linux>: Likewise.
36198 <riscv64-*-elf>: Likewise
36199 <riscv64-*-linux>: Likewise.
36201 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36204 * aarch64.opt(verbose-cost-dump): Fix typo.
36206 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36209 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36210 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36212 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36214 * reload.c (find_reloads): When reloading a nonoffsettable address,
36215 use RELOAD_OTHER for it and its address reloads.
36217 PR rtl-optimization/79910
36218 * combine.c (record_used_regs): New static function.
36219 (try_combine): Handle situations where there is an additional
36220 instruction between I2 and I3 which needs to have a LOG_LINK
36223 2017-03-17 Jeff Law <law@redhat.com>
36225 PR tree-optimization/71437
36226 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36227 conditional in the hash table first.
36228 (vrp_dom_walker::before_dom_children): Extract condition from
36229 ASSERT_EXPR. Record condition, its inverion and any implied
36230 conditions as well.
36232 2017-03-17 Marek Polacek <polacek@redhat.com>
36233 Markus Trippelsdorf <markus@trippelsdorf.de>
36235 PR tree-optimization/80079
36236 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36239 2017-03-17 Richard Biener <rguenther@suse.de>
36241 PR middle-end/80075
36242 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36243 Properly verify the LHS before the RHS possibly claims to be
36245 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36248 2017-03-17 Martin Jambor <mjambor@suse.cz>
36250 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36251 (List of -O2 options): Likewise.
36252 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36255 2017-03-17 Tom de Vries <tom@codesourcery.com>
36257 * gcov-dump.c (print_usage): Print bug_report_url.
36259 2017-03-17 Richard Biener <rguenther@suse.de>
36261 PR middle-end/80050
36262 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36263 (parser::peek): Likewise.
36265 2017-03-17 Richard Biener <rguenther@suse.de>
36267 PR tree-optimization/80048
36268 * sese.c (free_sese_info): Properly release rename_map and
36269 copied_bb_map elements.
36271 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36273 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36274 Add linked-list forward and backlinks. Insert on
36275 construction, remove on destruction.
36276 (class pass_store_merging): Add m_stores_head field.
36277 (pass_store_merging::terminate_and_process_all_chains):
36278 Iterate over m_stores_head list.
36279 (pass_store_merging::terminate_all_aliasing_chains):
36281 (pass_store_merging::execute): Check for debug stmts first.
36282 Push new chains onto the m_stores_head stack.
36284 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36287 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36288 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36289 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36291 2017-03-16 Jeff Law <law@redhat.com>
36293 PR tree-optimization/71437
36294 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36295 member function. Implementation moved into after_dom_children
36296 member function and into the threader's thread_outgoing_edges
36298 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36299 some code into new thread_outgoing_edges.
36300 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36301 definition. Simplify marker handling (do it here). Assume we always
36302 have the available expression and the const/copies tables.
36303 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36305 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36306 * tree-vrp.c (equiv_stack): No longer file scoped.
36307 (vrp_dom_walker): New class.
36308 (vrp_dom_walker::before_dom_children): New member function.
36309 (vrp_dom_walker::after_dom_children): Likewise.
36310 (identify_jump_threads): Setup domwalker. Use it rather than
36311 walking edges in a random order by hand. Simplify setup/finalization.
36312 (finalize_jump_threads): Remove.
36313 (vrp_finalize): Do not call identify_jump_threads here.
36314 (execute_vrp): Do it here instead and call thread_through_all_blocks
36317 PR tree-optimization/71437
36318 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36320 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36322 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36323 (dom_opt_dom_walker::thread_across_edge): Remove
36324 handle_dominating_asserts argument. All callers changed.
36325 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36326 changes. Remove calls to lhs_of_dominating_assert. Other
36327 uses of handle_dominating_asserts turn into unconditional code
36328 (simplify_control_stmt_condition_1): Likewise.
36329 (simplify_control_stmt_condition): Likewise.
36330 (thread_through_normal_block, thread_across_edge): Likewise.
36331 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36332 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36333 object if it is not an SSA_NAME.
36334 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36335 before calling into the VRP specific simplifiers.
36336 (identify_jump_threads): Remove handle_dominating_asserts
36339 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36342 * tree-diagnostic.c (default_tree_printer): No longer static.
36343 * tree-diagnostic.h (default_tree_printer): New prototype.
36345 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36347 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36348 Change ins into fmov.
36350 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36352 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36353 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36354 Use h_con constraint for operand 1.
36355 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36356 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36358 2017-03-15 Jeff Law <law@redhat.com>
36360 PR tree-optimization/71437
36361 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36362 (record_temporary_equivalences): Use it.
36364 PR tree-optimization/71437
36365 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36366 tree-ssa-scopedtables.
36367 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36368 (record_conditions, record_cond, vuse_eq): Likewise.
36369 (record_edge_info): Adjust to API tweak of record_conditions.
36370 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36371 (record_temporary_equivalences, optimize_stmt): Likewise.
36372 (eliminate_redundant_computations): Likewise.
36373 (record_equivalences_from_stmt): Likewise.
36374 * tree-ssa-scopedtables.c: Include options.h and params.h.
36375 (vuse_eq): New function, moved from tree-ssa-dom.c
36376 (build_and_record_new_cond): Likewise.
36377 (record_conditions): Likewise. Accept vector of conditions rather
36378 than edge_equivalence structure for first argument.
36379 for the first argument.
36380 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36381 from tree-ssa-dom.c.
36382 (avail_exprs_stack::record_cond): Likewise.
36383 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36384 from tree-ssa-dom.c.
36385 (avail_exprs_stack): Add new member functions lookup_avail_expr
36387 (record_conditions): Declare.
36389 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36392 * lra-constraints.c (process_alt_operands): Increase reject for
36393 reloading an input/output operand.
36395 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36398 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36399 insns to convert from signed/unsigned char/short to IEEE 128-bit
36401 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36403 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36406 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36407 subreg of inner mode for values already in registers.
36409 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36411 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36412 iteration reg is used after the loop.
36414 2017-03-14 Martin Sebor <msebor@redhat.com>
36416 PR tree-optimization/79800
36417 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36418 precision in negative-positive range.
36419 (format_floating): Call non-const overload with adjusted precision.
36421 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36424 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36427 2017-03-14 Martin Sebor <msebor@redhat.com>
36429 PR middle-end/80020
36430 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36431 * builtins.def (aligned_alloc): Use it.
36434 * Makefile.in (GTFILES): Add calls.c.
36435 * calls.c: Include "gt-calls.h".
36437 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36439 PR rtl-optimization/79728
36440 * regs.h (struct target_regs): New field
36441 x_contains_allocatable_regs_of_mode.
36442 (contains_allocatable_regs_of_mode): New macro.
36443 * reginfo.c (init_reg_sets_1): Initialize it, and change
36444 contains_reg_of_mode so it includes global regs as well.
36445 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36446 rather than contains_regs_of_mode.
36448 2017-03-14 Martin Liska <mliska@suse.cz>
36450 * doc/invoke.texi: Document options that can't be combined with
36451 -fcheck-pointer-bounds.
36453 2017-03-14 Martin Liska <mliska@suse.cz>
36455 PR middle-end/79831
36456 * doc/invoke.texi (-Wchkp): Document the option.
36458 2017-03-14 Martin Liska <mliska@suse.cz>
36460 * Makefile.in: Install gcov-dump.
36462 2017-03-14 Martin Liska <mliska@suse.cz>
36464 * multiple_target.c (expand_target_clones): Bail out for
36465 an invalid attribute.
36467 2017-03-14 Richard Biener <rguenther@suse.de>
36469 * alias.c (struct alias_set_entry): Pack properly.
36470 * cfgloop.h (struct loop): Likewise.
36471 * cse.c (struct set): Likewise.
36472 * ipa-utils.c (struct searchc_env): Likewise.
36473 * loop-invariant.c (struct invariant): Likewise.
36474 * lra-remat.c (struct cand): Likewise.
36475 * recog.c (struct change_t): Likewise.
36476 * rtl.h (struct address_info): Likewise.
36477 * symbol-summary.h (function_summary): Likewise.
36478 * tree-loop-distribution.c (struct partition): Likewise.
36479 * tree-object-size.c (struct object_size_info): Likewise.
36480 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36481 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36482 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36483 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36484 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36485 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36486 (struct _stmt_vec_info): Likewise.
36488 2017-03-14 Martin Liska <mliska@suse.cz>
36491 * multiple_target.c (create_dispatcher_calls): Check that
36492 a target can create a function dispatcher.
36494 2017-03-14 Martin Liska <mliska@suse.cz>
36497 * multiple_target.c (expand_target_clones): Drop local.local
36498 flag for default implementation.
36500 2017-03-14 Richard Biener <rguenther@suse.de>
36502 PR tree-optimization/80030
36503 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36505 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36507 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36508 gcc_fallthrough() instead of __attribute__((fallthrough));
36510 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36512 * doc/gcc.texi: Remove "up" link to (DIR).
36513 * doc/gccint.texi: Ditto.
36515 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36517 * doc/install.texi (Specific) <avr>: Remove reference to
36520 2017-03-13 Jeff Law <law@redhat.com>
36522 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36523 attribute rather than comments.
36525 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36526 match_scratch operand is highest.
36528 2017-03-13 Martin Liska <mliska@suse.cz>
36530 PR middle-end/78339
36531 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36532 is a CHKP clone, use original declaration.
36534 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36536 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36537 (arc_conditional_register_usage): Use a different allocation order
36538 when optimizing for size.
36539 * common/config/arc/arc-common.c (arc_option_optimization_table):
36540 Section anchors default on when optimizing for size.
36542 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36544 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36546 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36548 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36549 * config/arc/arc.md (cpu_facility): Add cd variant.
36550 (*movqi_insn): Add code density variant.
36551 (*movhi_insn): Likewise.
36552 (*movqi_insn): Likewise.
36553 (*addsi3_mixed): Likewise.
36554 (subsi3_insn): Likewise.
36556 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36558 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36560 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36562 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36563 expressions with MINUS and UNARY ops.
36565 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36568 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36570 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36571 between vec_select and vector argument.
36572 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36573 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36574 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36575 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36576 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36577 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36579 2017-03-13 Richard Biener <rguenther@suse.de>
36582 * params.def (vect-max-peeling-for-alignment): Fix typo.
36584 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36586 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36587 issue that only occurred with binutils below 2.18.
36589 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36591 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36592 refer to binutils 2.11/2.12 minimum.
36594 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36596 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36597 ftp.kernel.org and simplify binutils requirement.
36599 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36601 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36603 (Optimize Options): Ditto. Also remove redundancy.
36605 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36607 PR translation/79848
36608 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36610 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36611 to G_ to avoid double translation.
36613 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36615 PR translation/79923
36616 * auto-profile.c (get_combined_location): Convert leading
36617 character of diagnostics to lower case and remove trailing period.
36618 (read_profile): Likewise for various diagnostics.
36619 * config/arm/arm.c (arm_option_override): Remove trailing period
36620 from various diagnostics.
36621 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36622 (msp430_expand_delay_cycles): Likewise.
36624 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36627 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36628 full command-line argument, rather than just "str".
36629 (aarch64_validate_march): Likewise.
36630 (aarch64_validate_mtune): Likewise.
36632 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36634 PR rtl-optimization/78911
36635 * lra-assigns.c (must_not_spill_p): New function.
36636 (spill_for): Use it.
36638 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36640 PR tree-optimization/79981
36641 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36642 ATOMIC_COMPARE_EXCHANGE ifn result.
36643 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36644 IFN_ATOMIC_COMPARE_EXCHANGE.
36646 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36649 * opts.c (parse_sanitizer_options): Add missing question mark to
36650 "did you mean" message.
36652 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36654 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36656 (VMULEUH_UNS): Likewise.
36657 (VMULOUB_UNS): Likewise.
36658 (VMULOUH_UNS): Likewise.
36659 * config/rs6000/rs6000.c (builtin_function_type): Remove
36660 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36662 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36665 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36666 x with result of extra_parsing_for_operand_code_0.
36667 (function_reader::extra_parsing_for_operand_code_0): Convert
36668 return type from void to rtx, returning x. When reading
36669 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36670 larger size containing struct block_symbol.
36672 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36674 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36675 -mfloat128-hardware without -m64.
36677 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36680 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36681 entries to the case statement that marks unsigned arguments to
36682 overloaded functions.
36684 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36686 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36687 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36689 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36692 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36693 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36695 2017-03-10 Martin Liska <mliska@suse.cz>
36699 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36700 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36703 2017-03-10 Olivier Hainque <hainque@adacore.com>
36705 * tree-switch-conversion (array_value_type): Start by resetting
36706 candidate type to it's main variant.
36708 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36710 PR rtl-optimization/79909
36711 * combine.c (try_combine): Use simplify_replace_rtx on individual
36712 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36713 of the whole CALL_INSN_FUNCTION_USAGE.
36715 PR tree-optimization/79972
36716 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36717 get_range_info on SSA_NAMEs. Formatting fixes.
36719 2017-03-10 Richard Biener <rguenther@suse.de>
36720 Jakub Jelinek <jakub@redhat.com>
36722 PR tree-optimization/77975
36723 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36724 edge to be constant.
36725 (get_val_for): For constant x return it. Formatting fix.
36726 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36727 would use the same bases as the current one.
36729 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36731 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36732 instead of vec_select for V1TImode.
36733 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36735 (VSX_LE_128): Add V1TI to this mode iterator.
36736 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36737 (*vsx_le_perm_store_<mode>): Likewise.
36738 (pre-reload splitter for VSX stores): Likewise.
36739 (post-reload splitter for VSX stores): Likewise.
36740 (*vsx_xxpermdi2_le_<mode>): Likewise.
36741 (*vsx_lxvd2x2_le_<mode>): Likewise.
36742 (*vsx_stxvd2x2_le_<mode>): Likewise.
36744 2017-03-09 Michael Eager <eager@eagercon.com>
36746 Correct failures with --enable-checking=yes,rtl.
36748 * config/microblaze/microblaze.c (microblaze_expand_shift):
36749 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36750 test for const0_rtx.
36751 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36752 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36754 2017-03-09 Richard Biener <rguenther@suse.de>
36756 PR tree-optimization/79977
36757 * graphite-scop-detection.c (scop_detection::merge_sese):
36758 Handle the case of extra exits to blocks dominating the entry.
36760 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36762 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36765 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36767 PR rtl-optimization/79949
36768 * lra-constraints.c (process_alt_operands): Check memory when
36769 trying to predict a cycle. Print about the overall increase.
36771 2017-03-09 Richard Biener <rguenther@suse.de>
36773 PR middle-end/79971
36774 * gimple-expr.c (useless_type_conversion_p): Preserve
36775 TYPE_SATURATING for fixed-point types.
36777 2017-03-09 Richard Biener <rguenther@suse.de>
36780 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36781 alignment of BLKmode params.
36783 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36786 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36787 (VALL_NO_V2Q): Likewise.
36789 * config/aarch64/aarch64-simd.md
36790 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36792 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36793 VALL_NO_V2Q mode iterator.
36794 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36796 2017-03-09 Martin Liska <mliska@suse.cz>
36798 PR tree-optimization/79631
36799 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36800 tree_int_cst_sign_bit just for INTEGER constants.
36802 2017-03-09 Martin Liska <mliska@suse.cz>
36806 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36809 2017-03-09 Marek Polacek <polacek@redhat.com>
36812 * tree.c (inchash::add_expr): Handle TREE_VEC.
36814 2017-03-09 Martin Liska <mliska@suse.cz>
36817 (chkp_narrow_size_and_offset): New function.
36818 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36819 (void chkp_parse_bit_field_ref): New function.
36820 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36821 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36823 2017-03-09 Martin Liska <mliska@suse.cz>
36826 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36827 (chkp_find_bounds_1): Remove gcc_unreachable.
36829 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36832 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36833 BUILT_IN_SYNC*, determine the access type from the size suffix and
36834 always build a MEM_REF with that type. Handle forgotten
36835 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36838 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36839 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36840 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36841 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36842 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36843 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36844 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36845 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36846 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36847 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36848 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36849 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36850 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36851 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36852 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36853 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36854 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36855 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36856 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36857 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36858 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36859 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36860 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36861 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36862 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36863 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36864 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36865 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36866 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36867 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36868 definitions outside of __OPTIMIZE__ guarded section.
36871 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36872 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36873 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36874 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36877 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36879 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36880 ("vfenez<mode>"): Add missing constraints.
36882 2017-03-08 Martin Sebor <msebor@redhat.com>
36885 * config/nds32/nds32.c (nds32_option_override):
36886 Fix misspelled diagnostic.
36888 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36891 * gimplify.c (gimplify_omp_for): Replace index var in outer
36892 taskloop statement with an artificial variable and add
36893 OMP_CLAUSE_PRIVATE clause for it.
36895 2017-03-08 Richard Biener <rguenther@suse.de>
36897 PR tree-optimization/79955
36898 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36899 for accesses that are completely outside of the variable.
36901 2017-03-08 Andrew Haley <aph@redhat.com>
36903 PR tree-optimization/79943
36904 * tree-ssa-loop-split.c (compute_new_first_bound): When
36905 calculating the new upper bound, (END-BEG) should be added, not
36908 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36910 * config/avr/avr.md (setmemhi): Make sure match_dup
36911 operand number comes before match_scratch.
36913 2017-03-08 Richard Biener <rguenther@suse.de>
36915 PR tree-optimization/79920
36916 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36917 with ncopies == 1 to ...
36918 (vect_transform_slp_perm_load): ... here. Properly compute
36919 all element loads by iterating VF times over the group. Do
36920 not handle ncopies (computed in a broken way) in
36921 vect_create_mask_and_perm.
36923 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36926 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36927 is a uniform vector, use uniform_vector_p return value instead of
36928 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36930 2017-03-07 Marek Polacek <polacek@redhat.com>
36932 PR middle-end/79809
36933 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36934 (alloca_call_type): Likewise.
36936 2017-03-07 Martin Liska <mliska@suse.cz>
36938 * gcov.c (process_args): Put comment to correct location.
36940 2017-03-07 Martin Liska <mliska@suse.cz>
36942 PR middle-end/68270
36943 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36944 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36945 (chkp_narrow_bounds_for_field): Likewise.
36946 (chkp_parse_array_and_component_ref): Pass one more argument to
36949 2017-03-07 Richard Biener <rguenther@suse.de>
36951 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36954 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36956 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36957 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36959 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36962 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36963 to end of description.
36964 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36966 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36968 PR rtl-optimization/79901
36969 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36971 (*avx512f_<code><mode>3<mask_name>): ... this.
36972 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36973 iterator instead of VI8_AVX2_AVX512BW.
36975 PR rtl-optimization/79901
36976 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36977 min/max expander, expand it using expand_vec_cond_expr.
36980 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36983 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36986 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36987 to void * for PCH reasons.
36988 * dwarf2out.c (output_loc_operands, output_die): Cast
36989 v.val_vec.array to unsigned char *.
36991 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36994 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36997 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36999 PR rtl-optimization/79571
37000 * lra-constraints.c (process_alt_operands): Calculate static
37001 reject and subtract it from overall when only addresses will be
37004 2017-03-06 Julia Koval <julia.koval@intel.com>
37007 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37008 incoming stack boundary to 128 for 64-bit targets.
37010 2017-03-06 Richard Biener <rguenther@suse.de>
37012 PR tree-optimization/79894
37013 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37014 to NULL after folding it.
37016 2017-03-06 Richard Biener <rguenther@suse.de>
37018 PR tree-optimization/79824
37019 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37020 check disabling peeling for gaps.
37022 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37024 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37025 attributes): Document gettimeofday.
37027 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37029 * config/s390/s390.c (s390_option_override_internal): Set
37030 PARAM_MIN_VECT_LOOP_BOUND
37032 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37034 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37035 * config/s390/s390.md: Likewise.
37037 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37040 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37041 (<avx2_avx512>_perm<mode>): Rename to ...
37042 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37044 (<avx512>_perm<mode>_mask): Rename to ...
37045 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37047 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37048 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37049 instead of VI8F_256_512.
37050 (avx512f_perm<mode>): New define_expand.
37051 (avx512f_perm<mode>_mask): Likewise.
