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-stmts.c (vectorizable_call): Treat the number of
6 vectors as polynomial. Use build_index_vector for
9 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
10 Alan Hayward <alan.hayward@arm.com>
11 David Sherwood <david.sherwood@arm.com>
13 * tree-vect-stmts.c (get_load_store_type): Treat the number of
14 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
15 for variable-length vectors.
16 (vectorizable_mask_load_store): Treat the number of units as
17 polynomial, asserting that it is constant if the condition has
18 already been enforced.
19 (vectorizable_store, vectorizable_load): Likewise.
21 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22 Alan Hayward <alan.hayward@arm.com>
23 David Sherwood <david.sherwood@arm.com>
25 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
26 of units as polynomial. Punt if we can't tell at compile time
27 which vector contains the final result.
29 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30 Alan Hayward <alan.hayward@arm.com>
31 David Sherwood <david.sherwood@arm.com>
33 * tree-vect-loop.c (vectorizable_induction): Treat the number
34 of units as polynomial. Punt on SLP inductions. Use an integer
35 VEC_SERIES_EXPR for variable-length integer reductions. Use a
36 cast of such a series for variable-length floating-point
39 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
40 Alan Hayward <alan.hayward@arm.com>
41 David Sherwood <david.sherwood@arm.com>
43 * tree.h (build_index_vector): Declare.
44 * tree.c (build_index_vector): New function.
45 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
46 of units as polynomial, forcibly converting it to a constant if
47 vectorizable_reduction has already enforced the condition.
48 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
49 to create a {1,2,3,...} vector.
50 (vectorizable_reduction): Treat the number of units as polynomial.
51 Choose vectype_in based on the largest scalar element size rather
52 than the smallest number of units. Enforce the restrictions
55 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
56 Alan Hayward <alan.hayward@arm.com>
57 David Sherwood <david.sherwood@arm.com>
59 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
60 number of units as polynomial.
62 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
63 Alan Hayward <alan.hayward@arm.com>
64 David Sherwood <david.sherwood@arm.com>
66 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
67 * target.def (autovectorize_vector_sizes): Return the vector sizes
68 by pointer, using vector_sizes rather than a bitmask.
69 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
70 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
71 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
73 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
74 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
75 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
76 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
77 * omp-general.c (omp_max_vf): Likewise.
78 * omp-low.c (omp_clause_aligned_alignment): Likewise.
79 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
80 * tree-vect-loop.c (vect_analyze_loop): Likewise.
81 * tree-vect-slp.c (vect_slp_bb): Likewise.
82 * doc/tm.texi: Regenerate.
83 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
85 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
86 the vector size as a poly_uint64 rather than an unsigned int.
87 (current_vector_size): Change from an unsigned int to a poly_uint64.
88 (get_vectype_for_scalar_type): Update accordingly.
89 * tree.h (build_truth_vector_type): Take the size and number of
90 units as a poly_uint64 rather than an unsigned int.
91 (build_vector_type): Add a temporary overload that takes
92 the number of units as a poly_uint64 rather than an unsigned int.
93 * tree.c (make_vector_type): Likewise.
94 (build_truth_vector_type): Take the number of units as a poly_uint64
95 rather than an unsigned int.
97 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
98 Alan Hayward <alan.hayward@arm.com>
99 David Sherwood <david.sherwood@arm.com>
101 * target.def (get_mask_mode): Take the number of units and length
102 as poly_uint64s rather than unsigned ints.
103 * targhooks.h (default_get_mask_mode): Update accordingly.
104 * targhooks.c (default_get_mask_mode): Likewise.
105 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
106 * doc/tm.texi: Regenerate.
108 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
109 Alan Hayward <alan.hayward@arm.com>
110 David Sherwood <david.sherwood@arm.com>
112 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
113 * omp-general.c (omp_max_vf): Likewise.
114 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
115 (expand_omp_simd): Handle polynomial safelen.
116 * omp-low.c (omplow_simd_context): Add a default constructor.
117 (omplow_simd_context::max_vf): Change from int to poly_uint64.
118 (lower_rec_simd_input_clauses): Update accordingly.
119 (lower_rec_input_clauses): Likewise.
121 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
122 Alan Hayward <alan.hayward@arm.com>
123 David Sherwood <david.sherwood@arm.com>
125 * tree-vectorizer.h (vect_nunits_for_cost): New function.
126 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
127 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
128 (vect_analyze_slp_cost): Likewise.
129 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
130 (vect_model_load_cost): Likewise.
132 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
133 Alan Hayward <alan.hayward@arm.com>
134 David Sherwood <david.sherwood@arm.com>
136 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
137 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
138 from an unsigned int * to a poly_uint64_pod *.
139 (calculate_unrolling_factor): New function.
140 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
142 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
143 Alan Hayward <alan.hayward@arm.com>
144 David Sherwood <david.sherwood@arm.com>
146 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
147 from an unsigned int to a poly_uint64.
148 (_loop_vec_info::slp_unrolling_factor): Likewise.
149 (_loop_vec_info::vectorization_factor): Change from an int
151 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
152 (vect_get_num_vectors): New function.
153 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
154 (vect_get_num_copies): Use vect_get_num_vectors.
155 (vect_analyze_data_ref_dependences): Change max_vf from an int *
156 to an unsigned int *.
157 (vect_analyze_data_refs): Change min_vf from an int * to a
159 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
160 than an unsigned HOST_WIDE_INT.
161 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
162 (vect_analyze_data_ref_dependence): Change max_vf from an int *
163 to an unsigned int *.