37052 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37053 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37055 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37057 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37058 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37060 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37062 2017-03-06 Martin Liska <mliska@suse.cz>
37065 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37066 when having a SSA NAME w/o VAR_DECL assigned to it.
37068 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37070 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37071 msa_dpsub_<su>_d): Fix MODE for vec_select.
37073 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37075 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37077 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37079 2017-03-06 Richard Biener <rguenther@suse.de>
37081 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37082 * plugin.c (register_plugin_info): Likewise.
37083 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37085 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37087 * config/i386/sse.md (sse_storehps, sse_storelps,
37088 avx_<castmode><avxsizesuffix>_<castmode>,
37089 avx512f_<castmode><avxsizesuffix>_<castmode>,
37090 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37091 in condition that at least one operand is not a MEM.
37093 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37095 PR middle-end/79805
37096 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37097 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37099 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37100 gimple_call_nothrow_p flag based on whether original builtin can throw.
37101 If it can, emit following stmts on the fallthrough edge.
37102 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37103 don't create new bb if inserting just debug stmts on the edge, try to
37104 insert them on the fallthru bb or just reset debug stmts.
37106 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37109 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37110 restore recog_data (including the operand rtxes inside it) around
37111 the call to get_insn_template.
37113 2017-03-03 Martin Sebor <msebor@redhat.com>
37115 PR tree-optimization/79699
37116 * context.c (context::~context): Free MPFR caches to avoid
37117 a memory leak on program exit.
37119 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37121 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37122 Use wide_int::ulow () instead of .elt (0).
37124 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37126 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37127 (*pushxf): Limit oF constraint to 32bit targets and add oC
37128 constraint for 64bit targets.
37129 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37130 (*pushdf): Change rmF constraint to rmC.
37132 2017-03-03 Martin Liska <mliska@suse.cz>
37134 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37135 Remove unused variable.
37137 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37140 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37141 is a memory operand, increase num_memory.
37142 (ix86_expand_args_builtin): Likewise.
37144 2017-03-03 Jan Hubicka <jh@suse.cz>
37147 * ipa-devirt.c (maybe_record_node): Properly handle
37148 __cxa_pure_virtual visibility.
37150 2017-03-03 Martin Liska <mliska@suse.cz>
37152 PR tree-optimization/79803
37153 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37155 (pass_loop_prefetch::execute): Disabled optimization if an
37156 assumption about L1 cache size is not met.
37158 2017-03-03 Martin Liska <mliska@suse.cz>
37160 PR rtl-optimization/79574
37161 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37162 (hash_scan_set): Likewise.
37163 (dump_hash_table): Likewise.
37164 (hoist_code): Likewise.
37166 2017-03-03 Richard Biener <rguenther@suse.de>
37168 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37169 in place of uhigh ().
37170 (fixed_convert_from_real): Likewise.
37172 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37175 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37177 2017-03-03 Richard Biener <rguenther@suse.de>
37179 PR middle-end/79818
37180 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37181 TYPE_OVERFLOW_UNDEFINED check.
37183 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37185 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37187 (vector_ae_<mode>_p): Likewise.
37188 (vector_nez_<mode>_p): Likewise.
37189 (vector_ne_v2di_p): Likewise.
37190 (vector_ae_v2di_p): Likewise.
37191 (vector_ne_<mode>_p): Likewise.
37192 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37194 (vsx_tsqrt<mode>2_fe): Likewise.
37196 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37199 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37201 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37203 PR rtl-optimization/79780
37204 * cprop.c (one_cprop_pass): When second and further conditional trap
37205 in a single basic block is turned into an unconditional trap, turn it
37206 into a deleted note to avoid RTL verification failures.
37208 2017-03-02 Richard Biener <rguenther@suse.de>
37210 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37212 2017-03-02 Richard Biener <rguenther@suse.de>
37214 PR tree-optimization/79345
37216 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37217 param and abort the walk, returning -1 if it is hit.
37218 (walk_aliased_vdefs): Take a limit param and pass it on.
37219 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37220 defaulting to 0 and return a signed int.
37221 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37222 (check_defs): New helper.
37223 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37224 about uninitialized memory.
37225 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37226 bogus uninitialized warning.
37227 (fixed_convert_from_real): Likewise.
37229 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37231 PR tree-optimization/66768
37232 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37233 iv_use if base object can't be determined.
37235 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37237 PR tree-optimization/79345
37238 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37239 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37240 (get_pattern_stats): Initialize it.
37241 * genemit.c (gen_expand): Verify match_scratch numbers come after
37242 match_operand/match_dup numbers.
37243 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37244 match_scratch numbers.
37245 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37247 * config/s390/s390.md (trunctdsd2): Likewise.
37249 2017-03-02 Richard Biener <rguenther@suse.de>
37251 * wide-int.h (wide_int_storage::operator=): Implement in terms
37254 2017-03-02 Richard Biener <rguenther@suse.de>
37256 PR tree-optimization/79777
37257 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37258 the to insert expression to sth existing.
37260 2017-03-01 Martin Sebor <msebor@redhat.com>
37262 PR middle-end/79692
37263 * gimple-ssa-sprintf.c
37264 (directive::known_width_and_precision): New function.
37265 (format_integer): Use it.
37266 (get_mpfr_format_length): Consider the full range of precision
37267 when computing %g output with the # flag. Set the likely byte
37268 count to 3 rather than 1 when precision is indeterminate.
37269 (format_floating): Correct the lower bound of precision.
37271 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37273 * doc/invoke.texi: Document default code model for 64-bit Linux.
37275 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37278 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37279 udiv rather than div since input pattern is unsigned.
37281 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37283 * config/i386/i386.c (print_reg): Warn for values of
37284 unsupported size in integer register.
37286 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37289 * config/rs6000/predicates.md (current_file_function_operand): Do
37290 not allow self calls to be local if the function is replaceable.
37292 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37295 * config/rs6000/altivec.h (vec_ctz and others): Change the
37296 preprocessor macro that controls conditional compilation from
37297 _ARCH_PWR9 to __POWER9_VECTOR__.
37298 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37299 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37300 control (instead of _ARCH_PWR9 control) so that template
37301 definition uses power9-specific function.
37302 (vec_any_eq): Likewise.
37303 (vec_all_ne): Change macro definition to use a power9-specific
37304 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37305 _ARCH_PWR9 control).
37306 (vec_any_eq) Likewise.
37307 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37308 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37309 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37310 support for xvcmpnedp instruction.
37311 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37312 macro expansion so that Power9 implementation of vec_all_ne does
37313 not use the AltiVec predicate framework.
37314 (VCMPNEH_P): Likewise.
37315 (VCMPNEW_P): Likewise.
37316 (VCMPNED_P): Likewise.
37317 (VCMPNEFP_P): Likewise.
37318 (VCMPNEDP_P): Likewise.
37319 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37320 implementation of vec_any_eq to not use AltiVec predicate
37322 (VCMPAEH_P): Likewise.
37323 (VCMPAEW_P): Likewise.
37324 (VCMPAED_P): Likewise.
37325 (VCMPAEFP_P): Likewise.
37326 (VCMPAEDP_P): Likewise.
37327 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37328 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37329 not use the AltiVec predicate framework.
37330 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37331 of vec_any_eq to not use AltiVec predicate framework.
37332 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37333 support for predefined __POWER9_VECTOR__ macro to indicate that
37334 Power9 instruction selection is enabled.
37335 (altivec_overloaded_builtins): Remove extraneous
37336 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37337 function argument types RS6000_BTI_bool_V16QI and
37338 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37339 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37340 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37341 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37342 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37343 Power9 for implementations of vec_cmpne. Change the signature for
37344 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37345 (representing vec_all_ne) to remove the previously described first
37346 argument of type RS6000_BTI_INTSI, as this was an artifact of
37347 reliance on the AltiVec predicate framework, which is no longer
37348 used in the implementation of these functions. Add
37349 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37350 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37351 since, unlike the AltiVec predicate framework implementation, we
37352 do not share function descriptors between vec_alle and vec_anyeq.
37353 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37354 set of modes that receive special treatment even when
37355 TARGET_P9_VECTOR is true. The special treatment emits code that
37356 does not depend on Power9 instructions.
37357 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37358 define_expand to not rely on AltiVec predicate framework.
37359 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37361 (vector_ne_v2di_p): Change this define_expand to not rely on
37362 AltiVec predicate framework.
37363 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37365 (vector_ne_<mode>_p): Change this define_expand to not rely on
37366 AltiVec predicate framework.
37367 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37369 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37370 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37371 define_insn pattern.
37372 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37373 define_insn pattern because the xvcmpne<VSs>. instruction is not
37375 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37376 instruction is not supported.
37378 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37380 * config/nvptx/nvptx.c: Include intl.h.
37382 2017-03-01 Martin Jambor <mjambor@suse.cz>
37385 * ipa-prop.h (ipa_bits): Removed field known.
37386 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37387 to pointers. Adjusted their comments to warn about their sharing.
37388 (ipcp_transformation_summary): Change bits to a vector of pointers.
37389 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37390 (ipa_get_ipa_bits_for_value): Declare.
37391 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37392 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37393 (ipa_bits_hash_table): Likewise.
37394 (ipa_vr_ggc_hash_traits): Likewise.
37395 (ipa_vr_hash_table): Likewise.
37396 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37397 being pointers and vr_known being removed.
37398 (ipa_set_jf_unknown): Likewise.
37399 (ipa_get_ipa_bits_for_value): New function.
37400 (ipa_set_jfunc_bits): Likewise.
37401 (ipa_get_value_range): New overloaded functions.
37402 (ipa_set_jfunc_vr): Likewise.
37403 (ipa_compute_jump_functions_for_edge): Use the above functions to
37404 construct bits and vr parts of jump functions.
37405 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37406 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37408 (ipcp_grow_transformations_if_necessary): Also allocate
37409 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37411 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37412 them. Fix too long lines.
37413 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37414 vr_known being removed.
37415 (ipa_read_jump_function): Use new setter functions to construct bits
37416 and vr parts of jump functions or set them to NULL.
37417 (write_ipcp_transformation_info): Adjust for bits being pointers.
37418 (read_ipcp_transformation_info): Likewise.
37419 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37421 Include gt-ipa-prop.h.
37422 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37424 (ipcp_store_bits_results): Likewise.
37425 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37426 Do not write to existing jump functions but use a temporary instead.
37428 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37431 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37432 attempt to use its first operand as BIT_FIELD_REF base.
37434 2017-03-01 Richard Biener <rguenther@suse.de>
37436 PR middle-end/79721
37437 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37438 interpolating formula in wrapping arithmetic.
37439 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37441 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37443 PR tree-optimization/79734
37444 * tree-vect-generic.c (expand_vector_condition): Optimize
37445 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37446 Handle VEC_COND_EXPR where comparison has different inner width from
37447 type's inner width.
37449 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37451 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37452 markup, and similar issues. Remove @opindex entries for things
37453 that aren't options. Add missing -mmpy-option entries.
37455 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37457 PR tree-optimization/79737
37458 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37459 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37460 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37461 instead of byte_size. Formatting fix.
37462 (shift_bytes_in_array_right): Formatting fix.
37464 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37467 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37468 condition on optimize for the leaf function test.
37470 2017-02-28 Martin Liska <mliska@suse.cz>
37473 * read-rtl-function.c (function_reader::handle_unknown_directive):
37474 Bail out when one uses -flto.
37476 2017-02-28 Martin Liska <mliska@suse.cz>
37478 * common.opt: Replace space with tabular for options of <number>
37480 * config/i386/i386.opt: Show <number> value for
37481 -mlarge-data-threshold.
37482 * opts.c (print_filtered_help): Do not display number in hexadecimal
37485 2017-02-28 Martin Liska <mliska@suse.cz>
37487 * common.opt: Fix --help=option -Q for options which are of
37490 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37492 * config/i386/i386.c (print_reg): Error out for values
37493 of 8-bit size in invalid integer register.
37495 2017-02-28 Martin Sebor <msebor@redhat.com>
37497 PR tree-optimization/79691
37498 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37500 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37503 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37504 gcc_unreachable with output_operand_lossage.
37506 2017-02-28 Richard Biener <rguenther@suse.de>
37508 PR tree-optimization/79740
37509 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37511 (visit_nary_op): Insert the nary into the hashtable if we
37512 pattern-matched sth.
37513 * tree-ssa-pre.c (eliminate_insert): Robustify.
37515 2017-02-28 Richard Biener <rguenther@suse.de>
37517 PR middle-end/79731
37518 * fold-const.c (decode_field_reference): Reject out-of-bound
37521 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37523 * config/i386/i386.c: Include intl.h.
37524 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37525 instead of just cond ? "..." : "...".
37526 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37527 * coverage.c (read_counts_file): Likewise.
37528 * omp-offload.c: Include intl.h.
37529 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37530 of just cond ? "..." : "...".
37531 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37532 of just cond ? "..." : "...".
37534 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37537 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37539 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37540 'tune for' CPU name.
37541 * config/arm/arm-cpu-data.h: Regenerated.
37543 2017-02-28 Richard Biener <rguenther@suse.de>
37545 PR tree-optimization/79732
37546 * tree-inline.c (expand_call_inline): Do not shadow var.
37548 2017-02-28 Richard Biener <rguenther@suse.de>
37550 PR tree-optimization/79723
37551 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37552 address-space properly.
37554 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37556 * doc/optinfo.texi (Optimization groups): Fix option used for
37558 * doc/invoke.texi (-fopt-info): Document "omp".
37559 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37560 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37561 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37562 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37563 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37565 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37567 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37568 OPTGROUP_OMP with "omp".
37570 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37573 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37574 for arithmetic shift of unsigned V2DI.
37576 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37578 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37579 arc/linux.h headers.
37580 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37581 (LINK_SPEC): Likewise.
37582 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37583 (EXTRA_SPECS): Likewise.
37584 (STARTFILE_SPEC): Likewise.
37585 (ENDFILE_SPEC): Likewise.
37586 (LIB_SPEC): Likewise.
37587 (TARGET_SDATA_DEFAULT): Likewise.
37588 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37589 (MULTILIB_DEFAULTS): Likewise.
37590 (DWARF2_UNWIND_INFO): Likewise.
37591 * config/arc/big.h: New file.
37592 * config/arc/elf.h: Likewise.
37593 * config/arc/linux.h: Likewise.
37594 * config/arc/t-uClibc: Remove.
37596 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37598 PR tree-optimization/77536
37599 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37600 (tree_transform_and_unroll_loop): Use above function to compute the
37601 estimated niter of unrolled loop and use it when scaling profile.
37602 Also use count info rather than frequency if it's non-zero.
37603 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37604 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37605 (vect_transform_loop): Call above function.
37607 2017-02-27 Richard Biener <rguenther@suse.de>
37609 PR tree-optimization/45397
37610 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37611 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37612 (visit_nary_op): Add pattern matching for CSEing sign-changed
37613 or truncated operations with wider ones.
37615 2017-02-27 Richard Biener <rguenther@suse.de>
37617 PR tree-optimization/79690
37618 * tree-vect-stmts.c (vectorizable_store): Use vector type
37619 built from the DR with address-space.
37621 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37623 * doc/invoke.texi (Optimize Options): Refine the description
37624 of asan-use-after-return.
37626 2017-02-25 Alan Modra <amodra@gmail.com>
37628 PR rtl-optimization/79584
37629 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37630 base, not ad->base_term, the reg within base. Remove assertion
37631 that ad->base == ad->base_term. Replace gen_int_mode using
37632 bogus mode with const0_rtx.
37634 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37636 PR middle-end/79396
37637 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37638 FMA_EXPR like tcc_binary or tcc_unary.
37640 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37643 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37645 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37646 (output_loc_operands): Handle DW_OP_call_ref and
37647 DW_OP_GNU_variable_value.
37648 (struct variable_value_struct): New type.
37649 (struct variable_value_hasher): Likewise.
37650 (variable_value_hash): New variable.
37651 (string_types): Remove.
37652 (copy_loc_descr): New function.
37653 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37654 (prepend_loc_descr_to_each): New function.
37655 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37656 instead of add_loc_descr_to_each if the first argument is single
37657 location list and the second has multiple.
37658 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37659 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37660 when looking for variable value which doesn't have other location info.
37661 (loc_list_from_tree): Formatting fix.
37662 (gen_array_type_die): Simplify DW_AT_string_length handling.
37663 (adjust_string_types): Remove.
37664 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37665 string_types. Call resolve_variable_values.
37666 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37667 (resolve_addr_in_expr): Likewise. Add A argument.
37668 (copy_deref_exprloc): Remove deref argument. Adjust for the
37669 original expression being DW_OP_GNU_variable_value with optionally
37670 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37671 optionally after it.
37672 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37673 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37674 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37675 (variable_value_hasher::hash, variable_value_hasher::equal): New
37677 (resolve_variable_value_in_expr, resolve_variable_value,
37678 resolve_variable_values, note_variable_value_in_expr,
37679 note_variable_value): New functions.
37680 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37683 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37686 * opts.h (handle_generated_option): Add GENERATED_P argument.
37687 * opts-common.c (handle_option): Adjust function comment.
37688 (handle_generated_option): Add GENERATED_P argument, pass it to
37690 (control_warning_option): Pass false to handle_generated_option
37692 * opts.c (maybe_default_option): Pass true to handle_generated_option
37694 * optc-gen.awk: Likewise.
37696 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37698 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37699 a REG, look at the REG it is a SUBREG of.
37700 (splitter for cmpeqsi_t): Ditto.
37702 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37704 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37705 the special USEs with the pattern of the insn, not the insn itself.
37707 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37710 * doc/invoke.texi: Document -mload-store-pairs.
37712 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37713 Sandra Loosemore <sandra@codesourcery.com>
37715 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37716 argument isn't a CONST_INT.
37717 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37718 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37719 (nios2_validate_compare): Bypass alternate compare logic if *op2
37720 is not a CONST_INT.
37721 (ldstwm_operation_p): Return false if first_base is not a REG or
37722 if first_offset is not a CONST_INT.
37724 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37726 * config/cris/cris.md: Use correct operand in a define_peephole2.
37728 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37730 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37732 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37734 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37735 this_insn if it is an INSN or JUMP_INSN.
37736 (force_offsettable): Look at base, not at addr.
37737 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37738 on things that aren't necessarily CONST_INTs.
37740 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37742 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37743 -mfpmath=sse is the default also for x86-32 targets with SSE2
37744 instruction set when @option{-ffast-math} is enabled
37746 2017-02-24 Jeff Law <law@redhat.com>
37748 PR rtl-optimizatoin/79286
37749 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37752 2017-02-24 Richard Biener <rguenther@suse.de>
37754 PR tree-optimization/79389
37755 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37758 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37760 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37761 function comment to reflect reality.
37762 (loop_exits_before_overflow): Fix typo in function description.
37764 2017-02-24 Richard Biener <rguenther@suse.de>
37766 PR tree-optimization/79389
37767 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37768 properly that a threading opportunity exists. Detect conditional
37769 copy/constant propagation opportunities.
37771 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37773 * config/visium/visium.md (type): Add trap.
37774 (b): New mode attribute.
37775 (*btst): Rename into...
37776 (*btst<mode>): ...this and adjust.
37777 (*cbranchsi4_btst_insn): Rename into...
37778 (*cbranch<mode>4_btst_insn): ...this and adjust.
37779 (trap): New define_insn.
37781 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37783 PR tree-optimization/79389
37784 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37785 (noce_reversed_cond_code): New function.
37786 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37787 reversed_comparison_code. Formatting fix.
37788 (noce_try_store_flag): Test rev_cond != NULL in addition to
37789 reversed_comparison_code.
37790 (noce_try_store_flag_constants): Likewise.
37791 (noce_try_store_flag_mask): Likewise.
37792 (noce_try_addcc): Use rev_cond if non-NULL instead of
37793 reversed_comparison_code.
37794 (noce_try_cmove_arith): Likewise. Formatting fixes.
37795 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37796 (noce_find_if_block): Initialize rev_cond.
37797 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37798 instead of false as last argument never attempt to reverse it
37801 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37803 PR tree-optimization/79663
37804 * tree-predcom.c (combine_chains): Process refs in reverse order
37805 only for ZERO length chains, and add explaining comment.