164 (vect_analyze_data_ref_dependences): Likewise.
165 (vect_compute_data_ref_alignment): Handle polynomial vf.
166 (vect_enhance_data_refs_alignment): Likewise.
167 (vect_prune_runtime_alias_test_list): Likewise.
168 (vect_shift_permute_load_chain): Likewise.
169 (vect_supportable_dr_alignment): Likewise.
170 (dependence_distance_ge_vf): Take the vectorization factor as a
171 poly_uint64 rather than an unsigned HOST_WIDE_INT.
172 (vect_analyze_data_refs): Change min_vf from an int * to a
174 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
175 vfm1 as a poly_uint64 rather than an int. Make the same change
176 for the returned bound_scalar.
177 (vect_gen_vector_loop_niters): Handle polynomial vf.
178 (vect_do_peeling): Likewise. Update call to
179 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
180 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
182 * tree-vect-loop.c (vect_determine_vectorization_factor)
183 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
184 (vect_get_known_peeling_cost): Likewise.
185 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
186 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
187 (vect_transform_loop): Likewise. Use the lowest possible VF when
188 updating the upper bounds of the loop.
189 (vect_min_worthwhile_factor): Make static. Return an unsigned int
191 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
192 polynomial unroll factors.
193 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
194 (vect_make_slp_decision): Likewise.
195 (vect_supported_load_permutation_p): Likewise, and polynomial
197 (vect_analyze_slp_cost): Handle polynomial vf.
198 (vect_slp_analyze_node_operations): Likewise.
199 (vect_slp_analyze_bb_1): Likewise.
200 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
201 than an unsigned HOST_WIDE_INT.
202 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
203 (vectorizable_load): Handle polynomial vf.
204 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
206 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
208 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
209 Alan Hayward <alan.hayward@arm.com>
210 David Sherwood <david.sherwood@arm.com>
212 * match.pd: Handle bit operations involving three constants
213 and try to fold one pair.
215 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
217 * tree-vect-loop-manip.c: Include gimple-fold.h.
218 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
219 niters_maybe_zero parameters. Handle other cases besides a step of 1.
220 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
221 Add a path that uses a step of VF instead of 1, but disable it
223 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
224 and niters_no_overflow parameters. Update calls to
225 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
226 Create a new SSA name if the latter choses to use a ste other
227 than zero, and return it via niters_vector_mult_vf_var.
228 * tree-vect-loop.c (vect_transform_loop): Update calls to
229 vect_do_peeling, vect_gen_vector_loop_niters and
230 slpeel_make_loop_iterate_ntimes.
231 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
232 (vect_gen_vector_loop_niters): Update declarations after above changes.
234 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
236 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
237 128-bit round to integer instructions.
238 (ceil<mode>2): Likewise.
239 (btrunc<mode>2): Likewise.
240 (round<mode>2): Likewise.
242 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
244 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
245 unaligned VSX load/store on P8/P9.
246 (expand_block_clear): Allow the use of unaligned VSX
249 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
251 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
253 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
254 swap associated with both a load and a store.
256 2018-01-02 Andrew Waterman <andrew@sifive.com>
258 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
259 * config/riscv/riscv.md (clear_cache): Use it.
261 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
263 * web.c: Remove out-of-date comment.
265 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
267 * expr.c (fixup_args_size_notes): Check that any existing
268 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
269 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
270 (emit_single_push_insn): ...here.
272 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
274 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
275 (const_vector_encoded_nelts): New function.
276 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
277 (const_vector_int_elt, const_vector_elt): Declare.
278 * emit-rtl.c (const_vector_int_elt_1): New function.
279 (const_vector_elt): Likewise.
280 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
283 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
285 * expr.c: Include rtx-vector-builder.h.
286 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
287 directly on the tree encoding.
288 (const_vector_from_tree): Likewise.
289 * optabs.c: Include rtx-vector-builder.h.
290 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
291 sequence of "u" values.
292 * vec-perm-indices.c: Include rtx-vector-builder.h.
293 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
294 directly on the vec_perm_indices encoding.
296 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
298 * doc/rtl.texi (const_vector): Describe new encoding scheme.
299 * Makefile.in (OBJS): Add rtx-vector-builder.o.
300 * rtx-vector-builder.h: New file.
301 * rtx-vector-builder.c: Likewise.
302 * rtl.h (rtx_def::u2): Add a const_vector field.
303 (CONST_VECTOR_NPATTERNS): New macro.
304 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
305 (CONST_VECTOR_DUPLICATE_P): Likewise.
306 (CONST_VECTOR_STEPPED_P): Likewise.
307 (CONST_VECTOR_ENCODED_ELT): Likewise.
308 (const_vec_duplicate_p): Check for a duplicated vector encoding.
309 (unwrap_const_vec_duplicate): Likewise.
310 (const_vec_series_p): Check for a non-duplicated vector encoding.
311 Say that the function only returns true for integer vectors.
312 * emit-rtl.c: Include rtx-vector-builder.h.
313 (gen_const_vec_duplicate_1): Delete.
314 (gen_const_vector): Call gen_const_vec_duplicate instead of
315 gen_const_vec_duplicate_1.
316 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
317 (gen_const_vec_duplicate): Use rtx_vector_builder.
318 (gen_const_vec_series): Likewise.
319 (gen_rtx_CONST_VECTOR): Likewise.
320 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
321 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
322 Build a new vector rather than modifying a CONST_VECTOR in-place.
323 (handle_special_swappables): Update call accordingly.