37807 2017-02-23 Jeff Law <law@redhat.com>
37809 PR tree-optimization/79578
37810 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37811 in call to operand_equal_p.
37813 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37816 * config/i386/cpuid.h: Fix another undefined behavior.
37818 2017-02-23 Richard Biener <rguenther@suse.de>
37820 PR tree-optimization/79683
37821 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37822 vector types for data-refs.
37824 2017-02-23 Martin Liska <mliska@suse.cz>
37826 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37828 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37830 PR middle-end/79665
37831 * internal-fn.c (get_range_pos_neg): Moved to ...
37832 * tree.c (get_range_pos_neg): ... here. No longer static.
37833 * tree.h (get_range_pos_neg): New prototype.
37834 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37835 are known to be in between 0 and signed maximum inclusive, try to
37836 expand both unsigned and signed divmod and use the cheaper one from
37839 2017-02-22 Jeff Law <law@redhat.com>
37841 PR tree-optimization/79578
37842 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37843 to compare base operands.
37845 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37848 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37849 gpc_reg_operand instead of fpr_reg_operand.
37851 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37853 * config/mips/mips.c (mips_return_in_memory): Force FP
37854 vector types to be returned in memory for o32 ABI.
37856 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37858 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37859 instead of DW_TAG_member for static data member declarations and don't
37860 set no_linkage_name for static inline data members.
37861 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37864 2017-02-22 Martin Liska <mliska@suse.cz>
37866 * doc/invoke.texi: Replace inequality signs with square brackets
37869 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37872 * lra-constraints.c (simplify_operand_subreg): Handle
37873 WORD_REGISTER_OPERATIONS targets.
37875 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37878 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37879 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37880 elimination by swapping fld*.
37882 2017-02-22 Richard Biener <rguenther@suse.de>
37884 PR tree-optimization/79673
37885 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37886 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37887 irrelevant address-space qualifiers and avoiding a
37888 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37890 2017-02-22 Richard Biener <rguenther@suse.de>
37892 PR tree-optimization/79666
37893 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37894 to not symbolically negate if that may introduce undefined
37897 2017-02-22 Martin Liska <mliska@suse.cz>
37900 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37901 * data-streamer-out.c (streamer_write_gcov_count_stream):
37903 * value-prof.c (stream_out_histogram_value): Make assert more
37904 precise based on type of counter.
37906 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37909 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37910 Use nonimmediate_operand instead of memory_operand for operand 1.
37911 (float-extend standard_x87sse_constant_load splitter): Ditto.
37913 2017-02-21 Jeff Law <law@redhat.com>
37915 PR tree-optimization/79621
37916 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37917 blocks with edges to themselves.
37919 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37922 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37923 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37924 Use gimple_call_builtin_p.
37927 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37928 on temporarily removed DEBUG_INSNs.
37930 PR tree-optimization/79649
37931 * tree-loop-distribution.c (classify_partition): Give up on
37932 non-generic address space loads/stores.
37934 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37936 * doc/loop.texi (Loop manipulation): Remove nonexistent
37937 tree_ssa_loop_version from the documentation.
37938 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37940 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37943 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37944 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37945 * config/rs6000/rs6000.c: Include except.h.
37946 (rs6000_expand_split_stack_prologue): Call
37947 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37949 2017-02-21 Martin Jambor <mjambor@suse.cz>
37952 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37953 have been analyzed.
37955 2017-02-21 Martin Jambor <mjambor@suse.cz>
37957 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37958 for backward compatibility only.
37959 * doc/invoke.texi (Option Summary): Remove all references to
37960 -fipa-cp-alignment.
37962 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37966 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37968 * lra-constraints.c (curr_insn_transform): Handle
37969 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37971 2017-02-21 Martin Liska <mliska@suse.cz>
37973 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37975 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37977 PR translation/79638
37978 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37980 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37983 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37984 (arm_function_ok_for_sibcall): Return false for an indirect call by
37985 descriptor if all the argument registers are used.
37986 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37987 alignment of the function.
37989 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37991 PR tree-optimization/61441
37992 * simplify-rtx.c (simplify_const_unary_operation): For
37993 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37994 the sNaN unmodified.
37996 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37998 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37999 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
38000 instead of SYSTEM_HEADER_DIR.
38002 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
38003 Martin Liška <mliska@suse.cz>
38005 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
38006 Fix typos and grammar, use active voice, and clarify.
38008 2017-02-20 Marek Polacek <polacek@redhat.com>
38010 PR middle-end/79537
38011 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38014 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38016 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38019 * config/i386/i386.c (ix86_expand_builtin): Handle
38020 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38021 ix86_builtins_isa[fcode].isa as a requirement of those
38022 flags and any other flag in the bitmask.
38023 (ix86_init_mmx_sse_builtins): Use 0 instead of
38024 ~OPTION_MASK_ISA_64BIT as mask.
38025 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38026 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38027 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38028 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38030 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38033 * lra-constraints.c (split_reg): Check requested split mode
38034 is supported by the register.
38036 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38038 * lra-constraints.c (simplify_operand_subreg): Remove early
38041 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38044 * lra-constraints.c (curr_insn_transform): Tighten condition
38045 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38047 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38050 * lra-constraints.c (curr_insn_transform): Handle
38051 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38053 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38056 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38058 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38060 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38063 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38066 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38068 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38071 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38073 * final.c (last_columnnum, override_columnnum): New variables.
38074 (final_start_function): Set last_columnnum, pass it to begin_prologue
38075 hook and pass 0 to dwarf2out_begin_prologue.
38076 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38077 to source_line debug hook.
38078 (notice_source_line): Compute last_columnnum and for debug_column_info
38079 return true on column changes.
38080 * debug.h (struct gcc_debug_hooks): Add column argument to
38081 source_line and begin_prologue hooks.
38082 (debug_nothing_int_charstar_int_bool): Remove prototype.
38083 (debug_nothing_int_int_charstar,
38084 debug_nothing_int_int_charstar_int_bool): New prototypes.
38085 (dwarf2out_begin_prologue): Add column argument.
38086 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38087 begin_prologue hooks.
38088 (debug_nothing_int_charstar_int_bool): Remove.
38089 (debug_nothing_int_int_charstar,
38090 debug_nothing_int_int_charstar_int_bool): New functions.
38091 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38092 through to dwarf2out_source_line.
38093 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38094 (dwarf2out_source_line): Add column argument, emit it if requested.
38095 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38097 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38098 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38099 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38100 through to dwarf2out_begin_prologue.
38101 (vmsdbgout_source_line): Add column argument, pass it through to
38102 dwarf2out_source_line.
38103 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38104 dbxout_source_line caller.
38105 (dbxout_source_line): Add column argument.
38107 * common.opt (gno-column-info, gcolumn-info): New options.
38108 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38109 (check_die): Also test for multiple DW_AT_decl_column attributes.
38110 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38111 DW_AT_decl_column if requested.
38112 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38114 (gen_variable_die): Likewise.
38115 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38116 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38119 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38120 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38121 (ix86_handle_option): Handle OPT_m3dnowa.
38122 * doc/invoke.texi (-m3dnowa): Document.
38123 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38124 -m3dnowa instead of -m3dnow -march=athlon.
38127 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38128 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38130 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38133 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38134 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38135 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38136 generator for vsx_xxpermdi_<mode>_be.
38137 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38138 force big-endian semantics.
38139 (vsx_xxpermdi_<mode>_be): New define_expand with same
38140 implementation as previous version of vsx_xxpermdi_<mode>.
38142 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38144 PR tree-optimization/79327
38145 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38146 variable, its initialization and use.
38148 2017-02-17 Julia Koval <julia.koval@intel.com>
38150 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38151 (OPTION_MASK_ISA_PKU_UNSET): New.
38152 (ix86_handle_option): Handle -mrdpid.
38153 * config/i386/cpuid.h (bit_RDPID): New.
38154 * config/i386/driver-i386.c (host_detect_local_cpu):
38155 Detect RDPID feature.
38156 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38157 * config/i386/i386-c.c (ix86_target_macros_internal):
38159 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38160 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38161 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38162 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38163 * config/i386/i386.md (define_insn "rdpid"): New.
38164 * config/i386/i386.opt Add -mrdpid.
38165 * config/i386/immintrin.h (_rdpid_u32): New.
38167 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38169 PR rtl-optimization/79541
38170 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38171 instead of transforming it into USE.
38173 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38175 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38176 If HONOR_SNANS (SFmode) force the input to a register.
38177 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38178 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38179 an frsp or similar insn.
38181 2017-02-17 Martin Liska <mliska@suse.cz>
38183 PR rtl-optimization/79577
38184 * params.def (selsched-max-sched-times): Increase minimum to 1.
38186 2017-02-17 Martin Liska <mliska@suse.cz>
38188 PR rtl-optimization/79574
38189 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38191 2017-02-17 Martin Liska <mliska@suse.cz>
38193 PR tree-optimization/79529
38194 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38195 ssa_defined_default_def_p to handle cases which are implicitly
38197 * tree-ssa.c (ssa_defined_default_def_p): New function.
38198 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38199 which are implicitly defined.
38200 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38202 2017-02-17 Richard Biener <rguenther@suse.de>
38204 PR middle-end/79576
38205 * params.def (max-ssa-name-query-depth): Limit to 10.
38207 2017-02-17 Richard Biener <rguenther@suse.de>
38209 PR tree-optimization/79552
38210 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38213 2017-02-17 Richard Biener <rguenther@suse.de>
38216 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38218 2017-02-17 Marek Polacek <polacek@redhat.com>
38220 PR middle-end/79536
38221 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38222 (fold_negate_expr): New wrapper.
38224 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38226 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38227 Correct terminology and de-emphasize pre-standard behavior.
38229 2017-02-16 Alan Modra <amodra@gmail.com>
38231 PR rtl-optimization/79286
38232 * ira.c (def_dominates_uses): New function.
38233 (update_equiv_regs): Don't create an equivalence for insns that
38234 may trap where the register def does not dominate the use.
38236 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38238 PR rtl-optimization/78127
38239 * lra.c (lra): Call lra_eliminate before finish the loop after
38242 2017-02-16 Richard Biener <rguenther@suse.de>
38244 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38246 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38247 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38248 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38249 (isl_val_int_from_wi): New function.
38250 (extract_affine_gmp): Rename to ...
38251 (extract_affine_wi): ... this, take a widest_int.
38252 (extract_affine_int): Just wrap extract_affine_wi.
38253 (add_param_constraints): Use isl_val_int_from_wi.
38254 (add_loop_constraints): Likewise, and extract_affine_wi.
38256 2017-02-15 Jeff Law <law@redhat.com>
38258 PR middle-end/79521
38259 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38260 ira_init_register_move_cost_if_necessary.
38262 2017-02-15 Martin Sebor <msebor@redhat.com>
38264 PR middle-end/32003
38265 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38266 removed in a prior commit.
38268 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38270 PR tree-optimization/79347
38271 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38272 counters during peeling.
38274 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38276 * Makefile.in (site.exp): Remove "set ISLVER".
38278 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38281 * real.c (real_from_integer): Call real_convert even for decimal.
38283 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38286 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38288 2017-02-14 Andrew Pinski <apinski@cavium.com>
38290 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38291 cores and change the partno/implementer to be correct.
38292 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38293 the 'B" as the implementer.
38294 * config/aarch64/aarch64-tune.md: Regenerate.
38296 2017-02-14 Carl Love <cel@us.ibm.com>
38298 * config/rs6000/rs6000.c: Add case statement entry to make the
38299 xvcvuxdsp built-in argument unsigned.
38300 * config/rs6000/vsx.md: Fix the source and return operand types so they
38301 match the instruction definitions from the ISA document. Fix typo
38302 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38305 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38308 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38309 member early_clobber_alts.
38310 * lra-lives.c (reg_early_clobber_p): New.
38311 (process_bb_lives): Use it.
38312 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38313 (debug_operand_data): Initialize early_clobber_alts.
38314 (setup_operand_alternative): Set up early_clobber_alts.
38315 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38316 alternatives to new_insn_reg.
38317 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38319 (lra_update_insn_regno_info): Pass the new arg.
38321 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38323 PR middle-end/79505
38324 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38325 (new_oacc_loop_raw): Don't clear already cleared fields.
38328 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38329 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38330 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38332 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38335 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38337 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38340 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38341 the extra instruction to the right place to store 128-bit constant
38344 2017-02-14 Martin Sebor <msebor@redhat.com>
38346 PR middle-end/79448
38347 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38348 warning for strings of unknown length.
38350 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38352 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38354 2017-02-14 Jeff Law <law@redhat.com>
38357 * ira-costs.c (scan_one_insn): Initialize register move costs
38358 for pseudos seen in USE/CLOBBER insns.
38360 PR tree-optimization/79095
38361 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38362 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38363 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38364 if the operands are known to be not equal, then the resulting range
38366 (intersect_ranges): If the new range is ~[0,0] and the old range is
38367 wide, then prefer ~[0,0].
38368 * tree-vrp.c (overflow_comparison_p_1): New function.
38369 (overflow_comparison_p): New function.
38370 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38371 if NAME is used in an overflow test.
38372 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38373 overflow check that can be expressed as an equality test, then adjust
38374 ops to be that equality test.
38376 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38378 * config/s390/s390-builtin-types.def: Remove flags argument.
38379 * config/s390/s390.c (s390_init_builtins): Likewise.
38381 2017-02-14 Martin Liska <mliska@suse.cz>
38383 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38384 vector. Fix trailing white spaces.
38386 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38388 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38391 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38393 PR rtl-optimization/68664
38394 * config/arm/arm.c (arm_sched_can_speculate_insn):
38395 New function. Declare prototype.
38396 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38398 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38400 PR rtl-optimization/68664
38401 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38403 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38405 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38407 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38408 max skip bytes for function, loop and jump.
38410 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38412 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38413 ABS_EXPR for gimple dump.
38415 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38418 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38420 PR tree-optimization/79408
38421 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38422 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38423 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38424 also if rhs1 is INTEGER_CST.
38426 2017-02-14 Richard Biener <rguenther@suse.de>
38428 PR middle-end/79432
38429 * tree-into-ssa.c (insert_phi_nodes): When the function can
38430 have abnormal edges rewrite SSA names with broken use-def
38431 dominance out of SSA and register them for PHI insertion.
38433 2017-02-13 Martin Sebor <msebor@redhat.com>
38435 PR middle-end/79496
38436 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38437 clearing info.nowrite flag when snprintf size argument is a range.
38439 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38441 * cprop.c (cprop_jump): Add missing space in string literal.
38442 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38443 (get_constraint_for_component_ref): Likewise.
38444 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38445 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38446 * lra-constraints.c (process_alt_operands): Likewise.
38447 * ipa-inline.c (inline_small_functions): Likewise.
38448 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38449 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38450 * trans-mem.c (diagnose_tm_1_op): Likewise.
38451 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38452 (grid_parallel_clauses_gridifiable): Likewise.
38454 * config/nvptx/mkoffload.c (process): Add space in between
38457 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38458 "MOD4_SSE_REGS" and "ALL_REGS".
38460 * spellcheck.c (test_data): Add , in between "foo" and "food".
38462 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38465 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38466 boundary crossing check and subsequent code generation agree.
38468 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38470 * config/aarch64/aarch64.c (has_memory_op): Delete.
38471 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38474 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38476 PR rtl-optimization/79388
38477 PR rtl-optimization/79450
38478 * combine.c (distribute_notes): When removing TEM_INSN for which
38479 corresponding dest has last value recorded, invalidate that last
38482 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38484 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38485 of explicit '@'. Add missing assembly comment marker on branch costs
38488 2017-02-13 Nathan Sidwell <nathan@acm.org>
38490 * gengtype-lex.l (<in_struct>): Add '/'.
38492 2017-02-13 Martin Liska <mliska@suse.cz>
38495 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38497 2017-02-13 Richard Biener <rguenther@suse.de>
38499 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38501 * configure: Re-generate.
38502 * config.in: Likewise.
38503 * graphite-dependences.c: Simplify as if
38504 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38505 * graphite-isl-ast-to-gimple.c: Likewise.
38506 * graphite-optimize-isl.c: Likewise.
38507 * graphite-poly.c: Likewise.
38508 * graphite-sese-to-poly.c: Likewise.
38509 * graphite.h: Likewise.
38510 * toplev.c: Include isl/version.h and use isl_version () for
38511 printing the ISL version.
38512 * doc/install.texi: Update ISL requirement.
38514 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38516 * doc/standards.texi (Standards): Update reference to
38519 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38521 * doc/extend.texi (Named Address Spaces): sourceware.org now
38523 * doc/install.texi (Binaries): Ditto.
38526 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38528 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38529 standard terminology "stringize"/"stringizing" throughout.
38530 * doc/cppinternals.texi: Likewise.
38532 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38534 * doc/extend.texi: Fix some spelling mistakes and typos.
38535 * doc/invoke.texi: Likewise.
38537 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38540 * params.def (inline-min-speedup) Change from 10 to 8.
38542 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38544 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38547 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38550 * ipa-inline-analysis.c (get_minimal_bb): New function.
38551 (record_modified): Use it.
38552 (remap_edge_change_prob): Handle also ancestor functions.
38554 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38556 * doc/contrib.texi (Contributors): Remove broken link into
38557 the Mauve CVS repository.
38559 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38561 PR middle-end/79454
38562 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38563 result computation whenever lhs doesn't have vector mode, not
38564 just when it has BLKmode.
38566 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38568 * doc/makefile.texi (profiledbootstrap): Refer to the
38569 installation instructions only in textual form.
38571 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38574 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38576 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38578 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38579 (Specific): Update mingw-w64 reference.
38581 (Specific): Remove broken link to Renesas RX processor.
38583 2017-02-10 Richard Biener <rguenther@suse.de>
38585 * toplev.c (process_options): Do not mention obsolete graphite
38586 options when printing sorry message about missing graphite support.
38587 Mention -floop-nest-optimize.
38589 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38591 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38592 (vtst_p16): Likewise.
38593 (vtstq_p8): Likewise.
38594 (vtstq_p16): Likewise.
38596 (vtstq_p64): Likewise.
38597 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38598 (vset_lane_p64): New.
38599 (vsetq_lane_p64): New.
38601 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38603 PR tree-optimization/79411
38604 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38605 stmt operands are SSA_NAMEs used in abnormal phis.
38606 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38609 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38612 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38615 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38617 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38619 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38621 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38625 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38626 not arbitrary TREE_CONSTANT.
38629 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38630 "omp declare target link" attribute unless is_global_var.
38631 * omp-offload.c (find_link_var_op): Likewise.
38633 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38634 Chung-Lin Tang <cltang@codesourcery.com>
38636 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38638 (gimplify_adjust_omp_clauses): Don't delete TILE.
38639 (gimplify_omp_for): Deal with TILE.
38640 * internal-fn.c (expand_GOACC_TILE): New function.
38641 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38643 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38644 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38646 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38647 avoid DIV for outermost collapse var.
38648 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38649 Remove out of date comments, fix whitespace.
38650 * omp-general.c (omp_extract_for_data): Deal with tiling.
38651 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38652 adjust OLF_DIM_BASE value.
38653 (struct omp_for_data): Add tiling field.
38654 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38655 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38656 for auto loops. Remove default auto determining, moved to
38657 oacc_loop_fixed_partitions.
38658 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38659 stmts, add e_mask field.
38660 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38661 (oacc_thread_numbers): Use oacc_dim_call.
38662 (oacc_xform_tile): New.
38663 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38664 (finish_oacc_loop): Adjust for ifns vector.
38665 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38666 add case for GOACC_TILE fns.
38667 (oacc_loop_xform_loop): Delete.
38668 (oacc_loop_process): Iterate over call list directly, and add
38669 handling for GOACC_TILE fns.
38670 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38672 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38673 vector partitioning to outer loops. Assign 2 partitions to loops
38674 when available. Add TILE handling.
38675 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38676 (execite_oacc_device_lower): Process GOACC_TILE fns,
38677 ignore unknown specs.
38678 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38679 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38680 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38682 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38684 * configure.ac (ACX_BUGURL): Update.
38685 * configure: Regenerate.