324 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
325 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
326 Build a new vector rather than modifying a CONST_VECTOR in-place.
327 (handle_special_swappables): Update call accordingly.
329 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
331 * simplify-rtx.c (simplify_const_binary_operation): Use
332 CONST_VECTOR_ELT instead of XVECEXP.
334 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
336 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
337 the selector elements to be different from the data elements
338 if the selector is a VECTOR_CST.
339 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
340 ssizetype for the selector.
342 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
344 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
345 before testing each element individually.
346 * tree-vect-generic.c (lower_vec_perm): Likewise.
348 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
350 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
351 * selftest-run-tests.c (selftest::run_tests): Call it.
352 * vector-builder.h (vector_builder::operator ==): New function.
353 (vector_builder::operator !=): Likewise.
354 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
355 (vec_perm_indices::all_from_input_p): New function.
356 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
357 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
358 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
359 instead of reading the VECTOR_CST directly. Detect whether both
360 vector inputs are the same before constructing the vec_perm_indices,
361 and update the number of inputs argument accordingly. Use the
362 utility functions added above. Only construct sel2 if we need to.
364 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
366 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
367 the broadcast of the low byte.
368 (expand_mult_highpart): Use an explicit encoding for the permutes.
369 * optabs-query.c (can_mult_highpart_p): Likewise.
370 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
371 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
372 (vectorizable_bswap): Likewise.
373 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
374 explicit encoding for the power-of-2 permutes.
375 (vect_permute_store_chain): Likewise.
376 (vect_grouped_load_supported): Likewise.
377 (vect_permute_load_chain): Likewise.
379 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
381 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
382 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
383 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
384 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
385 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
386 (vect_gen_perm_mask_any): Likewise.
388 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
390 * int-vector-builder.h: New file.
391 * vec-perm-indices.h: Include int-vector-builder.h.
392 (vec_perm_indices): Redefine as an int_vector_builder.
393 (auto_vec_perm_indices): Delete.
394 (vec_perm_builder): Redefine as a stand-alone class.
395 (vec_perm_indices::vec_perm_indices): New function.
396 (vec_perm_indices::clamp): Likewise.
397 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
398 (vec_perm_indices::new_vector): New function.
399 (vec_perm_indices::new_expanded_vector): Update for new
400 vec_perm_indices class.
401 (vec_perm_indices::rotate_inputs): New function.
402 (vec_perm_indices::all_in_range_p): Operate directly on the
403 encoded form, without computing elided elements.
404 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
405 encoding. Update for new vec_perm_indices class.
406 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
407 the given vec_perm_builder.
408 (expand_vec_perm_var): Update vec_perm_builder constructor.
409 (expand_mult_highpart): Use vec_perm_builder instead of
410 auto_vec_perm_indices.
411 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
412 vec_perm_indices instead of auto_vec_perm_indices. Use a single
413 or double series encoding as appropriate.
414 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
415 vec_perm_indices instead of auto_vec_perm_indices.
416 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
417 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
418 (vect_permute_store_chain): Likewise.
419 (vect_grouped_load_supported): Likewise.
420 (vect_permute_load_chain): Likewise.
421 (vect_shift_permute_load_chain): Likewise.
422 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
423 (vect_transform_slp_perm_load): Likewise.
424 (vect_schedule_slp_instance): Likewise.
425 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
426 (vectorizable_mask_load_store): Likewise.
427 (vectorizable_bswap): Likewise.
428 (vectorizable_store): Likewise.
429 (vectorizable_load): Likewise.
430 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
431 vec_perm_indices instead of auto_vec_perm_indices. Use
432 tree_to_vec_perm_builder to read the vector from a tree.
433 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
434 vec_perm_builder instead of a vec_perm_indices.
435 (have_whole_vector_shift): Use vec_perm_builder and
436 vec_perm_indices instead of auto_vec_perm_indices. Leave the
437 truncation to calc_vec_perm_mask_for_shift.
438 (vect_create_epilog_for_reduction): Likewise.
439 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
440 from auto_vec_perm_indices to vec_perm_indices.
441 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
442 instead of changing individual elements.
443 (aarch64_vectorize_vec_perm_const): Use new_vector to install
444 the vector in d.perm.
445 * config/arm/arm.c (expand_vec_perm_d::perm): Change
446 from auto_vec_perm_indices to vec_perm_indices.
447 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
448 instead of changing individual elements.
449 (arm_vectorize_vec_perm_const): Use new_vector to install
450 the vector in d.perm.
451 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
452 Update vec_perm_builder constructor.
453 (rs6000_expand_interleave): Likewise.
454 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
455 (rs6000_expand_interleave): Likewise.
457 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
459 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
460 to qimode could truncate the indices.
461 * optabs.c (expand_vec_perm_var): Likewise.
463 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
465 * Makefile.in (OBJS): Add vec-perm-indices.o.
466 * vec-perm-indices.h: New file.
467 * vec-perm-indices.c: Likewise.
468 * target.h (vec_perm_indices): Replace with a forward class
470 (auto_vec_perm_indices): Move to vec-perm-indices.h.
471 * optabs.h: Include vec-perm-indices.h.
472 (expand_vec_perm): Delete.
473 (selector_fits_mode_p, expand_vec_perm_var): Declare.
474 (expand_vec_perm_const): Declare.
475 * target.def (vec_perm_const_ok): Replace with...
476 (vec_perm_const): ...this new hook.
477 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
478 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
479 * doc/tm.texi: Regenerate.