38687 2017-02-09 Richard Biener <rguenther@suse.de>
38689 PR tree-optimization/69823
38690 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38691 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38693 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38695 * config/arc/arc-c.def: Add __NPS400__ definition.
38696 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38697 (TARGET_NPS400): Define.
38699 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38701 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38703 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38704 pointer, arch_info.
38705 (arc_cpu_types): Fill the arch_info field with a pointer into the
38706 arc_arch_types table.
38707 (arc_selected_cpu): Declare.
38708 * config/arc/arc.c (arc_selected_cpu): Make global.
38709 (arc_selected_arch): Delete.
38710 (arc_base_cpu): Delete.
38711 (arc_override_options): Remove references to deleted variables,
38712 update access to arch information.
38713 (ARC_OPT): Update access to arch information.
38714 (ARC_OPTX): Likewise.
38715 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38716 (TARGET_ARC600): Update access to arch information.
38717 (TARGET_ARC601): Likewise.
38718 (TARGET_ARC700): Likewise.
38719 (TARGET_EM): Likewise.
38720 (TARGET_HS): Likewise.
38721 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38724 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38727 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38728 condition/operands for integer GE/LE/GEU/LEU operations.
38730 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38732 PR translation/79397
38733 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38736 2017-02-08 Martin Jambor <mjambor@suse.cz>
38739 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38740 whether allocation happened.
38741 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38742 nothing was allocated.
38744 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38746 PR tree-optimization/79408
38747 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38748 constant, but SSA_NAME with a known integer range, use the minimum
38749 of that range instead of op1 to determine if modulo can be replaced
38750 with its first operand.
38752 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38754 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38756 2017-02-08 Richard Biener <rguenther@suse.de>
38758 PR tree-optimization/71824
38759 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38760 Check all loops contained in the merged region.
38762 2017-02-07 Andrew Pinski <apinski@cavium.com>
38764 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38766 2017-02-07 Andrew Pinski <apinski@cavium.com>
38768 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38769 (thunderxt88): Likewise.
38770 (thunderxt81): Disable LSE and change v8.1 to v8.
38771 (thunderxt83): Likewise.
38773 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38774 Richard Biener <rguenther@suse.de>
38776 PR middle-end/79399
38777 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38778 type from int to size_t.
38779 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38781 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38783 PR rtl-optimization/79386
38784 * cprop.c (bypass_conditional_jumps): Initialize
38785 bypass_last_basic_block already before splitting bbs after
38786 unconditional traps...
38787 (bypass_conditional_jumps): ... rather than here.
38790 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38791 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38792 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38793 fix -masm=intel patterns.
38795 2017-02-07 Richard Biener <rguenther@suse.de>
38797 PR tree-optimization/79256
38798 PR middle-end/79278
38799 * builtins.c (get_object_alignment_2): Use min_align_of_type
38800 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38801 and ADJUST_FIELD_ALIGN.
38803 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38805 * doc/tm.texi: Regenerate.
38806 * stor-layout.c (layout_decl): Adjust.
38807 (update_alignment_for_field): Likewise.
38808 (place_field): Likewise.
38809 (min_align_of_type): Likewise.
38810 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38811 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38812 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38813 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38814 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38815 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38816 * config/i386/i386.c (x86_field_alignment): Likewise.
38817 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38818 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38819 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38820 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38821 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38822 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38826 2017-01-30 Richard Biener <rguenther@suse.de>
38828 PR tree-optimization/79256
38829 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38830 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38833 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38835 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38836 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38837 builtins to SImode and emit a zero-extend, if necessary.
38839 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38841 * docs/invoke.texi (RISC-V Options): Alphabetize.
38843 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38845 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38848 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38850 * config/riscv/riscv.c: New file.
38851 * common/config/riscv/riscv-common.c: Likewise.
38852 * config.gcc: Likewise.
38853 * config/riscv/constraints.md: Likewise.
38854 * config/riscv/elf.h: Likewise.
38855 * config/riscv/generic.md: Likewise.
38856 * config/riscv/linux.h: Likewise.
38857 * config/riscv/multilib-generator: Likewise.
38858 * config/riscv/peephole.md: Likewise.
38859 * config/riscv/pic.md: Likewise.
38860 * config/riscv/predicates.md: Likewise.
38861 * config/riscv/riscv-builtins.c: Likewise.
38862 * config/riscv/riscv-c.c: Likewise.
38863 * config/riscv/riscv-ftypes.def: Likewise.
38864 * config/riscv/riscv-modes.def: Likewise.
38865 * config/riscv/riscv-opts.h: Likewise.
38866 * config/riscv/riscv-protos.h: Likewise.
38867 * config/riscv/riscv.h: Likewise.
38868 * config/riscv/riscv.md: Likewise.
38869 * config/riscv/riscv.opt: Likewise.
38870 * config/riscv/sync.md: Likewise.
38871 * config/riscv/t-elf-multilib: Likewise.
38872 * config/riscv/t-linux: Likewise.
38873 * config/riscv/t-linux-multilib: Likewise.
38874 * config/riscv/t-riscv: Likewise.
38875 * configure.ac: Likewise.
38876 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38877 Waterman as RISC-V maintainers.
38878 * doc/install.texi: Add RISC-V entries.
38879 * doc/invoke.texi: Add RISC-V options section.
38880 * doc/md.texi: Add RISC-V constraints section.
38881 * configure: Regenerated.
38883 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38886 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38887 false values to be constant vectors with all 0 or all 1 bits set.
38888 (vcondu<mode><mode>): Likewise.
38889 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38891 (fpmask_comparison_operator): Update comment.
38892 (vecint_comparison_operator): New predicate.
38893 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38894 vector conditionals when the true and false values are constant
38895 vectors with all 0 bits or all 1 bits set.
38897 2017-02-06 Martin Sebor <msebor@redhat.com>
38899 PR tree-optimization/79376
38900 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38902 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38904 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38905 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38906 to simplify split condition.
38908 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38910 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38911 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38914 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38916 PR rtl-optimization/68664
38917 * target.def (can_speculate_insn): New hook.
38918 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38919 * doc/tm.texi: Regenerate.
38920 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38921 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38922 (rs6000_sched_can_speculate_insn): New function.
38924 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38926 PR tree-optimization/79284
38927 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38928 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38929 vectorizable_mask_load_store, vectorizable_operation,
38930 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38931 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38932 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38933 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38934 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38935 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38936 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38937 is_gimple_assign (stmt). Replace another such test with
38938 is_gimple_assign (stmt).
38940 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38943 * config/avr/avr.c (rtl-iter.h): Include it.
38944 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38945 (avr_legitimate_combined_insn): ...and implementation.
38947 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38949 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38950 * config/s390/s390.c (s390_const_operand_ok)
38951 (s390_canonicalize_comparison, s390_extract_part)
38952 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38953 (s390_contiguous_bitmask_p, s390_rtx_costs)
38954 (legitimize_pic_address): Likewise.
38955 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38956 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38957 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38958 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38959 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38961 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38963 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38964 REGNO($0) == REGNO($1).
38966 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38968 * config/s390/linux.h(SIZE_TYPE): Add comment.
38970 2017-02-06 Julian Brown <julian@codesourcery.com>
38971 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38972 Virendra Pathak <virendra.pathak@broadcom.com>
38974 * config/aarch64/aarch64-cores.def: Change the scheduler
38976 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38977 * config/aarch64/thunderx2t99.md: New file.
38979 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38981 * doc/standards.texi (Go Language): Update link to language
38984 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38986 * tree-eh.c (lower_resx): Sanitize profile.
38987 (cleanup_empty_eh_move_lp): Likewise.
38989 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38992 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38994 * cfgloopmanip.h (loop_version): Update prototype.
38995 * modulo-sched.c (sms_schedule): Update call of loop_version.
38996 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38997 * tree-parloops.c (gen_parallel_loop): Likewise.
38998 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38999 * tree-ssa-loop-split.c (split_loop): Likewise.
39000 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
39001 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
39003 2017-02-05 Martin Liska <mliska@suse.cz>
39006 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39008 (print_operand_address): Initialize a struct to zero.
39010 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39012 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39013 garbage collector only in textual form.
39015 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39017 * doc/extend.texi (x86 specific memory model extensions for
39018 transactional memory): Simplify a phrase.
39020 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39023 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39024 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39025 (atomic_storedi_1): Likewise.
39027 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39029 PR tree-optimization/79338
39030 * tree-parloops.c (gather_scalar_reductions): Don't call
39031 vect_analyze_loop_form for loop->inner before destroying loop's
39034 2017-02-03 Martin Sebor <msebor@redhat.com>
39036 PR tree-optimization/79327
39037 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39038 when precision has resulted in leading zeros.
39039 (format_integer): Adjust the likely counter to assume an unknown
39040 argument that may be zero is non-zero.
39042 2017-02-03 Jason Merrill <jason@redhat.com>
39045 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39046 avoid copying non-taken branch.
39048 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39050 PR tree-optimization/79340
39051 * tree-vect-loop.c (vectorizable_reduction): Release
39052 vec_defs elements after safe_splicing them into other vectors.
39055 PR tree-optimization/79327
39056 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39057 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39059 (format_integer): Use wide_int_to_tree instead of build_int_cst
39060 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39061 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39062 of shortest and longest sequence.
39064 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39066 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39067 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39069 2017-02-03 Walter Lee <walt@tilera.com>
39072 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39073 after initial stackframe link reg save.
39074 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39076 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39079 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39080 wu for stxssp alternative.
39082 2017-02-03 Martin Sebor <msebor@redhat.com>
39084 PR tree-optimization/79352
39085 * gimple-fold.c (get_range_strlen): Add argument.
39086 (get_range_strlen): Change return type to bool.
39087 (get_maxval_strlen): Pass in a dummy argument.
39088 * gimple-fold.h (get_range_strlen): Change return type to bool.
39089 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39090 * tree.h (array_at_struct_end_p): Add argument.
39091 * tree.c (array_at_struct_end_p): Handle it.
39093 2017-02-03 Martin Liska <mliska@suse.cz>
39096 * multiple_target.c (create_dispatcher_calls): Redirect edge
39097 from a caller of a dispatcher.
39098 (expand_target_clones): Make the clones local.
39099 (ipa_target_clone): Do both target clones and resolvers.
39100 (ipa_dispatcher_calls): Remove the pass.
39101 (pass_dispatcher_calls::gate): Likewise.
39102 (make_pass_dispatcher_calls): Likewise.
39103 * passes.def (pass_target_clone): Put as very first IPA early
39106 2017-02-03 Martin Liska <mliska@suse.cz>
39108 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39109 in case of a function with ifunc attribute.
39111 2017-02-03 Martin Liska <mliska@suse.cz>
39113 * cgraph.c (cgraph_node::dump): Dump function version info.
39114 * symtab.c (symtab_node::dump_base): Add missing new line.
39116 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39118 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39119 (ifcombine_ifandif): Use it.
39121 2017-02-03 Martin Liska <mliska@suse.cz>
39123 * doc/invoke.texi: Document default value for
39124 use-after-scope-direct-emission-threshold.
39126 2017-02-03 Martin Liska <mliska@suse.cz>
39128 PR tree-optimization/79339
39129 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39130 (format_floating): Likewise.
39132 2017-02-03 Martin Liska <mliska@suse.cz>
39135 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39137 (ipa_node_params_t::remove): Likewise.
39138 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39139 initialization from removed ipa_node_params_t::insert.
39140 (ipa_node_params::~ipa_node_params): Move from removed
39141 ipa_node_params_t::release.
39142 * symbol-summary.h (symbol_summary::m_released): New member.
39143 Do not release a summary twice. Do not allow to call finalizer
39144 for types of a summary that live in GGC memory.
39146 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39148 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39151 2017-02-02 Martin Sebor <msebor@redhat.com>
39153 PR middle-end/79275
39154 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39155 (format_string): Tighten up the range of output for non-constant
39156 strings and correct the expected range for wide non-constant strings.
39158 2017-02-02 Martin Sebor <msebor@redhat.com>
39160 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39162 PR middle-end/32003
39163 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39165 (-fdump-tree-@var): Add to index and document how to come up
39166 with pass-specific option and dump file names.
39167 (-fdump-passes): Clarify where to look for output.
39169 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39171 PR middle-end/77445
39172 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39173 statistics of the analyzed path; allow threading for speed when
39174 any of BBs along the path are optimized for speed.
39176 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39178 PR middle-end/78468
39179 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39180 settings of the virtual registers.
39183 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39185 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39186 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39189 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39191 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39192 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39194 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39196 * config/s390/s390.md: Add missing comments with the expanded
39198 * config/s390/vector.md: Likewise.
39199 * config/s390/vx-builtins.md: Likewise.
39201 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39204 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39205 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39206 conditions on a single line.
39208 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39210 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39211 __S390_VX__ to __VX__.
39213 2017-02-01 Andrew Pinski <apinski@cavium.com>
39215 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39216 stmt_info to record_stmt_cost.
39217 (vect_get_known_peeling_cost): Pass stmt_info if known to
39219 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39220 cpu_vector_cost field into
39221 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39222 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39223 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39224 splitting of scalar_stmt_cost and vec_stmt_cost.
39225 (thunderx_vector_cost): Likewise.
39226 (cortexa57_vector_cost): LIkewise.
39227 (exynosm1_vector_cost): Likewise.
39228 (xgene1_vector_cost): Likewise.
39229 (thunderx2t99_vector_cost): Improve after the splitting of the two
39231 (aarch64_builtin_vectorization_cost): Update for the splitting of
39232 scalar_stmt_cost and vec_stmt_cost.
39234 2017-02-01 Torvald Riegel <triegel@redhat.com>
39235 Richard Henderson <rth@redhat.com>
39237 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39238 conditional on existance of a fast atomic load.
39239 * optabs-query.c (can_atomic_load_p): New function.
39240 * optabs-query.h (can_atomic_load_p): Declare it.
39241 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39242 no fast atomic load is available for the particular size of access.
39243 (expand_atomic_compare_and_swap): Likewise.
39244 (expand_atomic_load): Likewise.
39245 (expand_atomic_store): Likewise.
39246 (expand_atomic_fetch_op): Likewise.
39247 * testsuite/lib/target-supports.exp
39248 (check_effective_target_sync_int_128): Remove x86 because it provides
39249 no fast atomic load.
39250 (check_effective_target_sync_int_128_runtime): Likewise.
39252 2017-02-01 Richard Biener <rguenther@suse.de>
39254 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39255 (graphite_transform_loops): Provide opt-info for optimized nests.
39256 * tree-parloop.c (parallelize_loops): Provide opt-info for
39257 parallelized loops.
39259 2017-02-01 Richard Biener <rguenther@suse.de>
39261 PR middle-end/79315
39262 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39263 was not set before.
39265 2017-02-01 Richard Biener <rguenther@suse.de>
39267 PR tree-optimization/71824
39268 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39269 Verify the loops are valid in the merged SESE region.
39270 (scop_detection::can_represent_loop_1): Check analyzing the
39271 evolution of the number of iterations in the region succeeds.
39273 2017-01-31 Ian Lance Taylor <iant@golang.org>
39275 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39276 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39278 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39280 PR preprocessor/79210
39281 * input.c (get_substring_ranges_for_loc): Replace line_width
39282 assertion with error-handling.
39284 2017-01-31 Richard Biener <rguenther@suse.de>
39286 PR tree-optimization/77318
39287 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39288 (create_pw_aff_from_tree): Take loop parameter.
39289 (add_condition_to_pbb): Pass loop of the condition to
39290 create_pw_aff_from_tree.
39292 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39294 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39295 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39297 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39301 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39303 (convert_int_to_float128): Likewise.
39304 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39305 (convert_int_to_float128): Likewise.
39306 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39307 (UNSPEC_IEEE128_CONVERT): Likewise.
39308 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39309 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39310 Use local variables for IBM extended format.
39311 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39312 (fix_trunc<mode>si2_fprs): Likewise.
39313 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39314 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39315 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39316 to know that we can now have integers of all sizes in vector
39318 (fix<uns>_<mode>di2_hw): Likewise.
39319 (float<uns>_<mode>si2_hw): Likewise.
39320 (fix_<mode>si2_hw): Likewise.
39321 (fixuns_<mode>si2_hw): Likewise.
39322 (float<uns>_<mode>di2_hw): Likewise.
39323 (float_<mode>di2_hw): Likewise.
39324 (float_<mode>si2_hw): Likewise.
39325 (floatuns_<mode>di2_hw): Likewise.
39326 (floatuns_<mode>si2_hw): Likewise.
39327 (xscvqp<su>wz_<mode>): Delete, no longer used.
39328 (xscvqp<su>dz_<mode>): Likewise.
39329 (xscv<su>dqp_<mode>): Likewise.
39330 (ieee128_mfvsrd_64bit): Likewise.
39331 (ieee128_mfvsrd_32bit): Likewise.
39332 (ieee128_mfvsrwz): Likewise.
39333 (ieee128_mtvsrw): Likewise.
39334 (ieee128_mtvsrd_64bit): Likewise.
39335 (ieee128_mtvsrd_32bit): Likewise.
39337 2017-01-31 Martin Liska <mliska@suse.cz>
39340 * ipa-prop.c (ipa_free_all_node_params): Call release method
39341 instead of ~sumbol_summary to not to trigger double times
39344 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39346 PR tree-optimization/71691
39347 * bitmap.h (class auto_bitmap): New.
39348 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39349 is_maybe_undefined instead of ssa_undefined_value_p.
39351 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39353 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39354 __S390_ARCH_LEVEL__ to __ARCH__.
39356 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39358 PR tree-optimization/79267
39359 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39360 if should_remove_lhs_p is true.
39362 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39365 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39366 (add_alignment_attribute): New.
39367 (base_type_die): Add alignment attribute.
39368 (subrange_type_die): Likewise.
39369 (modified_type_die): Likewise.
39370 (gen_array_type_die): Likewise.
39371 (gen_descr_array_type_die: Likewise.
39372 (gen_enumeration_type_die): Likewise.
39373 (gen_subprogram_die): Likewise.
39374 (gen_variable_die): Likewise.
39375 (gen_field_die): Likewise.
39376 (gen_ptr_to_mbr_type_die): Likewise.
39377 (gen_struct_or_union_type_die): Likewise.
39378 (gen_subroutine_type_die): Likewise.
39379 (gen_typedef_die): Likewise.
39380 (base_type_cmp): Compare alignment attribute.
39382 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39385 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39386 (setb_unsigned) New pattern for setb with CCUNS.
39387 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39388 subfc./subfe sequence to avoid overflow problems. Generate a
39389 shorter sequence with cmpld/setb for power9.
39390 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39391 for generating subfc. instruction.
39392 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39393 now uses this instruction.
39395 2017-01-30 Ian Lance Taylor <iant@google.com>
39398 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39399 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39401 2017-01-30 Martin Sebor <msebor@redhat.com>
39403 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39404 Move constant to the right of a relational operator.
39405 (get_mpfr_format_length, format_character, format_string): Ditto.
39406 (should_warn_p, maybe_warn): Same.
39408 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39410 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39413 * asan.c (get_translation_unit_decl): Remove function.
39414 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39416 2017-01-30 Martin Liska <mliska@suse.cz>
39418 PR gcov-profile/79259
39419 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39420 -fprofile-generate.
39422 2017-01-30 Martin Liska <mliska@suse.cz>
39425 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39426 Initialize variables with NULL value.
39428 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39431 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39433 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39435 2017-01-30 Richard Biener <rguenther@suse.de>
39437 PR tree-optimization/79276
39438 * tree-vrp.c (process_assert_insertions): Properly adjust common
39439 when removing a duplicate.
39441 2017-01-30 Richard Biener <rguenther@suse.de>
39443 PR tree-optimization/79256
39444 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39445 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39447 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39449 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39452 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39453 ("*r<noxa>sbg_<mode>_sll_bitmask")
39454 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39455 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39456 Use contiguous_bitmask_nowrap_operand.
39458 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39461 * config/rs6000/altivec.h (vec_xl): Revise #define.
39462 (vec_xst): Likewise.
39464 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39466 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39468 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39470 PR rtl-optimization/79194
39471 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39472 traps before call to bypass_conditional_jumps.
39474 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39476 PR tree-optimization/71374
39477 * lra-constraints.c (check_conflict_input_operands): New.
39478 (match_reload): Use it.
39480 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39483 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39484 account to calculate conflict_set.
39486 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39488 PR rtl-optimization/78559
39489 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39490 other_insn in combine.
39492 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39494 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39495 uint16_type_node for BT_UINT16.
39497 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39499 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39500 "RTL Tests" to menu.
39501 (GIMPLE Tests): New node.
39502 (RTL Tests): New node.
39504 2017-01-27 Richard Biener <rguenther@suse.de>
39506 PR tree-optimization/79245
39507 * tree-loop-distribution.c (distribute_loop): Apply cost
39508 modeling also to detected patterns.
39510 2017-01-27 Richard Biener <rguenther@suse.de>
39512 PR tree-optimization/71433
39513 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39514 (compare_assert_loc): New function.
39515 (process_assert_insertions): Sort and optimize assert locations
39516 to remove duplicates and push down identical assertions on
39517 edges to their destination block.
39519 2017-01-27 Richard Biener <rguenther@suse.de>
39521 PR tree-optimization/79244
39522 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39523 out SSA names even if abnormal.
39525 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39527 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39528 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39529 instead of MPFR_RNDN.
39531 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39534 * arm.c (arm_option_override): Don't call build_target_option_node
39535 until after doing all option overrides.
39536 (arm_valid_target_attribute_tree): Likewise.
39538 2017-01-27 Martin Liska <mliska@suse.cz>
39540 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39541 for {cd}tors and C++ {cd}tors.
39543 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39545 * config/s390/s390.md ("*setmem_long_and")
39546 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39548 2017-01-26 Martin Sebor <msebor@redhat.com>
39550 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39553 2017-01-26 Martin Sebor <msebor@redhat.com>
39555 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39556 HAVE_DFmode before using XFmode or DFmode.
39557 (parse_directive): Avoid using the z length modifier to avoid
39558 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39560 PR middle-end/78703
39561 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39562 to accept adjustment as an array.
39563 (get_int_range): New function.
39564 (struct directive): Make width and prec arrays.
39565 (directive::set_width, directive::set_precision): Call get_int_range.
39566 (format_integer, format_floating): Handle width and precision ranges.
39567 (format_string, parse_directive): Same.
39569 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39572 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39573 comdat_type_p set, just clone them, but keep the children in the
39577 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39578 which have direct callers with -fvar-tracking-assignments enabled
39580 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39581 inside of type units.
39583 2017-01-26 Martin Sebor <msebor@redhat.com>
39585 PR middle-end/78703
39586 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39588 (struct format_result): Replace number_chars, number_chars_min,
39589 and number_chars_max with a single member of struct result_range.
39591 (format_result::operator+=): Adjust.
39592 (struct fmtresult): Remove bounded. Handle likely and unlikely
39594 (fmtresult::adjust_for_width_or_precision): New function.
39595 (fmtresult:type_max_digits): New function.
39596 (bytes_remaining): Handle likely and unlikely counters.
39597 (min_bytes_remaining): Remove.
39598 (format_percent): Simplify.
39599 (format_integer, format_floating): Set likely and unlikely counters.
39600 (get_string_length, format_character, format_string): Same.
39601 (format_plain, should_warn_p): New function.
39602 (maybe_warn): Call should_warn_p. Update diagnostic messages
39603 and handle those for all directives, including plain strings.
39604 (format_directive): Handle likely and unlikely counters.
39605 Remove unnecessary quoting from diagnostics. Add an informational
39607 (add_bytes): Remove.
39608 (pass_sprintf_length::compute_format_length): Simplify.
39609 (try_substitute_return_value): Handle likely and unlikely counters.
39611 2017-01-26 Carl Love <cel@us.ibm.com>
39613 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39614 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39616 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39619 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39620 endianess for subregs into account.
39621 * lra-constraints.c (lra_constraints): Do risky transformations
39622 always on the first iteration.
39623 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39624 last_call_used_reg_set.
39625 (process_bb_lives): Define and use last_call_used_reg_set.
39626 * lra.c (lra): Always continue after lra_constraints on the first
39629 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39631 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39633 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39635 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39637 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39638 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39639 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39640 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39641 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39642 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39643 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39644 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39645 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39647 2017-01-26 Marek Polacek <polacek@redhat.com>
39650 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39651 for the third operand.
39653 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39655 PR middle-end/79236
39656 * omp-low.c (struct omp_context): Add simt_stmt field.
39657 (scan_omp_for): Return omp_context *.
39658 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39659 context to the _simt_ SIMD stmt.
39660 (lower_omp_for): For combined SIMD with sibling _simt_
39661 SIMD, make sure to use the same decls in _looptemp_
39662 clauses as in the sibling.
39664 2017-01-26 David Sherwood <david.sherwood@arm.com>
39666 PR middle-end/79212
39667 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39670 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39673 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39674 emit fld b; fld a; if possible.
39676 * brig-builtins.def: Update copyright years.
39677 * config/arm/arm_acle_builtins.def: Update copyright years.
39679 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39682 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39683 constraint instead of o for the stxsd instruction.
39685 2017-01-25 Carl Love <cel@us.ibm.com>
39687 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39688 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39690 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39692 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39694 2017-01-25 Richard Biener <rguenther@suse.de>
39696 PR tree-optimization/69264
39697 * target.def (vector_alignment_reachable): Improve documentation.
39698 * doc/tm.texi: Regenerate.
39699 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39701 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39702 earlier changes with respect to TYPE_USER_ALIGN.
39703 (vector_alignment_reachable_p): Likewise. Improve dumping.
39705 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39708 * config/arm/arm.md (xordi3): Force constant operand into a register
39711 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39713 * doc/invoke.texi (-fstore-merging): Correct default optimization
39714 levels at which it is enabled.
39715 (-O): Move -fstore-merging from list to...
39718 2017-01-25 Richard Biener <rguenther@suse.de>
39721 * omp-expand.c: Include debug.h.
39722 (expand_omp_taskreg): Make sure to generate early debug before
39723 outlining anything from a function.
39724 (expand_omp_target): Likewise.
39725 (grid_expand_target_grid_body): Likewise.
39727 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39730 * asan.c (get_translation_unit_decl): New function.
39731 (asan_add_global): Extract modules file name from globals
39732 TRANSLATION_UNIT_DECL name.
39734 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39737 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39738 for long calls with APCS frame and VFP.
39740 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39742 * cfg.c (original_copy_tables_initialized_p): New function.
39743 * cfg.h (original_copy_tables_initialized_p): New decl.
39744 * cfgrtl.c (relink_block_chain): Guard the call to
39745 free_original_copy_tables with a call to
39746 original_copy_tables_initialized_p.
39747 * cgraph.h (symtab_node::native_rtl_p): New decl.
39748 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39749 (symtab_node::needed_p): Don't assert for early assembly output
39750 for __RTL functions.
39751 (cgraph_node::finalize_function): Set "force_output" for __RTL
39753 (cgraph_node::analyze): Bail out early for __RTL functions.
39754 (analyze_functions): Update assertion to support __RTL functions.
39755 (cgraph_node::expand): Bail out early for __RTL functions.
39756 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39758 * function.h (struct function): Update comment for field
39760 * gimple-expr.c: Include "tree-pass.h".
39761 (gimple_has_body_p): Return false for __RTL functions.
39762 * Makefile.in (OBJS): Add run-rtl-passes.o.
39763 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39765 (gcc::pass_manager::get_clean_slate): New accessor.
39766 * passes.c: Include "insn-addr.h".
39767 (should_skip_pass_p): Add logging. Update logic for running
39768 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39769 property-provider override so it is only done for gimple passes.
39771 (skip_pass): New function.
39772 (execute_one_pass): Call skip_pass when skipping passes.
39773 * read-md.c (md_reader::read_char): Support filtering
39774 the input to a subset of line numbers.
39775 (md_reader::md_reader): Initialize fields
39776 m_first_line and m_last_line.
39777 (md_reader::read_file_fragment): New function.
39778 * read-md.h (md_reader::read_file_fragment): New decl.
39779 (md_reader::m_first_line): New field.
39780 (md_reader::m_last_line): New field.
39781 * read-rtl-function.c (function_reader::create_function): Only
39782 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39783 curr_properties. Set DECL_INITIAL to a dummy block.
39784 (read_rtl_function_body_from_file_range): New function.
39785 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39787 * run-rtl-passes.c: New file.
39788 * run-rtl-passes.h: New file.
39790 2017-01-24 Jeff Law <law@redhat.com>
39792 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39795 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39797 PR tree-optimization/79159
39798 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39799 (record_nonwrapping_iv): Improve boundary using above function if no
39800 value range information.
39802 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39803 Martin Jambor <mjambor@suse.cz>
39805 * brig-builtins.def: New file.
39806 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39807 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39808 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39809 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39810 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39811 * builtin-types.def (BT_INT8): New.
39812 (BT_INT16): Likewise.
39813 (BT_UINT8): Likewise.
39814 (BT_UINT16): Likewise.
39815 (BT_FN_ULONG): Likewise.
39816 (BT_FN_UINT_INT): Likewise.
39817 (BT_FN_UINT_ULONG): Likewise.
39818 (BT_FN_UINT_LONG): Likewise.
39819 (BT_FN_UINT_PTR): Likewise.
39820 (BT_FN_ULONG_PTR): Likewise.
39821 (BT_FN_INT8_FLOAT): Likewise.
39822 (BT_FN_INT16_FLOAT): Likewise.
39823 (BT_FN_UINT32_FLOAT): Likewise.
39824 (BT_FN_UINT16_FLOAT): Likewise.
39825 (BT_FN_UINT8_FLOAT): Likewise.
39826 (BT_FN_UINT64_FLOAT): Likewise.
39827 (BT_FN_UINT16_UINT32): Likewise.
39828 (BT_FN_UINT32_UINT16): Likewise.
39829 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39830 (BT_FN_INT_PTR_INT): Likewise.
39831 (BT_FN_UINT_PTR_UINT): Likewise.
39832 (BT_FN_LONG_PTR_LONG): Likewise.
39833 (BT_FN_ULONG_PTR_ULONG): Likewise.
39834 (BT_FN_VOID_UINT64_UINT64): Likewise.
39835 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39836 (BT_FN_INT8_INT8_INT8): Likewise.
39837 (BT_FN_INT16_INT16_INT16): Likewise.
39838 (BT_FN_INT_INT_INT): Likewise.
39839 (BT_FN_UINT_FLOAT_UINT): Likewise.
39840 (BT_FN_FLOAT_UINT_UINT): Likewise.
39841 (BT_FN_ULONG_UINT_UINT): Likewise.
39842 (BT_FN_ULONG_UINT_PTR): Likewise.
39843 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39844 (BT_FN_UINT_UINT_UINT): Likewise.
39845 (BT_FN_VOID_UINT_PTR): Likewise.
39846 (BT_FN_UINT_UINT_PTR: Likewise.
39847 (BT_FN_UINT32_UINT64_PTR): Likewise.
39848 (BT_FN_INT_INT_UINT_UINT): Likewise.
39849 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39850 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39851 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39852 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39853 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39854 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39855 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39856 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39857 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39858 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39859 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39860 * doc/frontends.texi: List BRIG FE.
39861 * doc/install.texi (Testing): Add BRIG tesring requirements.
39862 * doc/invoke.texi (Overall Options): Mention BRIG.
39863 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39865 2017-01-24 Richard Biener <rguenther@suse.de>
39867 PR translation/79208
39868 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39870 2017-01-24 Martin Jambor <mjambor@suse.cz>
39873 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39874 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39875 and known_contexts.
39877 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39879 PR middle-end/79123
39880 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39881 casts from signed to unsigned really don't have a range.
39883 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39885 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39886 GMP_RNDx for compatiblity.
39888 2017-01-24 Martin Liska <mliska@suse.cz>
39891 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39892 that would prevent us to call alloca with -1 as argument.
39894 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39896 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39897 Avoid -Wformat-security warning.
39899 2017-01-23 Andrew Pinski <apinski@cavium.com>
39901 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39904 2017-01-23 Martin Sebor <msebor@redhat.com>
39906 PR middle-end/78703
39907 * gimple-ssa-sprintf.c (warn_level): New global.
39908 (format_integer): Use it here and throughout the rest of the file.
39909 Use the same switch to compute sign as base.
39910 (maybe_warn): New function.
39911 (format_directive): Factor out warnings into maybe_warn.
39912 Add debugging output. Use warn_level.
39913 (add_bytes): Use warn_level.
39914 (pass_sprintf_length::compute_format_length): Add debugging output.
39915 (try_substitute_return_value): Same.
39916 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39918 PR middle-end/78703
39919 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39920 (struct fmtresult, format_integer, format_floating): Adjust.
39921 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39922 (get_string_length, format_string,format_directive): Same.
39923 (pass_sprintf_length::compute_format_length): Same.
39924 (try_substitute_return_value): Simplify slightly.
39926 PR middle-end/78703
39927 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39928 (fmtresult::operator+=): Outlined.
39929 (struct fmtresult): Add ctors.
39930 (struct conversion_spec): Rename...
39931 (struct directive): ...to this. Add and remove data members.
39932 (directive::set_width, directive::set_precision): New functions.
39933 (format_percent): Use fmtresult ctor.
39934 (get_width_and_precision): Remove.
39935 (format_integer): Make naming changes. Avoid computing width and
39937 (format_floating): Same. Adjust indentation.
39938 (format_character, format_none): New functions.
39939 (format_string): Moved character handling to format_character.
39940 (format_directive): Remove arguments, change return type.
39941 (parse_directive): New function.
39942 (pass_sprintf_length::compute_format_length): Move directive
39943 parsing to parse_directive.
39945 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39947 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39948 (assign_assembler_name_if_needed): ... this.
39949 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39950 (assign_assembler_name_if_needed): ... this.
39951 (free_lang_data_in_cgraph): Adjust callers.
39952 * cgraphunit.c (cgraph_node::analyze): Likewise.
39953 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39956 2017-01-23 Richard Biener <rguenther@suse.de>
39958 PR tree-optimization/79088
39959 PR tree-optimization/79188
39960 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39961 resetting loop bounds after last path deletion. Reset loop
39962 bounds of the target loop, make code match the comments.
39963 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39964 Make sure loops need no fixups.
39966 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39968 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39969 exponent support with double type for first argument.
39970 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39971 type returned by __builtin_vec_extract_sig,
39972 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39973 functions from "vector int" to "vector unsigned int" or from
39974 "vector long long int" to "vector unsigned long long int".
39975 Changed type returned by __builtin_vec_extract_exp,
39976 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39977 functions from "vector int" to "vector unsigned int" or from
39978 "vector long long int" to "vector unsigned long long int".
39979 Changed return type of __builtin_vec_test_data_class,
39980 __builtin_vec_test_data_class_sp, and
39981 __builtin_vec_test_data_class_dp from "vector int" to
39982 "vector bool int" or from "vector long long int" to "vector bool
39983 long long int" and changed second argument type from "unsigned
39984 int" to "int". Added new overloaded function forms "vector float
39985 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39986 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39987 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39988 double, vector unsigned long long int)" and "vector double
39989 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39990 long int)". Changed return type of
39991 __builtin_scalar_test_data_class and
39992 __builtin_scalar_test_data_class_sp and
39993 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39994 int" and changed second argument from "unsigned int" to "int".
39995 Changed type returned by __builtin_scalar_test_neg,
39996 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39997 from "int" to "bool int". Added new overloaded function form
39998 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39999 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
40000 exponent double-precision with floating point first argument.
40001 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
40002 documentation of scalar_test_data_class, scalar_test_neg,
40003 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
40004 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
40005 vec_test_data_class built-in functions to reflect refinements in
40006 their type signatures.
40008 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40010 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40012 (aarch64_elf_asm_destructor): Likewise.
40014 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40016 PR rtl-optimization/78634
40017 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40018 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40019 * ifcvt.c (noce_try_cmove): Add missing cost check.
40021 PR rtl-optimization/71724
40022 * combine.c (if_then_else_cond): Look for situations where it is
40023 beneficial to undo the work of one of the recursive calls.
40025 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40027 PR tree-optimization/70754
40028 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40029 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40030 combined stmt before it if not NULL.
40031 (combine_chains): Process refs reversely and compute dominance point
40034 2017-01-23 Martin Liska <mliska@suse.cz>
40036 PR tree-optimization/79196
40037 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40038 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40040 (strlen_optimize_stmt): Call the renamed function.
40042 2017-01-23 Michael Matz <matz@suse.de>
40044 PR tree-optimization/78384
40045 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40047 2017-01-23 Richard Biener <rguenther@suse.de>
40049 PR tree-optimization/79186
40050 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40051 both incoming edges before moving an assert.
40053 2017-01-23 Martin Jambor <mjambor@suse.cz>
40055 * ipa-prop.c (load_from_param_1): Removed.
40056 (load_from_unmodified_param): Bits from load_from_param_1 put back
40058 (load_from_param): Removed.
40059 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40060 with stmt. Reverted back to use of load_from_unmodified_param.
40062 2017-01-23 Martin Jambor <mjambor@suse.cz>
40065 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40066 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40067 field a pointer to garbage collected vector, mark lattices and
40068 ipcp_orig_node with GTY((skip)).
40069 (ipa_get_param_count): Adjust to descriptors being a pointer.
40070 (ipa_get_param): Likewise.
40071 (ipa_get_type): Likewise.
40072 (ipa_get_param_move_cost): Likewise.
40073 (ipa_set_param_used): Likewise.
40074 (ipa_get_controlled_uses): Likewise.
40075 (ipa_set_controlled_uses): Likewise.
40076 (ipa_is_param_used): Likewise.
40077 (ipa_node_params_t): Move into garbage collector. New methods insert
40079 (ipa_node_params_sum): Annotate wth GTY(()).
40080 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40082 (ipa_load_from_parm_agg): Adjust declaration.
40083 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40084 * ipa-profile.c (ipa_profile): Likewise.
40085 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40086 (ipa_populate_param_decls): Make descriptors parameter garbage
40088 (ipa_dump_param): Adjust to descriptors being a pointer.
40089 (ipa_alloc_node_params): Likewise.
40090 (ipa_initialize_node_params): Likewise.
40091 (load_from_param_1): Make descriptors parameter garbage collected.
40092 (load_from_unmodified_param): Likewise.
40093 (load_from_param): Likewise.
40094 (ipa_load_from_parm_agg): Likewise.
40095 (ipa_node_params::~ipa_node_params): Removed.
40096 (ipa_free_all_node_params): Remove call to delete operator.
40097 (ipa_node_params_t::insert): New.
40098 (ipa_node_params_t::remove): Likewise.
40099 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40100 copy known_csts and known_contexts vectors.
40101 (ipa_read_node_info): Adjust to descriptors being a pointer.
40102 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40104 (ipcp_transform_function): Make descriptors variable garbage
40107 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40109 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40110 * config/i386/avx512dqintrin.h: Ditto.
40111 * config/i386/avx512fintrin.h: Ditto.
40112 * config/i386/i386.c: Handle new builtins.
40113 * config/i386/i386-builtin.def: Add new builtins.
40114 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40115 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40117 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40118 Martin Liska <mliska@suse.cz>
40120 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40121 * asan.c (asan_expand_poison_ifn): Support stores and use
40122 appropriate ASAN report function.
40123 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40124 * internal-fn.def (ASAN_POISON_USE): Declare.
40125 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40126 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40127 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40128 ASAN_POISON calls w/o LHS.
40129 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40130 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40131 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40132 * gimplify.c (asan_poison_variables): Add attribute
40133 use_after_scope_memory to variables that really needs to live
40135 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40136 having the attribute.
40138 2017-01-23 Martin Liska <mliska@suse.cz>
40140 * asan.c (create_asan_shadow_var): New function.
40141 (asan_expand_poison_ifn): Likewise.
40142 * asan.h (asan_expand_poison_ifn): New declaration.
40143 * internal-fn.c (expand_ASAN_POISON): Likewise.
40144 * internal-fn.def (ASAN_POISON): New builtin.
40145 * sanopt.c (pass_sanopt::execute): Expand
40146 asan_expand_poison_ifn.