480 * optabs.def (vec_perm_const): Delete.
481 * doc/md.texi (vec_perm_const): Likewise.
482 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
483 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
484 expand_vec_perm for constant permutation vectors. Assert that
485 the mode of variable permutation vectors is the integer equivalent
486 of the mode that is being permuted.
487 * optabs-query.h (selector_fits_mode_p): Declare.
488 * optabs-query.c: Include vec-perm-indices.h.
489 (selector_fits_mode_p): New function.
490 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
491 is defined, instead of checking whether the vec_perm_const_optab
492 exists. Use targetm.vectorize.vec_perm_const instead of
493 targetm.vectorize.vec_perm_const_ok. Check whether the indices
494 fit in the vector mode before using a variable permute.
495 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
496 vec_perm_indices instead of an rtx.
497 (expand_vec_perm): Replace with...
498 (expand_vec_perm_const): ...this new function. Take the selector
499 as a vec_perm_indices rather than an rtx. Also take the mode of
500 the selector. Update call to shift_amt_for_vec_perm_mask.
501 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
502 Use vec_perm_indices::new_expanded_vector to expand the original
503 selector into bytes. Check whether the indices fit in the vector
504 mode before using a variable permute.
505 (expand_vec_perm_var): Make global.
506 (expand_mult_highpart): Use expand_vec_perm_const.
507 * fold-const.c: Includes vec-perm-indices.h.
508 * tree-ssa-forwprop.c: Likewise.
509 * tree-vect-data-refs.c: Likewise.
510 * tree-vect-generic.c: Likewise.
511 * tree-vect-loop.c: Likewise.
512 * tree-vect-slp.c: Likewise.
513 * tree-vect-stmts.c: Likewise.
514 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
516 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
517 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
518 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
519 (aarch64_vectorize_vec_perm_const): ...this new function.
520 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
521 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
522 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
523 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
524 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
525 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
526 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
528 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
529 check for NEON modes.
530 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
531 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
532 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
533 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
535 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
536 the old VEC_PERM_CONST conditions.
537 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
538 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
539 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
540 (ia64_vectorize_vec_perm_const_ok): Merge into...
541 (ia64_vectorize_vec_perm_const): ...this new function.
542 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
543 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
544 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
545 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
546 * config/mips/mips.c (mips_expand_vec_perm_const)
547 (mips_vectorize_vec_perm_const_ok): Merge into...
548 (mips_vectorize_vec_perm_const): ...this new function.
549 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
550 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
551 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
552 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
553 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
554 (rs6000_expand_vec_perm_const): Delete.
555 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
557 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
558 (altivec_expand_vec_perm_const_le): Take each operand individually.
559 Operate on constant selectors rather than rtxes.
560 (altivec_expand_vec_perm_const): Likewise. Update call to
561 altivec_expand_vec_perm_const_le.
562 (rs6000_expand_vec_perm_const): Delete.
563 (rs6000_vectorize_vec_perm_const_ok): Delete.
564 (rs6000_vectorize_vec_perm_const): New function.
565 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
566 an element count and rtx array.
567 (rs6000_expand_extract_even): Update call accordingly.
568 (rs6000_expand_interleave): Likewise.
569 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
570 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
571 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
572 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
573 (rs6000_expand_vec_perm_const): Delete.
574 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
575 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
576 (altivec_expand_vec_perm_const_le): Take each operand individually.
577 Operate on constant selectors rather than rtxes.
578 (altivec_expand_vec_perm_const): Likewise. Update call to
579 altivec_expand_vec_perm_const_le.
580 (rs6000_expand_vec_perm_const): Delete.
581 (rs6000_vectorize_vec_perm_const_ok): Delete.
582 (rs6000_vectorize_vec_perm_const): New function. Remove stray
583 reference to the SPE evmerge intructions.
584 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
585 an element count and rtx array.
586 (rs6000_expand_extract_even): Update call accordingly.
587 (rs6000_expand_interleave): Likewise.
588 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
589 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
591 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
593 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
595 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
596 vector mode and that that mode matches the mode of the data
598 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
599 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
600 directly using expand_vec_perm_1 when forcing selectors into
602 (expand_vec_perm_var): New function, split out from expand_vec_perm.
604 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
606 * optabs-query.h (can_vec_perm_p): Delete.
607 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
608 * optabs-query.c (can_vec_perm_p): Split into...
609 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
610 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
611 particular selector is valid.
612 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
613 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
614 (vect_grouped_load_supported): Likewise.
615 (vect_shift_permute_load_chain): Likewise.
616 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
617 (vect_transform_slp_perm_load): Likewise.
618 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
619 (vectorizable_bswap): Likewise.
620 (vect_gen_perm_mask_checked): Likewise.
621 * fold-const.c (fold_ternary_loc): Likewise. Don't take
622 implementations of variable permutation vectors into account
623 when deciding which selector to use.
624 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
625 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
626 with a false third argument.
627 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
628 to test whether the constant selector is valid and can_vec_perm_var_p
629 to test whether a variable selector is valid.
631 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
633 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
634 * optabs-query.c (can_vec_perm_p): Likewise.
635 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
636 instead of vec_perm_indices.
637 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
638 (vect_gen_perm_mask_checked): Likewise,
639 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
640 (vect_gen_perm_mask_checked): Likewise,
642 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
644 * optabs-query.h (qimode_for_vec_perm): Declare.
645 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
646 (qimode_for_vec_perm): ...this new function.
647 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
649 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
651 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
652 does not have a conditional at the top.
654 2018-01-02 Richard Biener <rguenther@suse.de>
656 * ipa-inline.c (big_speedup_p): Fix expression.
658 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
661 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
664 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
668 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
669 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
670 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
671 cond_taken_branch_cost 3->4.
673 2018-01-01 Jakub Jelinek <jakub@redhat.com>
675 PR tree-optimization/83581
676 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
677 TODO_cleanup_cfg if any changes have been made.
680 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
681 convert_modes if target mode has the right side, but different mode
685 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
686 last argument when extracting from CONCAT. If either from_real or
687 from_imag is NULL, use expansion through memory. If result is not
688 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
689 the parts directly to inner mode, if even that fails, use expansion
693 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
694 check for bswap in mode rather than HImode and use that in expand_unop
697 2017-12-31 Jakub Jelinek <jakub@redhat.com>
700 * config/i386/i386.c (ix86_attribute_table): Remove excess
701 initializer for "shared" attribute.
703 2017-12-30 Tom de Vries <tom@codesourcery.com>
706 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
708 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
709 offload_funcs entries that no longer have a corresponding cgraph_node.
710 Mark the remaining ones as DECL_PRESERVE_P.
711 (output_lto): Call prune_offload_funcs.
713 2017-12-30 Jakub Jelinek <jakub@redhat.com>
715 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
716 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
717 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
718 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
719 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
720 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
721 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
722 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
723 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
724 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
725 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
726 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
727 vaesenc_<mode>, vpclmulqdq_<mode>,
728 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
730 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
732 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
733 _Float<N> and _Float<N>X variants for rounding built-in
735 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
736 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
737 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
738 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
739 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
740 * builtins.c (mathfn_built_in_2): Likewise.
741 * internal-fn.def (CEIL): Likewise.
743 (NEARBYINT): Likewise.
747 * convert.c (convert_to_integer_1): Likewise.
748 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
749 (integer_valued_real_call_p): Likewise.
750 * fold-const-call.c (fold_const_call_ss): Likewise.
751 * gencfn-macros.c (print_case_cfn): Change CFN and operator
752 printers to take a const char * suffix instead of a bool.
753 (print_define_operator_list): Likewise.
754 (fltall_suffixes): New list of suffixes, that include the
755 traditional suffixes as well as all of the _Float<N> and
757 (main): For _Float<N> and _Float<N>X functions, emit both
758 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
759 has the _Float<N> and _Float<N>X case names or operators. The
760 <name>_ALL variant has both the traditional and the
761 _Float<N>/_Float<N>X case names or operators.
762 * match.pd (COPYSIGN optimizations): Provide optimizations for
763 _Float<N> and _Float<N>X types where possible.
764 (MIN/MAX optimizations): Likewise.
765 (sqrt optimizations): Likewise.
766 (rounding optimizations): Likewise.
768 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
770 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
771 gen_int_mode rather than GEN_INT.
772 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
774 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
776 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
777 instead of CONSTANT_P.
778 (gen_vec_series): Likewise.
779 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
781 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
783 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
786 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
788 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
789 (aarch64_ld1x2<VDC:mode>): Likewise.
790 (aarch64_simd_ld1<mode>_x2): Likewise.
791 (aarch64_simd_ld1<mode>_x2): Likewise.
792 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
793 (vld1_s8_x2): Likewise.
794 (vld1_u16_x2): Likewise.
795 (vld1_s16_x2): Likewise.
796 (vld1_u32_x2): Likewise.
797 (vld1_s32_x2): Likewise.
798 (vld1_u64_x2): Likewise.
799 (vld1_s64_x2): Likewise.
800 (vld1_f16_x2): Likewise.
801 (vld1_f32_x2): Likewise.
802 (vld1_f64_x2): Likewise.
803 (vld1_p8_x2): Likewise.
804 (vld1_p16_x2): Likewise.
805 (vld1_p64_x2): Likewise.
806 (vld1q_u8_x2): Likewise.
807 (vld1q_s8_x2): Likewise.
808 (vld1q_u16_x2): Likewise.
809 (vld1q_s16_x2): Likewise.
810 (vld1q_u32_x2): Likewise.
811 (vld1q_s32_x2): Likewise.
812 (vld1q_u64_x2): Likewise.
813 (vld1q_s64_x2): Likewise.
814 (vld1q_f16_x2): Likewise.
815 (vld1q_f32_x2): Likewise.
816 (vld1q_f64_x2): Likewise.
817 (vld1q_p8_x2): Likewise.
818 (vld1q_p16_x2): Likewise.
819 (vld1q_p64_x2): Likewise.
821 2017-12-27 Martin Liska <mliska@suse.cz>
823 PR tree-optimization/83552
824 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
825 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
827 2017-12-27 Tom de Vries <tom@codesourcery.com>
829 * config/nvptx/nvptx.c (nvptx_option_override): Disable
830 -gstatement-frontiers.
832 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
834 PR rtl-optimization/83513
835 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
836 before priority comparison.
838 2017-12-25 Jakub Jelinek <jakub@redhat.com>
841 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
842 ix86_isa_flags2 to ix86_isa_flags.
843 * config/i386/i386-c.c (ix86_target_macros_internal): Test
844 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
845 isa_flags rather than isa_flags2.
846 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
847 and -mavx512bitalg from isa2_opts to isa_opts.
848 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
849 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
850 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
852 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
853 OPTION_MASK_ISA_AVX512F ored with another option similarly to
854 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
855 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
856 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
857 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
858 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
859 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
860 builtins from bdesc_args2 to bdesc_args section.