40147 * tree-inline.c (copy_decl_for_dup_finish): Make function
40149 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40150 * tree-ssa.c (is_asan_mark_p): New function.
40151 (execute_update_addresses_taken): Rewrite local variables
40152 (identified just by use-after-scope as addressable) into SSA.
40154 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40156 * doc/install.texi (Specific): opensource.apple.com uses https
40157 now. Remove trailing slash.
40159 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40161 * README.Portability: Remove note on an Irix compatibility issue.
40163 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40165 * gcov.c (INCLUDE_ALGORITHM): Define.
40166 (INCLUDE_VECTOR): Define.
40167 No longer include <vector> and <algorithm> directly.
40169 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40171 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40173 * doc/invoke.texi (Code Gen Options): Ditto.
40175 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40178 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40180 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40182 rtl-optimization/79125
40183 * cprop.c (local_cprop_pass): Handle cases where we make an
40184 unconditional trap.
40186 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40190 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40191 read from, for big endian.
40193 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40195 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40196 register pauth builtins for LP64 only.
40198 2017-01-20 Marek Polacek <polacek@redhat.com>
40201 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40204 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40206 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40208 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40209 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40210 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40212 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40215 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40216 in big-endian mode when they are not a single duplicated value.
40218 2017-01-20 Richard Biener <rguenther@suse.de>
40220 * BASE-VER: Bump to 7.0.1.
40222 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40224 * omp-low.c (omplow_simd_context): New struct. Use it...
40225 (lower_rec_simd_input_clauses): ...here and...
40226 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40227 references to idx, lane, max_vf, is_simt.
40229 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40231 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40234 2017-01-20 Martin Jambor <mjambor@suse.cz>
40236 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40237 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40239 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40240 (GTFILES): Rename hsa.c to hsa-common.c.
40241 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40242 * hsa-dump.c: Likewise.
40243 * hsa-gen.c: Likewise.
40244 * hsa-regalloc.c: Likewise.
40245 * ipa-hsa.c: Likewise.
40246 * omp-expand.c: Likewise.
40247 * omp-low.c: Likewise.
40248 * toplev.c: Likewise.
40250 2017-01-20 Marek Polacek <polacek@redhat.com>
40253 * doc/invoke.texi: Document -Wduplicated-branches.
40254 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40255 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40256 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40257 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40258 return 0 only when not OEP_LEXICOGRAPHIC.
40259 (fold_build_cleanup_point_expr): Use the expression
40260 location when building CLEANUP_POINT_EXPR.
40261 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40262 * tree.c (add_expr): Handle error_mark_node.
40264 2017-01-20 Martin Liska <mliska@suse.cz>
40267 * tree-profile.c (init_ic_make_global_vars): Do not call
40269 (gimple_init_gcov_profiler): Likewise.
40271 2017-01-20 Martin Liska <mliska@suse.cz>
40274 * cgraph.h (maybe_create_reference): Remove argument and
40276 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40278 * ipa-cp.c (create_specialized_node): Likewise.
40279 * symtab.c (symtab_node::maybe_create_reference): Handle
40280 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40282 2017-01-20 Martin Liska <mliska@suse.cz>
40284 * read-rtl-function.c (function_reader::create_function): Use
40285 build_decl instread of build_decl_stat.
40287 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40289 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40290 * config/i386/avx512dqintrin.h: Ditto.
40291 * config/i386/avx512fintrin.h: Ditto.
40292 * config/i386/i386-builtin-types.def: Add new types.
40293 * config/i386/i386.c: Handle new types.
40294 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40295 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40296 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40297 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40298 (__builtin_ia32_kshiftridi): New.
40299 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40301 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40305 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40306 define to rs6000_init_stack_protect_guard.
40307 (rs6000_init_stack_protect_guard): New function.
40309 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40310 Yunqiang Su <yunqiang.su@imgtec.com>
40312 * config.gcc (supported_defaults): Add madd4.
40313 (with_madd4): Add validation.
40314 (all_defaults): Add madd4.
40315 * config/mips/mips.opt (mmadd4): New option.
40316 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40318 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40320 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40321 (ISA_HAS_FUSED_MADD4): Likewise.
40322 * doc/invoke.texi (-mmadd4): Document the new option.
40323 * doc/install.texi (--with-madd4): Document the new option.
40325 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40327 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40328 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40329 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40330 (aarch64_init_pauth_hint_builtins): New.
40331 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40332 (aarch64_expand_builtin): Expand new builtins.
40334 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40336 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40337 * combine-stack-adj.c (no_unhandled_cfa): Handle
40338 REG_CFA_TOGGLE_RA_MANGLE.
40339 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40340 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40341 info for return address signing.
40342 (aarch64_expand_epilogue): Likewise.
40344 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40346 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40347 * config/aarch64/aarch64-protos.h
40348 (aarch64_return_address_signing_enabled): New declaration.
40349 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40351 (aarch64_expand_prologue): Sign return address before it's pushed onto
40353 (aarch64_expand_epilogue): Authenticate return address fetched from
40355 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40356 (aarch64_attributes): New function attributes for "sign-return-address".
40357 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40358 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40359 ("*do_return"): Generate combined instructions according to key index.
40360 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40361 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40363 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40364 * config/aarch64/aarch64.opt (msign-return-address=): New.
40365 * doc/extend.texi (AArch64 Function Attributes): Documents
40366 "sign-return-address=".
40367 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40369 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40371 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40372 overall option summary.
40374 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40376 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40377 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40378 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40379 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40381 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40383 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40384 -mpower9-minmax by default for -mcpu=power9.
40385 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40386 128-bit floating point.
40388 2017-01-20 Alan Modra <amodra@gmail.com>
40390 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40391 optimizing for size.
40393 2017-01-20 Alan Modra <amodra@gmail.com>
40396 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40397 for strcmp and strncmp from corresponding builtin decl.
40399 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40401 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40402 instead of i386/rtems-64.h.
40403 * config/i386/rtems-64.h: Remove.
40405 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40409 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40411 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40413 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40415 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40416 Change int to HOST_WIDE_INT.
40417 * config/aarch64/aarch64-protos.h
40418 (aarch64_simd_gen_const_vector_dup): Likewise.
40419 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40421 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40423 * langhooks-def.h (lhd_type_for_size): New decl.
40424 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40425 * langhooks.c (lhd_type_for_size): New function, taken from
40428 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40430 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40431 define_bypass for CR latency.
40432 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40433 (power9-alu2): Add define_bypass for CR latency.
40435 (power9-mul): Update insn latency.
40436 (power9-mul-compare): Update insn latency, bypass latency and remove
40439 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40441 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40443 * config/aarch64/aarch64.md
40444 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40445 aarch64_nopcrelative_literal_loads.
40446 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40448 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40450 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40451 TARGET_LOONGSON_3A.
40452 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40454 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40457 * config.gcc (supported_defaults): Add lxc1-sxc1.
40458 (with_lxc1_sxc1): Add validation.
40459 (all_defaults): Add lxc1-sxc1.
40460 * config/mips/mips.opt (mlxc1-sxc1): New option.
40461 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40463 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40464 __mips_no_lxc1_sxc1.
40465 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40466 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40467 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40469 2017-01-19 Richard Biener <rguenther@suse.de>
40471 PR tree-optimization/72488
40472 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40473 sure to restore SSA info.
40474 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40476 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40478 PR rtl-optimization/79121
40479 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40480 of the inner type when shifting an extended value.
40482 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40485 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40486 interposable aliases.
40488 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40491 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40492 Use the evmergelohi instruction.
40493 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40494 (mov_sitf_e500_subreg8_2_be): Likewise.
40495 (mov_sitf_e500_subreg12_2_le): Likewise.
40496 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40497 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40498 (mov_sitf_e500_subreg8_2_le): Likewise.
40499 (mov_sitf_e500_subreg12_2_be): Likewise.
40501 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40503 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40504 attribute from vecsimple to vecperm.
40505 (altivec_vbpermq2): Likewise.
40507 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40510 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40512 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40513 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40514 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40515 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40516 case where N arg is SIZE_MAX.
40517 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40518 (cmpstrsi): Add pattern.
40520 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40522 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40523 __builtin_vec_revb builtins.
40524 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40525 built-in functions to support generation of the ISA 3.0 XXBR<x>
40526 vector byte reverse instructions.
40527 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40528 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40529 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40530 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40531 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40532 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40533 (P9V_BUILTIN_VEC_REVB): Likewise.
40534 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40535 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40536 (p9_xxbrq_v16qi): Likewise.
40537 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40538 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40539 (p9_xxbrh_v8hi): Likewise.
40540 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40541 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40542 vec_revb built-in functions.
40544 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40546 PR rtl-optimization/78952
40547 * config/i386/i386.md (any_extract): New code iterator.
40548 (*insvqi_2): Use any_extract for source operand.
40549 (*insvqi_3): Use any_shiftrt for source operand.
40551 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40553 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40555 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40557 2017-01-18 Matthias Klose <doko@ubuntu.com>
40559 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40561 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40563 * config/rs6000/altivec.h (vec_bperm): Change #define.
40564 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40565 (altivec_vbpermq2): New define_insn.
40566 (altivec_vbpermd): Likewise.
40567 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40568 function interface.
40569 (VBPERMD): Likewise.
40570 (VBPERM): New polymorphic function interface.
40571 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40572 Add entries for P9V_BUILTIN_VEC_VBPERM.
40573 * doc/extend.texi: Add interfaces for vec_bperm.
40575 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40577 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40578 first letter of error messages.
40579 (s390_resolve_overloaded_builtin): Likewise.
40580 * config/s390/s390.c (s390_expand_builtin): Likewise.
40581 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40582 (s390_valid_target_attribute_inner_p): Likewise.
40583 * config/s390/s390.md ("tabort"): Likewise.
40585 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40587 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40588 (ISA_AVOID_DIV_HILO): New macro.
40589 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40590 (ISA_HAS_DDIV): Likewise.
40592 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40594 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40596 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40598 * doc/invoke.texi (fabi-version): Spelling fix.
40600 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40603 * doc/invoke.texi (fabi-version): Mention mangling fix for
40606 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40609 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40611 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40614 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40615 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40617 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40619 (stack_protect_test): Ditto.
40620 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40621 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40623 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40624 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40625 -mstack-protector-guard-offset=.
40626 (RS/6000 and PowerPC Options): Ditto.
40628 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40630 * config/i386/i386.h (MASK_CLASS_P): New define.
40631 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40632 there are no registers from different register sets also when
40633 mask registers are used. Update function comment.
40634 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40635 to (*k/*r) and (*k/*km) alternatives.
40637 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40639 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40640 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40641 (EH_RETURN_HANDLER_RTX): New define.
40642 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40643 Force frame pointer in EH return functions.
40644 (aarch64_expand_epilogue): Add barrier for eh_return.
40645 (aarch64_final_eh_return_addr): Remove.
40646 (aarch64_eh_return_handler_rtx): New function.
40647 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40649 (aarch64_eh_return_handler_rtx): New prototype.
40651 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40653 * config/rs6000/altivec.h (vec_rlmi): New #define.
40654 (vec_vrlnm): Likewise.
40655 (vec_rlnm): Likewise.
40656 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40657 (UNSPEC_VRLNM): Likewise.
40658 (VIlong): New mode iterator.
40659 (altivec_vrl<VI_char>mi): New define_insn.
40660 (altivec_vrl<VI_char>nm): Likewise.
40661 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40663 (VRLDNM): Likewise.
40664 (RLNM): New polymorphic function entry.
40665 (VRLWMI): New monomorphic function entry.
40666 (VRLDMI): Likewise.
40667 (RLMI): New polymorphic function entry.
40668 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40669 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40670 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40673 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40676 * dwarf2out.c (field_byte_offset): Restore the
40677 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40678 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40679 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40682 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40685 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40687 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40690 * dwarf2out.c (add_data_member_location_attribute): For constant
40691 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40692 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40693 DW_AT_byte_size attributes.
40695 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40697 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40698 after forcing to constant memory when the code model is medium.
40700 2017-01-17 Julia Koval <julia.koval@intel.com>
40703 * config/i386/avx512fintrin.h
40704 (_mm512_i32gather_ps): Change __addr type to void const*.
40705 (_mm512_mask_i32gather_ps): Ditto.
40706 (_mm512_i32gather_pd): Ditto.
40707 (_mm512_mask_i32gather_pd): Ditto.
40708 (_mm512_i64gather_ps): Ditto.
40709 (_mm512_mask_i64gather_ps): Ditto.
40710 (_mm512_i64gather_pd): Ditto.
40711 (_mm512_mask_i64gather_pd): Ditto.
40712 (_mm512_i32gather_epi32): Ditto.
40713 (_mm512_mask_i32gather_epi32): Ditto.
40714 (_mm512_i32gather_epi64): Ditto.
40715 (_mm512_mask_i32gather_epi64): Ditto.
40716 (_mm512_i64gather_epi32): Ditto.
40717 (_mm512_mask_i64gather_epi32): Ditto.
40718 (_mm512_i64gather_epi64): Ditto.
40719 (_mm512_mask_i64gather_epi64): Ditto.
40720 (_mm512_i32scatter_ps): Change __addr type to void*.
40721 (_mm512_mask_i32scatter_ps): Ditto.
40722 (_mm512_i32scatter_pd): Ditto.
40723 (_mm512_mask_i32scatter_pd): Ditto.
40724 (_mm512_i64scatter_ps): Ditto.
40725 (_mm512_mask_i64scatter_ps): Ditto.
40726 (_mm512_i64scatter_pd): Ditto.
40727 (_mm512_mask_i64scatter_pd): Ditto.
40728 (_mm512_i32scatter_epi32): Ditto.
40729 (_mm512_mask_i32scatter_epi32): Ditto.
40730 (_mm512_i32scatter_epi64): Ditto.
40731 (_mm512_mask_i32scatter_epi64): Ditto.
40732 (_mm512_i64scatter_epi32): Ditto.
40733 (_mm512_mask_i64scatter_epi32): Ditto.
40734 (_mm512_i64scatter_epi64): Ditto.
40735 (_mm512_mask_i64scatter_epi64): Ditto.
40736 * config/i386/avx512pfintrin.h
40737 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40738 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40739 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40740 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40741 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40742 (_mm512_prefetch_i32scatter_ps): Ditto.
40743 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40744 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40745 (_mm512_prefetch_i64scatter_pd): Ditto.
40746 (_mm512_prefetch_i64scatter_ps): Ditto.
40747 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40748 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40749 * config/i386/avx512vlintrin.h
40750 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40751 (_mm_mmask_i32gather_ps): Ditto.
40752 (_mm256_mmask_i32gather_pd): Ditto.
40753 (_mm_mmask_i32gather_pd): Ditto.
40754 (_mm256_mmask_i64gather_ps): Ditto.
40755 (_mm_mmask_i64gather_ps): Ditto.
40756 (_mm256_mmask_i64gather_pd): Ditto.
40757 (_mm_mmask_i64gather_pd): Ditto.
40758 (_mm256_mmask_i32gather_epi32): Ditto.
40759 (_mm_mmask_i32gather_epi32): Ditto.
40760 (_mm256_mmask_i32gather_epi64): Ditto.
40761 (_mm_mmask_i32gather_epi64): Ditto.
40762 (_mm256_mmask_i64gather_epi32): Ditto.
40763 (_mm_mmask_i64gather_epi32): Ditto.
40764 (_mm256_mmask_i64gather_epi64): Ditto.
40765 (_mm_mmask_i64gather_epi64): Ditto.
40766 (_mm256_i32scatter_ps): Change __addr type to void*.
40767 (_mm256_mask_i32scatter_ps): Ditto.
40768 (_mm_i32scatter_ps): Ditto.
40769 (_mm_mask_i32scatter_ps): Ditto.
40770 (_mm256_i32scatter_pd): Ditto.
40771 (_mm256_mask_i32scatter_pd): Ditto.
40772 (_mm_i32scatter_pd): Ditto.
40773 (_mm_mask_i32scatter_pd): Ditto.
40774 (_mm256_i64scatter_ps): Ditto.
40775 (_mm256_mask_i64scatter_ps): Ditto.
40776 (_mm_i64scatter_ps): Ditto.
40777 (_mm_mask_i64scatter_ps): Ditto.
40778 (_mm256_i64scatter_pd): Ditto.
40779 (_mm256_mask_i64scatter_pd): Ditto.
40780 (_mm_i64scatter_pd): Ditto.
40781 (_mm_mask_i64scatter_pd): Ditto.
40782 (_mm256_i32scatter_epi32): Ditto.
40783 (_mm256_mask_i32scatter_epi32): Ditto.
40784 (_mm_i32scatter_epi32): Ditto.
40785 (_mm_mask_i32scatter_epi32): Ditto.
40786 (_mm256_i32scatter_epi64): Ditto.
40787 (_mm256_mask_i32scatter_epi64): Ditto.
40788 (_mm_i32scatter_epi64): Ditto.
40789 (_mm_mask_i32scatter_epi64): Ditto.
40790 (_mm256_i64scatter_epi32): Ditto.
40791 (_mm256_mask_i64scatter_epi32): Ditto.
40792 (_mm_i64scatter_epi32): Ditto.
40793 (_mm_mask_i64scatter_epi32): Ditto.
40794 (_mm256_i64scatter_epi64): Ditto.
40795 (_mm256_mask_i64scatter_epi64): Ditto.
40796 (_mm_i64scatter_epi64): Ditto.
40797 (_mm_mask_i64scatter_epi64): Ditto.
40798 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40799 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40800 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40801 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40802 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40803 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40804 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40805 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40806 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40807 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40808 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40809 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40810 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40811 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40812 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40813 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40814 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40815 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40816 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40817 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40818 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40819 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40820 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40821 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40822 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40823 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40824 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40825 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40826 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40827 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40828 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40829 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40830 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40831 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40832 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40833 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40834 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40835 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40836 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40837 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40838 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40839 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40840 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40841 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40842 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40843 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40844 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40845 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40846 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40847 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40848 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40849 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40850 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40851 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40852 definitions accordingly.
40854 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40855 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40858 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40861 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40864 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40865 hard regs for inner regno.
40867 2017-01-17 Martin Liska <mliska@suse.cz>
40870 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40871 assumption and add comment.
40873 2017-01-17 Nathan Sidwell <nathan@acm.org>
40875 * ipa-visibility.c (localize_node): New function, broken out of ...
40876 (function_and_variable_visibility): ... here. Call it.
40878 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40880 PR middle-end/77445
40881 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40882 correctly set frequency of oudgoing edge.
40883 (duplicate_thread_path): Fix profile updating.
40885 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40888 * configure.ac: Add GCC_BASE_VER.
40889 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40890 version from BASE-VER file.
40891 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40892 (gcc.o): Depend on $(BASEVER).
40893 * common.opt (dumpfullversion): New option.
40894 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40895 * doc/invoke.texi: Document -dumpfullversion.
40896 * doc/install.texi: Document --with-gcc-major-version-only.
40897 * configure: Regenerated.
40899 2017-01-17 Richard Biener <rguenther@suse.de>
40901 PR tree-optimization/71433
40902 * tree-vrp.c (register_new_assert_for): Merge same asserts
40903 on all incoming edges.
40904 (process_assert_insertions_for): Handle insertions at the
40907 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40909 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40910 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40912 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40915 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40918 2017-01-17 Alan Modra <amodra@gmail.com>
40921 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40922 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40923 symbolic stack limit when pic.
40925 2017-01-16 Martin Sebor <msebor@redhat.com>
40927 PR tree-optimization/78608
40928 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40930 2017-01-16 Jeff Law <law@redhat.com>
40933 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40934 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40935 for several include directories that may be relative to sysroot.
40936 * config/i386/x-mingw32 (gplus_includedir): Define.
40937 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40938 (native_system_includedir): Likewise.
40939 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40940 override if TARGET_SYSTEM_ROOT is defined.
40941 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40943 PR tree-optimization/79090
40944 PR tree-optimization/33562
40945 PR tree-optimization/61912
40946 PR tree-optimization/77485
40947 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40948 and computed trims into the dump file.
40950 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40952 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40954 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40957 * gimplify.c (gimplify_init_constructor): If want_value and
40958 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40962 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40963 sequence. Formatting fixes.
40964 (doloop_optimize): Formatting fixes.
40967 * gcc.c (debug_level_greater_than_spec_func): New function.
40968 (static_spec_functions): Add debug-level-gt spec function.
40969 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40971 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40972 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40973 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40974 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40975 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40976 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40978 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40980 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40981 QImode fixups to general and mask registers only.
40983 2017-01-16 Carl Love <cel@us.ibm.com>
40985 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40986 for built-in functions
40987 vector signed char vec_nabs (vector signed char)
40988 vector signed short vec_nabs (vector signed short)
40989 vector signed int vec_nabs (vector signed int)
40990 vector signed long long vec_nabs (vector signed long long)
40991 vector float vec_nabs (vector float)
40992 vector double vec_nabs (vector double)
40993 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40995 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40996 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40997 * doc/extend.texi: Update the documentation file for the new built-in
41000 2017-01-16 Martin Sebor <msebor@redhat.com>
41002 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
41005 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41007 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41008 UNSPEC_VSX__XXSPLTD to require special splat handling.
41010 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41013 * system.h: Poison strndup.
41015 2017-01-16 Alan Modra <amodra@gmail.com>
41018 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41021 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41023 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41025 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41027 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41028 call recog here. Assert that INSN_CODE (insn) is non-negative.
41030 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41033 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41035 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41036 in the currently scheduled RTL fragment.
41038 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41040 PR rtl-optimization/78751
41041 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41044 2017-01-14 Jeff Law <law@redhat.com>
41046 PR tree-optimization/79090
41047 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41048 variable length stores.
41049 (compute_trims): Delete dead assignment to *trim_tail.
41050 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41053 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41055 PR rtl-optimization/78626
41056 PR rtl-optimization/78727
41057 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41058 of a block, and split such blocks after everything else is finished.
41060 2017-01-14 Alan Modra <amodra@gmail.com>
41063 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41064 target legitimate_combined_insn.
41065 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41066 (rs6000_legitimate_combined_insn): New function.
41067 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41069 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41070 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41071 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41073 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41075 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41077 2017-01-13 Jeff Law <law@redhat.com>
41079 PR tree-optimization/33562
41080 PR tree-optimization/61912
41081 PR tree-optimization/77485
41082 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41084 (delete_dead_assignment): Likewise.
41085 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41086 statement to delete_dead_call and delete_dead_assignment.
41088 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41091 * substring-locations.c (format_warning_va): Strengthen case 1 so
41092 that both endpoints of the substring must be within the format
41093 range for just the substring to be printed.
41095 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41097 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41098 * config/i386/i386.c (ix86_target_string): Add missing options
41099 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41100 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41101 flags_other and ix86_target_other to flags2_other. Display unknown
41103 (ix86_valid_target_attribute_inner_p): Add missing options and
41104 reorder options by implied ISAs, as in ix86_target_string.
41106 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41108 * hash-table.h (hash_table::too_empty_p): New function.
41109 (hash_table::expand): Use it.
41110 (hash_table::traverse): Likewise.
41111 (hash_table::empty_slot): Use sizeof (value_type) instead of
41112 sizeof (PTR) to convert bytes to elements. Shrink the table
41113 if the current size is excessive for the current number of
41116 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41118 * ira-costs.c (record_reg_classes): Break from the inner loop
41119 early once alt_fail is known to be true. Update outer loop
41120 handling accordingly.
41122 2017-01-13 Jeff Law <law@redhat.com>
41124 * tree-ssa-dse.c (decrement_count): New function.
41125 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41126 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41127 when we know the partially dead statement is a mem* function.
41129 PR tree-optimization/61912
41130 PR tree-optimization/77485
41131 * tree-ssa-dse.c: Include expr.h.
41132 (maybe_trim_constructor_store): New function.
41133 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41135 PR tree-optimization/33562
41136 PR tree-optimization/61912
41137 PR tree-optimization/77485
41138 * doc/invoke.texi: Document new dse-max-object-size param.
41139 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41140 * tree-ssa-dse.c: Include params.h.
41141 (dse_store_status): New enum.
41142 (initialize_ao_ref_for_dse): New, partially extracted from
41144 (valid_ao_ref_for_dse, normalize_ref): New.
41145 (setup_live_bytes_from_ref, compute_trims): Likewise.
41146 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41147 (maybe_trim_partially_dead_store): Likewise.
41148 (maybe_trim_complex_store): Likewise.
41149 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41150 Track what bytes live from the original store. Return tri-state
41151 for dead, partially dead or live.
41152 (dse_dom_walker): Add constructor, destructor and new private members.
41153 (delete_dead_call, delete_dead_assignment): New extracted from
41155 (dse_optimize_stmt): Make a member of dse_dom_walker.
41156 Use initialize_ao_ref_for_dse.
41158 PR tree-optimization/33562
41159 PR tree-optimization/61912
41160 PR tree-optimization/77485
41161 * sbitmap.h (bitmap_count_bits): Prototype.
41162 (bitmap_clear_range, bitmap_set_range): Likewise.
41163 * sbitmap.c (bitmap_clear_range): New function.
41164 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41166 2017-01-13 Martin Liska <mliska@suse.cz>
41169 * function.c (set_cfun): Add new argument force.
41170 * function.h (set_cfun): Likewise.
41171 * ipa-inline-transform.c (inline_call): Use the function when
41172 strict alising from is dropped for function we inline to.
41174 2017-01-13 Richard Biener <rguenther@suse.de>
41176 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41177 for dumping GIMPLE INTEGER_CSTs.
41179 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41181 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41182 to 201112L since C++17.
41184 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41187 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41188 if -fsanitize=kernel-address is present.
41190 2017-01-13 Richard Biener <rguenther@suse.de>
41192 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41193 as _Literal ( type ) number in case usual suffixes do not
41194 preserve all information.
41196 2017-01-13 Richard Biener <rguenther@suse.de>
41198 PR tree-optimization/77283
41199 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41200 and ssa-iterators.h.
41201 (is_feasible_trace): Implement a cost model based on joiner
41204 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41207 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41208 char or short to __float128/_Float128 directly.
41210 2017-01-12 Martin Sebor <msebor@redhat.com>
41212 to -Wformat-overflow.
41213 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41214 (min_bytes_remaining): Same.
41215 (get_string_length): Same.
41216 (format_string): Same.
41217 (format_directive): Same.
41219 (pass_sprintf_length::handle_gimple_call): Same.
41221 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41223 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41224 info.nowrite calls with no lhs that can't throw. Return bool
41225 whether gsi_remove has been called or not.
41226 (pass_sprintf_length::handle_gimple_call): Return bool whether
41227 try_substitute_return_value called gsi_remove. Formatting fix.
41228 (pass_sprintf_length::execute): Don't use gsi_remove if
41229 handle_gimple_call returned true.
41232 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41233 be removed due to side-effects, don't remove following barrier nor
41234 turn the successor edge into fallthru edge.
41236 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41239 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41240 element-reversing loads and stores as not swappable.
41242 2017-01-12 Nathan Sidwell <nathan@acm.org>
41243 Nicolai Stange <nicstange@gmail.com>
41245 * combine.c (try_combine): Don't ignore result of overlap checking
41246 loop. Combine overlap & asm check into single loop.
41248 2017-01-12 Richard Biener <rguenther@suse.de>
41250 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41251 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41253 2017-01-12 Richard Biener <rguenther@suse.de>
41255 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41256 and TS_TARGET_OPTION directly derive from TS_BASE.
41257 * tree-core.h (tree_optimization_option): Derive from tree_base.
41258 (tree_target_option): Likewise.
41260 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41262 * config/i386/i386.c (memory_address_length): Increase len
41263 only when rip_relative_addr_p returns false.
41265 2017-01-11 Julia Koval <julia.koval@intel.com>
41267 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41268 (OPTION_MASK_ISA_SGX_SET): New.
41269 (ix86_handle_option): Handle OPT_msgx.
41270 * config.gcc: Added sgxintrin.h.
41271 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41272 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41273 * config/i386/i386.c (ix86_target_string): Add -msgx.
41275 (ix86_option_override_internal): Handle new options.
41276 (ix86_valid_target_attribute_inner_p): Add sgx.
41277 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41278 * config/i386/i386.opt: Add msgx.
41279 * config/i386/sgxintrin.h: New file.
41280 * config/i386/x86intrin.h: Add sgxintrin.h.
41282 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41285 * fold-const.c (maybe_nonzero_address): Return 1 for function
41287 (tree_single_nonzero_warnv_p): Don't handle function local objects
41291 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41294 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41297 * opts.c: Include "spellcheck.h"
41298 (struct string_fragment): New struct.
41299 (struct edit_distance_traits<const string_fragment &>): New
41301 (get_closest_sanitizer_option): New function.
41302 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41304 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41306 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41308 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41309 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41310 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41311 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41312 for initial die_offset if dwarf_split_debug_info.
41313 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41314 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41316 (output_skeleton_debug_sections): Formatting fix. Use
41317 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41318 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41320 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41322 * config/arm/cortex-a53.md: Add bypasses for
41323 cortex_a53_r2f_cvt.
41324 (cortex_a53_r2f): Only use for transfers.
41325 (cortex_a53_f2r): Likewise.
41326 (cortex_a53_r2f_cvt): Add reservation for conversions.
41327 (cortex_a53_f2r_cvt): Likewise.
41329 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41331 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41332 to all inlined functions, change static to extern.
41334 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41337 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41339 (arm_assemble_integer): Likewise.
41341 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41343 * config.gcc: Use new awk script to check CPU, FPU and architecture
41344 parameters for --with-... options.
41345 * config/arm/parsecpu.awk: New file
41346 * config/arm/arm-cpus.in: New file.
41347 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41349 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41351 * config/arm/t-arm: Update dependency rules.
41352 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41353 of processing .def files.
41354 * config/arm/genopt.sh: Deleted.
41355 * config/arm/gentune.sh: Deleted.
41356 * config/arm/arm-cores.def: Deleted.
41357 * config/arm/arm-arches.def: Deleted.
41358 * config/arm/arm-fpus.def: Deleted.
41359 * config/arm/arm-tune.md: Regenerated.
41360 * config/arm/arm-tables.opt: Regenerated.
41361 * config/arm/arm-cpu.h: New generated file.
41362 * config/arm/arm-cpu-data.h: New generated file.
41363 * config/arm/arm-cpu-cdata.h: New generated file.
41365 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41368 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41370 (input_varpool_node): Unpack dynamically_initialized bit.
41372 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41374 PR rtl-optimization/79032
41375 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41376 the alignment of the adjusted memory reference against that of MODE,
41377 instead of the alignment of the original memory reference.
41379 2017-01-11 Martin Jambor <mjambor@suse.cz>
41381 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41383 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41384 decorated functions.
41386 2017-01-11 Richard Biener <rguenther@suse.de>
41388 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41389 set range/nonnull info for PHI results. Do not set it on
41390 stmts marked for removal.
41392 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41394 * expr.c (store_field): In the bitfield case, fetch the return value
41395 from the registers before applying a single big-endian adjustment.
41396 Always do a final load for a BLKmode value not larger than a word.
41398 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41401 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41402 that we correctly handle column numbers greater than
41403 LINE_MAP_MAX_COLUMN_NUMBER.
41405 2017-01-10 Martin Sebor <msebor@redhat.com>
41407 PR middle-end/78245
41408 * gimple-ssa-sprintf.c (get_destination_size): Call
41409 {init,fini}object_sizes.
41410 * tree-object-size.c (addr_object_size): Adjust.
41411 (pass_through_call): Adjust.
41412 (pass_object_sizes::execute): Adjust.
41413 * tree-object-size.h (fini_object_sizes): Declare.
41415 2017-01-10 Martin Sebor <msebor@redhat.com>
41417 PR tree-optimization/78775
41418 * builtins.c (get_size_range): Move...
41419 * calls.c: ...to here.
41420 (alloc_max_size): Accept zero argument.
41421 (operand_signed_p): Remove.
41422 (maybe_warn_alloc_args_overflow): Call get_size_range.
41423 * calls.h (get_size_range): Declare.
41425 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41427 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41428 from TI's devices.csv file as of September 2016.
41429 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41431 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41433 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41434 * doc/invoke.texi: Likewise.
41435 * doc/md.texi: Likewise.
41436 * doc/objc.texi: Likewise.
41438 2017-01-10 Joshua Conner <joshconner@google.com>
41440 * config/arm/fuchsia-elf.h: New file.
41441 * config/fuchsia.h: New file.
41442 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41443 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41445 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41447 2016-01-10 Richard Biener <rguenther@suse.de>
41449 PR tree-optimization/79034
41450 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41451 Propagate out degenerate PHIs in the joiner.
41453 2017-01-10 Martin Liska <mliska@suse.cz>
41455 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41456 (sort_congruence_classes_by_decl_uid): Likewise.
41457 (sort_congruence_class_groups_by_decl_uid): Likewise.
41458 (sem_item_optimizer::merge_classes): Sort class, groups in these
41459 classes and members in the groups by DECL_UID of declarations.
41460 This would make merge operations stable.
41462 2017-01-10 Martin Liska <mliska@suse.cz>
41464 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41465 usage of m_classes_vec.
41466 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41467 (sem_item_optimizer::get_group_by_hash): Likewise.
41468 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41469 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41470 (sem_item_optimizer::verify_classes): Likewise.
41471 (sem_item_optimizer::process_cong_reduction): Likewise.
41472 (sem_item_optimizer::dump_cong_classes): Likewise.
41473 (sem_item_optimizer::merge_classes): Likewise.
41474 * ipa-icf.h (congruence_class_hash): Rename from
41475 congruence_class_group_hash. Remove declaration of m_classes_vec.
41477 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41479 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41480 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41481 * config.gcc: Add avx512vpopcntdqintrin.h.
41482 * config/i386/avx512vpopcntdqintrin.h: New.
41483 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41484 * config/i386/i386-builtin-types.def: Add new types.
41485 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41486 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41487 __builtin_ia32_vpopcountq_v8di_mask): New.
41488 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41489 __AVX512VPOPCNTDQ__.
41490 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41491 (PTA_AVX512VPOPCNTDQ): Define.
41492 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41493 TARGET_AVX512VPOPCNTDQ_P): Define.
41494 * config/i386/i386.opt: Add mavx512vpopcntdq.
41495 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41496 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41498 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41500 PR middle-end/77484
41501 * predict.def (PRED_CALL): Set to 67.
41503 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41505 * expr.c (store_field): In the bitfield case, if the value comes from
41506 a function call and is of an aggregate type returned in registers, do
41507 not modify the field mode; extract the value in all cases if the mode
41508 is BLKmode and the size is not larger than a word.
41510 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41513 * config/i386/cpuid.h: Fix undefined behavior.
41515 2017-01-04 Jeff Law <law@redhat.com>
41517 PR tree-optimization/79007
41518 PR tree-optimization/67955
41519 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41520 conservative for pt.null when flag_non_call_exceptions is on.
41522 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41524 PR translation/79019
41525 PR translation/79020
41526 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41527 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41528 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41530 * config/avr/avr.opt (maccumulate-args): Likewise.
41531 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41532 * common.opt (freport-bug): Likewise.
41533 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41534 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41535 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41536 translatable string.
41537 * config/i386/i386.c (function_value_32): Likewise.
41538 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41539 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41541 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41542 * common/config/msp430/msp430-common.c (msp430_handle_option):
41544 * symtab.c (symtab_node::verify_base): Likewise.
41545 * opts.c (set_debug_level): Likewise.
41546 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41547 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41548 missing whitespace to translatable strings.
41549 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41550 * config/sh/superh.h: Likewise.
41551 * config/i386/xopintrin.h: Likewise.
41552 * config/i386/znver1.md: Likewise.
41553 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41554 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41555 * double-int.h (struct double_int): Likewise.
41556 * double-int.c (div_and_round_double): Likewise.
41557 * wide-int.cc: Likewise.
41558 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41559 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41560 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41561 (crossjumps_occurred): ... this.
41562 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41565 PR tree-optimization/78899
41566 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41567 returning bool return struct loop *, NULL for failure and the new
41569 (versionable_outer_loop_p): Don't version outer loop if it has
41570 dont_vectorized bit set.
41571 (tree_if_conversion): When versioning outer loop, ensure
41572 tree_if_conversion is performed also on the inner loop of the
41573 non-vectorizable outer loop copy.
41574 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41575 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41576 prevent vectorization of it.
41577 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41578 the outer loop vectorization of the non-scalar version is attempted
41579 before vectorization of the inner loop in scalar version. If
41580 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41581 vectorization of its inner loop.
41582 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41583 has 2 inner loops, rename also on edges from bb whose single pred
41584 is outer_loop->header. Fix typo in function comment.
41586 2017-01-09 Martin Sebor <msebor@redhat.com>
41589 * asan.c (asan_emit_stack_protection): Increase local buffer size
41590 to avoid snprintf truncation warning.
41592 2017-01-09 Andrew Pinski <apinski@cavium.com>
41594 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41595 to reference thunderx2t99 for the tuning structure
41596 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41598 (thunderx2t99_extra_costs): This.
41599 * config/aarch64/aarch64-tune.md: Regenerate.
41600 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41601 (vulcan_addrcost_table): This.
41602 (vulcan_regmove_cost): Rename to ...
41603 (thunderx2t99_regmove_cost): This.
41604 (vulcan_vector_cost): Rename to ...
41605 (thunderx2t99_vector_cost): this.
41606 (vulcan_branch_cost): Rename to ...
41607 (thunderx2t99_branch_cost): This.
41608 (vulcan_tunings): Rename to ...
41609 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41610 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41612 2017-01-09 Martin Jambor <mjambor@suse.cz>
41616 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41617 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41618 (propagate_vr_accross_jump_function): Use the above function for all
41619 value range computations for pass-through jump functions and type
41620 converasion from explicit value range values.
41621 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41622 parameters from TYPE_ARG_TYPES.
41623 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41624 (ipa_write_node_info): Stream type of the actual argument.
41625 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41627 2017-01-09 Martin Liska <mliska@suse.cz>
41630 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41631 (lookup_compiler): Do not show error message with have_E.
41633 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41635 PR tree-optimization/78938
41636 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41637 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41638 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41639 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41642 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41644 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41647 2017-01-09 Richard Biener <rguenther@suse.de>
41649 PR tree-optimization/78997
41650 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41651 name condition properly.
41653 2017-01-09 Richard Biener <rguenther@suse.de>
41656 * dwarf2out.c (is_cxx): New overload with context.
41657 (is_naming_typedef_decl): Use it.
41659 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41661 * invoke.texi (Option Summary): Correct spacing in option lists
41662 and add line breaks to fix over-long lines.
41664 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41666 PR middle-end/17660
41668 * extend.texi (Common Variable Attributes): Add xref to GCC
41669 Internals manual to explain mode attribute keywords.
41671 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41674 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41675 and Preprocessor Options.
41676 (Options for Linking): Document -pthread here....
41677 (RS/6000 and PowerPC Options): ...not here.
41678 (Solaris 2 Options): ...or here.
41679 * doc/cppopts.texi: Document -pthread.
41681 2017-01-08 Martin Sebor <msebor@redhat.com>
41683 PR middle-end/77708
41684 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41685 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41686 New member functions.
41687 (format_directive): Used them.
41689 (pass_sprintf_length::handle_gimple_call): Same.
41690 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41691 to avoid truncation for any argument.
41692 (extract_affine_mul): Same.
41693 * tree.c (get_file_function_name): Same.
41695 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41697 PR middle-end/77484
41698 * predict.def (PRED_INDIR_CALL): Set to 86.
41700 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41702 PR preprocessor/54124
41703 * doc/cppopts.texi: Reformat -d subtable to list the full name
41704 of the options. Add cross-reference to the docs for the general
41705 compiler -d options.
41706 * doc/invoke.texi (Developer Options): Add cross-reference to the
41707 preprocessor-specific -d option documentation.
41709 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41711 PR preprocessor/13498
41712 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41713 redudant material, and reflect new command-line options.
41714 (System Headers): Likewise.
41716 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41718 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41719 -isystem, and -idirafter. Copy-edit.