861 (__builtin_ia32_compressstoreuqi512_mask,
862 __builtin_ia32_compressstoreuhi512_mask,
863 __builtin_ia32_compressstoreuqi256_mask,
864 __builtin_ia32_expandloadqi512_mask,
865 __builtin_ia32_expandloadqi512_maskz,
866 __builtin_ia32_expandloadhi512_mask,
867 __builtin_ia32_expandloadhi512_maskz,
868 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
869 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
870 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
871 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
872 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
873 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
874 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
875 __builtin_ia32_vpshldv_v32hi_maskz,
876 __builtin_ia32_vpopcountb_v64qi_mask,
877 __builtin_ia32_vpopcountw_v32hi_mask,
878 __builtin_ia32_vpshufbitqmb512_mask,
879 __builtin_ia32_vpshufbitqmb256_mask): Add
880 " | OPTION_MASK_ISA_AVX512BW".
881 (__builtin_ia32_expandloadqi256_mask,
882 __builtin_ia32_expandloadqi256_maskz,
883 __builtin_ia32_vpopcountb_v32qi_mask): Add
884 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
885 (__builtin_ia32_expandloadhi256_mask,
886 __builtin_ia32_expandloadhi256_maskz,
887 __builtin_ia32_expandloadqi128_mask,
888 __builtin_ia32_expandloadqi128_maskz,
889 __builtin_ia32_expandloadhi128_mask,
890 __builtin_ia32_expandloadhi128_maskz,
891 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
892 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
893 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
894 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
895 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
896 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
897 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
898 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
899 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
900 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
901 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
902 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
903 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
904 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
905 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
906 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
907 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
908 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
909 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
910 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
911 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
912 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
913 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
914 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
915 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
916 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
917 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
918 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
919 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
920 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
921 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
922 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
923 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
924 __builtin_ia32_vpopcountw_v8hi_mask): Add
925 " | OPTION_MASK_ISA_AVX512VL".
926 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
927 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
928 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
929 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
930 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
931 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
932 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
933 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
934 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
935 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
936 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
937 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
938 _mm_mask_bitshuffle_epi64_mask): Likewise.
939 * common/config/i386/i386-common.c
940 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
941 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
942 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
943 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
944 OPTION_MASK_ISA_AVX512BITALG_UNSET.
945 (OPTION_MASK_ISA2_AVX512F_UNSET,
946 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
947 (ix86_handle_option): For -mno-general-regs-only, clear from
948 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
949 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
950 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
951 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
952 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
953 for moving from ix86_isa_flags2 to ix86_isa_flags.
955 2017-12-23 Jakub Jelinek <jakub@redhat.com>
958 * fold-const.c (struct contains_label_data): New type.
959 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
960 inside of a SWITCH_BODY seen during the walk.
961 (contains_label_p): Use walk_tree instead of
962 walk_tree_without_duplicates, prepare data for contains_label_1 and
965 2017-12-22 Jakub Jelinek <jakub@redhat.com>
968 * tree-iterator.c (alloc_stmt_list): Start with cleared
969 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
973 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
974 enable avx512f explicitly in #pragma GCC target.
975 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
976 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
977 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
978 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
979 __builtin_ia32_vpdpbusds_v8si_mask,
980 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
981 __builtin_ia32_vpdpbusds_v4si_mask,
982 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
983 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
984 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
985 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
986 __builtin_ia32_vpdpwssds_v8si_mask,
987 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
988 __builtin_ia32_vpdpwssds_v4si_mask,
989 __builtin_ia32_vpdpwssds_v4si_maskz): Use
990 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
991 just OPTION_MASK_ISA_AVX512VNNI.
993 2017-12-22 Martin Jambor <mjambor@suse.cz>
996 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
999 2017-12-22 Julia Koval <julia.koval@intel.com>
1000 Sebastian Peryt <sebastian.peryt@intel.com>
1002 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
1003 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
1004 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
1005 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
1006 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
1007 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
1008 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
1009 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
1010 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
1011 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
1012 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
1013 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
1014 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
1015 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
1016 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
1017 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
1018 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
1019 _mm256_maskz_popcnt_epi64): New intrinsics.
1020 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
1021 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
1022 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
1023 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
1024 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
1025 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
1026 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
1027 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
1028 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
1029 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
1030 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
1031 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
1032 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
1033 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
1034 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1035 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
1036 __builtin_ia32_vpshufbitqmb256_mask,
1037 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
1038 * config/i386/i386-c.c (__AVX512BITALG__): New.
1039 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
1040 (ix86_valid_target_attribute_inner_p): Ditto.
1041 (ix86_expand_args_builtin): Handle new types.
1042 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
1043 * config/i386/i386.opt: Add -mavx512bitalg.
1044 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
1045 avx512bitalgintrin.h.
1046 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
1047 (vpopcount<mode><mask_name>): Add more types.
1048 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
1049 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
1051 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1053 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
1054 Or in OPTION_MASK_ISA_AVX512F_SET.
1055 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1056 OPTION_MASK_ISA_AVX512VNNI_UNSET.
1057 (ix86_handle_option): Adjust for
1058 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
1059 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1061 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1062 isa_flag instead of isa_flag2.
1063 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1064 isa_opts2 to isa_opts.
1065 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1068 2017-12-22 Mike Stump <mikestump@comcast.net>
1069 Eric Botcazou <ebotcazou@adacore.com>
1071 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1073 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1076 * config/i386/i386.c (ix86_function_arg_boundary): Return
1077 PARM_BOUNDARY for TYPE_EMPTY_P types.
1080 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1081 if navail is >= dir.len.
1083 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1085 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1088 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1090 PR rtl-optimization/80747
1091 PR rtl-optimization/83512
1092 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1093 succ edge from ENTRY, copy partition from e->dest to the newly
1095 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1096 ENTRY, use BB_PARTITION of its successor block as current_partition.
1097 Don't copy partition when splitting succ edge from ENTRY.
1099 PR tree-optimization/83523
1100 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1101 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1102 (convert_mult_to_fma): Likewise.
1104 PR tree-optimization/83521
1105 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1106 gimple_build_assign without code on result of
1107 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1108 a VIEW_CONVERT_EXPR.
1110 2017-12-21 Andrew Pinski <apinski@cavium.com>
1111 Steve Ellcey <sellcey@cavium.com>
1113 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1114 multi-arch for ilp32.
1116 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1119 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1120 constraints to operand 2.
1121 (*ashl<mode>3_mask_1): Ditto.
1122 (*<shift_insn><mode>3_mask): Ditto.
1123 (*<shift_insn><mode>3_mask_1): Ditto.
1124 (*<rotate_insn><mode>3_mask): Ditto.
1125 (*<rotate_insn><mode>3_mask_1): Ditto.
1127 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1129 * reorg.c (make_return_insns): Reemit each insn with its own location.
1131 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1133 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1134 generation for cases where splatting a value is not useful.
1135 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1136 across a vec_duplicate and a paradoxical subreg forming a vector
1137 mode to a vec_concat.
1139 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1141 * combine.c (simplify_set): Do not transform subregs to zero_extends
1142 if the destination is not a scalar int mode.
1144 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1147 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1148 shared zero if the input has overflowed.
1150 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1152 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1153 support for Cortex-A55 and Cortex-A75.
1155 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1157 * common/config/arm/arm-common.c (compare_opt_names): Add function
1158 comment. Use strcmp instead of manual loop.
1160 2017-12-21 Martin Liska <mliska@suse.cz>
1162 PR gcov-profile/83509
1163 * gcov-dump.c (dump_gcov_file): Do not read info about
1164 support_unexecuted_blocks for gcda files.
1166 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1168 PR rtl-optimization/82973
1169 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1170 (valid_for_const_vector_p): ... this.
1171 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1172 (valid_for_const_vector_p): ... this. Adjust function comment.
1173 (gen_vec_duplicate): Adjust caller.
1174 * optabs.c (expand_vector_broadcast): Likewise.
1175 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1176 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1178 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1179 instead of GET_CODE == CONST_FIXED.
1180 (simplify_subreg): Use CONST_FIXED_P macro instead of
1181 GET_CODE == CONST_FIXED.
1184 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1185 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1186 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1187 (ix86_option_override_internal): Adjust for
1188 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1189 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1190 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1191 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1192 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1193 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1194 def_builtin2 for CET builtins.
1195 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1196 up formatting in IX86_BUILTIN_RDPID code.
1197 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1198 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1199 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1200 ix86_isa_flags2 to ix86_isa_flags.
1201 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1203 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1204 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1205 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1207 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1208 Or in OPTION_MASK_ISA_AVX512F_SET.
1209 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1210 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1211 (ix86_handle_option): Adjust for
1212 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1213 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1215 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1216 Alan Hayward <alan.hayward@arm.com>
1217 David Sherwood <david.sherwood@arm.com>
1219 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1220 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1221 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1222 Track polynomial offsets.
1224 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1225 Alan Hayward <alan.hayward@arm.com>
1226 David Sherwood <david.sherwood@arm.com>
1228 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1229 Treat drb->init as a poly_int. Fail if its misalignment wrt
1230 vector_alignment isn't known.
1232 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1233 Alan Hayward <alan.hayward@arm.com>
1234 David Sherwood <david.sherwood@arm.com>
1236 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1238 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1239 (vect_loop_versioning): Take the loop versioning threshold as a
1241 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1242 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1243 versioning_threshold.
1244 (vect_analyze_loop_2): Compute the loop versioning threshold
1245 whenever loop versioning is needed, and store it in the new
1246 field rather than combining it with the cost model threshold.
1247 (vect_transform_loop): Update call to vect_loop_versioning.
1248 Try to combine the loop versioning and cost thresholds here.
1250 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1251 Alan Hayward <alan.hayward@arm.com>
1252 David Sherwood <david.sherwood@arm.com>
1254 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1255 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1256 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1257 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1258 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1259 (group_compare_offset): Update accordingly.
1260 (split_small_address_groups_p): Likewise.
1261 (record_use): Take addr_offset as a poly_uint64 rather than
1262 an unsigned HOST_WIDE_INT.
1263 (strip_offset): Return the offset as a poly_uint64 rather than
1264 an unsigned HOST_WIDE_INT.
1265 (record_group_use, split_address_groups): Track polynomial offsets.
1266 (add_iv_candidate_for_use): Likewise.
1267 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1268 than a HOST_WIDE_INT.
1269 (strip_offset_1): Return the offset as a poly_int64 rather than
1272 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1273 Alan Hayward <alan.hayward@arm.com>
1274 David Sherwood <david.sherwood@arm.com>
1276 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1277 rather than a HOST_WIDE_INT.
1278 * tree.c (get_binfo_at_offset): Likewise.