41720 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41721 default for -ftrack-macro-expansion. Delete obsolete and
41722 badly-formatted implementation details about -fdebug-cpp output.
41723 * doc/cppwarnopts.texi: Copy-edit.
41725 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41728 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41729 that the transition from a max line width >= 1<<10 to narrower
41730 lines works correctly.
41732 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41734 * doc/options.texi (PerFunction): New.
41735 * opt-functions.awk (switch_flags): Map both Optimization and
41736 PerFunction to CL_OPTIMIZATION.
41737 * opth-gen.awk: Test for PerFunction flag along with
41739 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41740 it only when the latter is present. Skip those that don't in
41741 the hash function generator.
41742 * common.opt (fvar-tracking): Mark as PerFunction instead of
41744 (fvar-tracking-assignments): Likewise.
41745 (fvar-tracking-assignments-toggle): Likewise.
41746 (fvar-tracking-uninit): Likewise.
41748 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41750 PR translation/79018
41751 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41754 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41757 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41758 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41759 TARGET_LONG_JUMP_TABLE_OFFSETS.
41760 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41761 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41762 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41763 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41764 * config/m68k/m68k.md (tablejump expander): Likewise.
41765 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41766 TARGET_LONG_JUMP_TABLE_OFFSETS.
41767 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41768 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41770 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41771 David Holsgrove <david.holsgrove@xilinx.com>
41773 * common/config/microblaze/microblaze-common.c
41774 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41775 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41777 * config/microblaze/microblaze.c (microblaze_must_save_register)
41778 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41780 (microblaze_eh_return): New function.
41781 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41782 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41783 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41784 * config/microblaze/microblaze.md (eh_return): New pattern.
41786 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41788 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41789 GCC_DIAGNOSTIC_STRINGIFY): Define.
41791 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41793 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41795 * config/arm/arm.md (<mcrr>): New.
41797 * config/arm/arm.c (arm_arch5te): New.
41798 (arm_option_override): Set arm_arch5te.
41799 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41801 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41802 (arm_mcrr_qualifiers): ... this. New.
41803 (MRRC_QUALIFIERS): Define to...
41804 (arm_mrrc_qualifiers): ... this. New.
41805 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41807 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41808 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41809 (MRRCI, mrrc, MRRC): New.
41810 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41811 VUNSPEC_MRRC2): New.
41813 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41815 * config/arm/arm.md (<mcr>): New.
41817 * config/arm/arm.c (arm_coproc_builtin_available): Add
41818 support for mcr, mrc, mcr2 and mrc2.
41819 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41820 (arm_mcr_qualifiers): ... this. New.
41821 (MRC_QUALIFIERS): Define to ...
41822 (arm_mrc_qualifiers): ... this. New.
41823 (MCR_QUALIFIERS): Define to ...
41824 (arm_mcr_qualifiers): ... this. New.
41825 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41827 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41828 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41829 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41830 VUNSPEC_MRC2): New.
41832 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41834 * config/arm/arm.md (*ldc): New.
41838 * config/arm/arm.c (arm_coproc_builtin_available): Add
41839 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41840 (arm_coproc_ldc_stc_legitimate_address): New.
41841 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41842 'qualifier_const_pointer'.
41843 (LDC_QUALIFIERS): Define to...
41844 (arm_ldc_qualifiers): ... this. New.
41845 (STC_QUALIFIERS): Define to...
41846 (arm_stc_qualifiers): ... this. New.
41847 * config/arm/arm-protos.h
41848 (arm_coproc_ldc_stc_legitimate_address): New.
41849 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41850 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41851 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41852 stc2, stcl, stc2l): New.
41853 * config/arm/constraints.md (Uz): New.
41854 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41855 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41856 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41857 VUNSPEC_STC2L): New.
41859 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41861 * config/arm/arm.md (<cdp>): New.
41862 * config/arm/arm.c (neon_const_bounds): Rename this ...
41863 (arm_const_bounds): ... this.
41864 (arm_coproc_builtin_available): New.
41865 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41866 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41867 (CDP_QUALIFIERS): Define to...
41868 (arm_cdp_qualifiers): ... this. New.
41870 (arm_expand_builtin_args): Add case for 6 arguments.
41871 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41872 (arm_const_bounds): ... this.
41873 (arm_coproc_builtin_available): New.
41874 * config/arm/arm_acle.h (__arm_cdp): New.
41876 * config/arm/arm_acle_builtins.def (cdp): New.
41878 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41879 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41880 'arm_const_bounds'.
41881 * config/arm/types.md (coproc): New.
41882 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41883 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41884 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41885 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41887 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41889 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41890 (UBINOP_QUALIFIERS): New.
41892 (acle_builtin_data): New. Change comment.
41893 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41894 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41895 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41896 arm_acle_builtins.def.
41897 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41898 (arm_init_acle_builtins): New.
41899 (CRC32_BUILTIN): Remove.
41900 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41901 crc32cb, crc32ch and crc32cw.
41902 (arm_init_crc32_builtins): Remove.
41903 (arm_init_builtins): Use arm_init_acle_builtins rather
41904 than arm_init_crc32_builtins.
41905 (arm_expand_acle_builtin): New.
41906 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41907 * config/arm/arm_acle_builtins.def: New.
41909 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41911 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41912 (arm_builtin_datum): ... this.
41913 (arm_init_neon_builtin): Rename to ...
41914 (arm_init_builtin): ... this. Add a new parameters PREFIX
41915 and USE_SIG_IN_NAME.
41916 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41917 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41918 'arm_builtin_datum'.
41919 (arm_init_vfp_builtins): Likewise.
41920 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41921 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41922 (arm_expand_neon_args): Rename to ...
41923 (arm_expand_builtin_args): ... this. Rename builtin_arg
41924 enum values and differentiate between ARG_BUILTIN_MEMORY
41925 and ARG_BUILTIN_NEON_MEMORY.
41926 (arm_expand_neon_builtin_1): Rename to ...
41927 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41928 values, arm_expand_builtin_args and add bool parameter NEON.
41929 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41930 (arm_expand_vfp_builtin): Likewise.
41931 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41933 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41935 PR middle-end/77484
41936 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41937 * predict.c (tree_estimate_probability_bb): Reverse direction of
41938 polymorphic call predictor.
41940 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41942 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41943 (determine_pass_name_match): ...this new function and...
41944 (should_skip_pass_p): ...this new function.
41946 2017-01-06 Nathan Sidwell <nathan@acm.org>
41948 * ipa-visibility.c (function_and_variable_visibility): Reformat
41949 comments and long lines. Remove extrneous if.
41950 * symtab.c (symtab_node::make_decl_local): Fix code format.
41951 (symtab_node::set_section_for_node): Fix comment typo.
41953 2017-01-06 Martin Liska <mliska@suse.cz>
41956 * lra-constraints.c: Rename invariant to lra_invariant.
41957 * predict.c (set_even_probabilities): Initialize e to NULL.
41959 2017-01-05 Martin Sebor <msebor@redhat.com>
41961 PR tree-optimization/78910
41962 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41963 (format_integer): Correct off-by-one error in the handling
41964 of precision with negative numbers in signed conversions..
41966 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41968 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41970 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41972 PR tree-optimization/71016
41973 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41974 factor_out_conditional_conversion. Formatting fix.
41975 (factor_out_conditional_conversion): Add cond_stmt argument.
41976 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41977 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41980 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41982 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41983 read-rtl-function.o, and selftest-rtl.o.
41984 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41985 (selftest::aarch64_test_loading_full_dump): New function.
41986 (selftest::aarch64_run_selftests): New function.
41987 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41988 selftest::aarch64_run_selftests.
41989 * config/i386/i386.c
41990 (selftest::ix86_test_loading_dump_fragment_1): New function.
41991 (selftest::ix86_test_loading_call_insn): New function.
41992 (selftest::ix86_test_loading_full_dump): New function.
41993 (selftest::ix86_test_loading_unspec): New function.
41994 (selftest::ix86_run_selftests): Call the new functions.
41995 * emit-rtl.c (maybe_set_max_label_num): New function.
41996 * emit-rtl.h (maybe_set_max_label_num): New decl.
41997 * function.c (instantiate_decls): Guard call to
41998 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41999 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
42001 * gensupport.c (gen_reader::gen_reader): Pass "false"
42002 for new "compact" param of rtx_reader.
42003 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
42004 rather than an empty string for NULL strings.
42005 * read-md.c: Potentially include config.h rather than bconfig.h.
42006 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42007 (have_error): New global, copied from errors.c.
42008 (md_reader::read_name): Rename to...
42009 (md_reader::read_name_1): ...this, adding "out_loc" param,
42010 and converting "missing name or number" to returning false, rather
42012 (md_reader::read_name): Reimplement in terms of read_name_1.
42013 (md_reader::read_name_or_nil): New function.
42014 (md_reader::read_string): Handle "(nil)" by returning NULL.
42015 (md_reader::md_reader): Add new param "compact".
42016 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42017 (md_reader::read_file): New method.
42018 * read-md.h (md_reader::md_reader): Add new param "compact".
42019 (md_reader::read_file): New method.
42020 (md_reader::is_compact): New accessor.
42021 (md_reader::read_name): Convert return type from void to file_location.
42022 (md_reader::read_name_or_nil): New decl.
42023 (md_reader::read_name_1): New decl.
42024 (md_reader::m_compact): New field.
42025 (noop_reader::noop_reader): Pass "false" for new "compact" param
42027 (rtx_reader::rtx_reader): Add new "compact" param.
42028 (rtx_reader::read_rtx_operand): Make virtual and convert return
42029 type from void to rtx.
42030 (rtx_reader::read_until): New decl.
42031 (rtx_reader::handle_any_trailing_information): New virtual function.
42032 (rtx_reader::postprocess): New virtual function.
42033 (rtx_reader::finalize_string): New virtual function.
42034 (rtx_reader::m_in_call_function_usage): New field.
42035 (rtx_reader::m_reuse_rtx_by_id): New field.
42036 * read-rtl-function.c: New file.
42037 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42038 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42039 (selftest::verify_three_block_rtl_cfg): New decl.
42040 * read-rtl-function.h: New file.
42041 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42042 For host, include function.h, memmodel.h, and emit-rtl.h.
42043 (one_time_initialization): New function.
42044 (struct compact_insn_name): New struct.
42045 (compact_insn_names): New array.
42046 (find_code): Handle insn codes in compact dumps.
42047 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42048 (bind_subst_iter_and_attr): Likewise.
42049 (add_condition_to_string): Likewise.
42050 (add_condition_to_rtx): Likewise.
42051 (apply_attribute_uses): Likewise.
42052 (add_current_iterators): Likewise.
42053 (apply_iterators): Likewise.
42054 (initialize_iterators): Guard usage of apply_subst_iterator with
42055 #ifdef GENERATOR_FILE.
42056 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42057 (md_reader::read_mapping): Likewise.
42058 (add_define_attr_for_define_subst): Likewise.
42059 (add_define_subst_attr): Likewise.
42060 (read_subst_mapping): Likewise.
42061 (check_code_iterator): Likewise.
42062 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42064 (one_time_initialization): New function.
42065 (rtx_reader::read_until): New method.
42066 (read_flags): New function.
42067 (parse_reg_note_name): New function.
42068 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42069 Handle reuse_rtx ids.
42070 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42071 Add parsing support for RTL dumps, mirroring the special-cases in
42072 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42073 values, and calling handle_any_trailing_information.
42074 (rtx_reader::read_rtx_operand): Convert return type from void
42075 to rtx, returning return_rtx. Handle case 'e'. Call
42076 finalize_string on XSTR and XTMPL fields.
42077 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42078 "(nil)" values were omitted. Call the postprocess vfunc on the
42080 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42081 class ctor. Initialize m_in_call_function_usage. Call
42082 one_time_initialization.
42083 * rtl-tests.c (selftest::test_uncond_jump): Call
42084 set_new_first_and_last_insn.
42085 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42086 * selftest-rtl.c: New file.
42087 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42088 (selftest::get_insn_by_uid): New decl.
42089 * selftest-run-tests.c (selftest::run_tests): Call
42090 read_rtl_function_c_tests.
42091 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42092 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42095 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42097 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42098 operands in a special way. Assert that pos+len <= mode precision.
42100 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42102 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42103 3 argument Alias with unlimited for the negative form.
42104 (fno-vect-cost-model): Removed.
42106 2017-01-05 Martin Liska <mliska@suse.cz>
42108 * hsa-gen.c (gen_hsa_divmod): New function.
42109 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42111 2017-01-05 Martin Liska <mliska@suse.cz>
42114 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42117 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42119 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42120 small constant length operands.
42122 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42124 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42125 between loop iterations.
42127 2017-01-05 Martin Liska <mliska@suse.cz>
42130 * gimplify.c (gimplify_decl_expr): Compare to
42131 asan_poisoned_variables instread of checking flags.
42132 (gimplify_target_expr): Likewise.
42133 (gimplify_expr): Likewise.
42134 (gimplify_function_tree): Conditionally initialize
42135 asan_poisoned_variables.
42137 2017-01-04 Jeff Law <law@redhat.com>
42139 PR tree-optimizatin/78812
42140 * rtl.h (contains_mem_rtx_p): Prototype.
42141 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42142 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42143 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42144 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42145 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42147 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42149 * input.c (assert_char_at_range): Default-initialize actual_range.
42151 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42153 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42154 to match the caller.
42156 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42158 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42159 insns after final jump in test to emit dummy move.
42161 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42163 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42164 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42166 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42168 * multiple_target.c (create_dispatcher_calls): Init e_next.
42169 * tree-ssa-loop-split.c (split_loop): Init border.
42170 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42173 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42178 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42179 (altivec_register_operand): Do not return true if the operand
42180 contains a SUBREG mixing SImode and SFmode.
42181 (vsx_register_operand): Likewise.
42182 (vsx_reg_sfsubreg_ok): New predicate.
42183 (vfloat_operand): Do not return true if the operand contains a
42184 SUBREG mixing SImode and SFmode.
42185 (vint_operand): Likewise.
42186 (vlogical_operand): Likewise.
42187 (gpc_reg_operand): Likewise.
42188 (int_reg_operand): Likewise.
42189 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42190 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42191 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42193 (rs6000_emit_move_si_sf_subreg): New helper function.
42194 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42195 fixup SUBREGs involving SImode and SFmode.
42196 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42197 numbers for the new peephole2 optimization.
42198 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42199 the GLIBC math library that do AND/IOR/XOR operations on single
42200 precision floating point.
42201 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42202 target macros to say whether we need to avoid SUBREGs mixing
42204 (TARGET_ALLOW_SF_SUBREG): Likewise.
42205 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42206 (UNSPEC_SI_FROM_SF): Likewise.
42207 (iorxor): Change spacing.
42208 (and_ior_xor): New iterator for AND, IOR, and XOR.
42209 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42210 (movdi_from_sf_zero_ext): Likewise.
42211 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42212 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42213 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42214 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42215 (fms<mode>4): Likewise.
42216 (fnma<mode>4): Likewise.
42217 (fnms<mode>4): Likewise.
42218 (nfma<mode>4): Likewise.
42219 (nfms<mode>4): Likewise.
42221 2017-01-04 Marek Polacek <polacek@redhat.com>
42224 * doc/invoke.texi: Document -Wpointer-compare.
42226 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42228 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42231 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42232 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42235 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42238 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42239 documentation of the powerpc_popcntb_ok attribute.
42240 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42241 code to issue warning messages if a requested CPU configuration is
42242 not supported by the binary (assembler and loader) toolchain.
42243 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42244 made to define a built-in function that has been disabled.
42245 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42246 made to define a built-in function that has been disabled.
42247 (altivec_init_builtins): Add comment explaining why definition
42248 of the DST built-in functions is not preceded by an assertion
42249 check. Add assertions to prevent ICE if attempts are made to
42250 define an altivec predicate or an abs* built-in function that has
42252 (htm_init_builtins): Add comment explaining why definition of the
42253 htm built-in functions is not preceded by an assertion check.
42255 2017-01-04 Jeff Law <law@redhat.com>
42257 PR tree-optimizatin/67955
42258 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42259 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42260 the points-to solution does not include pt_null. Use DECL_PT_UID
42263 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42265 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42266 Use gen_int_mode instead of gen_lopwart for const_int operands.
42268 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42270 PR tree-optimization/71563
42271 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42272 out of range value - has low bits known to be zero.
42274 2017-01-04 Alan Modra <amodra@gmail.com>
42276 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42277 * configure: Regenerate.
42278 * config.in: Regenerate.
42280 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42283 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42284 a substring of the message, but strcmp with the whole message. Ifdef
42285 ENABLE_NLS, translate the message first using dgettext.
42287 2017-01-03 Jeff Law <law@redhat.com>
42289 PR tree-optimizatin/78856
42290 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42291 (mark_threaded_blocks): Remove code to truncate thread paths that
42292 cross multiple loop headers. Instead invalidate the cached loop
42293 iteration information and handle case of a thread path walking
42294 into an irreducible region.
42296 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42299 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42300 assertions. Add support for doing the signbit if the IEEE 128-bit
42301 floating point value is in a GPR.
42302 * config/rs6000/rs6000.md (Fsignbit): Delete.
42303 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42304 Update the length attribute if the value is in a GPR.
42305 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42306 the sign or zero extension instruction, since the value is always 0/1.
42307 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42310 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42311 extracting SImode to a GPR register so that we can generate a
42312 store, limit the vector to be in a traditional Altivec register
42313 for the vextuwrx instruction.
42315 2017-01-03 Ian Lance Taylor <iant@google.com>
42317 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42319 2017-01-03 Martin Sebor <msebor@redhat.com>
42321 PR tree-optimization/78696
42322 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42323 precision. Use MPFR for %f for greater fidelity. Correct handling
42325 (pass_sprintf_length::compute_format_length): Set width and precision
42326 specified by asrerisk to void_node for vararg functions.
42327 (try_substitute_return_value): Adjust dump output.
42329 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42331 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42333 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42335 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42336 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42337 -mlra/-mno-lra was passed to the compiler.
42339 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42341 PR rtl-optimization/65618
42342 * emit-rtl.c (try_split): Move initialization of "before" and
42343 "after" to just before the call to emit_insn_after_setloc.
42345 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42347 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42349 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42351 * dwarf2out.c (gen_enumeration_type_die): When
42352 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42354 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42356 PR tree-optimization/78965
42357 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42358 Change first argument from const call_info & to call_info &. For %n
42359 set info.nowrite to false.
42361 PR middle-end/78901
42362 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42363 possibly throwing calls.
42365 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42366 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42367 and fns handling, rather than in a separate case SSA_NAME.
42369 2017-01-02 Jeff Law <law@redhat.com>
42371 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42372 fixes for first_period and second_period variables.
42374 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42377 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42378 (*insvqi_1): New insn pattern.
42379 (*insvqi_1_mem_rex64): Ditto.
42380 (*insvqi_2): Ditto.
42381 (*insvqi_3): Rename from *insvqi.
42383 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42385 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42387 * doc/cfg.texi (Edges): Remove reference to Java.
42388 (Maintaining the CFG): Ditto.
42390 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42392 PR middle-end/77674
42393 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42394 transparent aliases.
42396 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42398 PR middle-end/77484
42399 * predict.def (PRED_CALL): Update hitrate.
42400 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42401 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42402 into direct/indirect/polymorphic variants.
42404 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42406 Update copyright years.
42408 * gcc.c (process_command): Update copyright notice dates.
42409 * gcov-dump.c (print_version): Ditto.
42410 * gcov.c (print_version): Ditto.
42411 * gcov-tool.c (print_version): Ditto.
42412 * gengtype.c (create_file): Ditto.
42413 * doc/cpp.texi: Bump @copying's copyright year.
42414 * doc/cppinternals.texi: Ditto.
42415 * doc/gcc.texi: Ditto.
42416 * doc/gccint.texi: Ditto.
42417 * doc/gcov.texi: Ditto.
42418 * doc/install.texi: Ditto.
42419 * doc/invoke.texi: Ditto.
42421 Copyright (C) 2017 Free Software Foundation, Inc.
42423 Copying and distribution of this file, with or without modification,
42424 are permitted in any medium without royalty provided the copyright
42425 notice and this notice are preserved.