1280 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1281 Alan Hayward <alan.hayward@arm.com>
1282 David Sherwood <david.sherwood@arm.com>
1284 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1285 rather than a HOST_WIDE_INT.
1286 * tree-sra.c (build_ref_for_offset): Likewise.
1288 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1289 Alan Hayward <alan.hayward@arm.com>
1290 David Sherwood <david.sherwood@arm.com>
1292 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1294 * tree.c (mem_ref_offset): Likewise.
1295 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1296 * builtins.c (get_object_alignment_2): Likewise.
1297 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1298 * gimple-fold.c (get_base_constructor): Likewise.
1299 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1300 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1302 * ipa-polymorphic-call.c
1303 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1304 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1305 (get_ancestor_addr_info): Likewise.
1306 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1307 * match.pd: Likewise.
1308 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1309 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1310 * tree-eh.c (tree_could_trap_p): Likewise.
1311 * tree-object-size.c (addr_object_size): Likewise.
1312 * tree-ssa-address.c (copy_ref_info): Likewise.
1313 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1314 (indirect_refs_may_alias_p): Likewise.
1315 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1316 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1317 (non_rewritable_mem_ref_base): Likewise.
1318 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1319 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1320 * varasm.c (decode_addr_const): Likewise.
1322 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1323 Alan Hayward <alan.hayward@arm.com>
1324 David Sherwood <david.sherwood@arm.com>
1326 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1327 offsets for MEM_REFs.
1329 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1330 Alan Hayward <alan.hayward@arm.com>
1331 David Sherwood <david.sherwood@arm.com>
1333 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1334 as a poly_uint64 rather than an unsigned int.
1335 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1337 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1338 Alan Hayward <alan.hayward@arm.com>
1339 David Sherwood <david.sherwood@arm.com>
1341 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1343 (decode_addr_const): Update accordingly.
1345 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1346 Alan Hayward <alan.hayward@arm.com>
1347 David Sherwood <david.sherwood@arm.com>
1349 * tree.h (bit_field_size, bit_field_offset): New functions.
1350 * hsa-gen.c (gen_hsa_addr): Use them.
1351 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1352 (simplify_vector_constructor): Likewise.
1353 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1354 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1355 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1356 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1359 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1360 Alan Hayward <alan.hayward@arm.com>
1361 David Sherwood <david.sherwood@arm.com>
1363 * expr.h (emit_group_load, emit_group_load_into_temps)
1364 (emit_group_store): Take the size as a poly_int64 rather than an int.
1365 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1366 (emit_group_load_into_temp, emit_group_store): Likewise.
1368 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1369 Alan Hayward <alan.hayward@arm.com>
1370 David Sherwood <david.sherwood@arm.com>
1372 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1373 an unsigned int to a poly_uint64.
1374 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1375 sizes as poly_uint64s rather than unsigned ints.
1376 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1379 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1380 Alan Hayward <alan.hayward@arm.com>
1381 David Sherwood <david.sherwood@arm.com>
1383 * lra-constraints.c (emit_inc): Change inc_amount from an int
1386 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1387 Alan Hayward <alan.hayward@arm.com>
1388 David Sherwood <david.sherwood@arm.com>
1390 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1392 (add_stack_var, stack_var_cmp, partition_stack_vars)
1393 (dump_stack_var_partition): Update accordingly.
1394 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1395 than a HOST_WIDE_INT.
1396 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1397 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1398 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1399 rather than a HOST_WIDE_INT.
1401 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1402 Alan Hayward <alan.hayward@arm.com>
1403 David Sherwood <david.sherwood@arm.com>
1405 * target.def (return_pops_args): Treat both the input and output
1406 sizes as poly_int64s rather than HOST_WIDE_INTS.
1407 * targhooks.h (default_return_pops_args): Update accordingly.
1408 * targhooks.c (default_return_pops_args): Likewise.
1409 * doc/tm.texi: Regenerate.
1410 * emit-rtl.h (incoming_args): Change pops_args, size and
1411 outgoing_args_size from int to poly_int64_pod.
1412 * function.h (expr_status): Change x_pending_stack_adjust and
1413 x_stack_pointer_delta from int to poly_int64.
1414 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1415 (ARGS_SIZE_RTX): Update accordingly.
1416 * calls.c (highest_outgoing_arg_in_use): Change from int to
1418 (stack_usage_watermark, stored_args_watermark): New variables.
1419 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1420 (emit_call_1): Change the stack_size and rounded_stack_size
1421 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1423 (save_fixed_argument_area): Check stack_usage_watermark.
1424 (initialize_argument_information): Change old_pending_adj from
1425 a HOST_WIDE_INT * to a poly_int64_pod *.
1426 (compute_argument_block_size): Return the size as a poly_int64
1428 (finalize_must_preallocate): Track polynomial argument sizes.
1429 (compute_argument_addresses): Likewise.
1430 (internal_arg_pointer_based_exp): Track polynomial offsets.
1431 (mem_overlaps_already_clobbered_arg_p): Rename to...
1432 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1433 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1434 Check stored_args_used_watermark.
1435 (load_register_parameters): Update accordingly.
1436 (check_sibcall_argument_overlap_1): Likewise.
1437 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1438 args size as a poly_int64 rather than an int. Return a bool
1439 indicating whether the optimization was possible and return
1440 the new adjustment by reference.
1441 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1442 Update stored_args_watermark.
1443 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1444 (expand_call): Likewise. Maintain stack_usage_watermark and
1445 stored_args_watermark. Update calls to
1446 combine_pending_stack_adjustment_and_call.
1447 (emit_library_call_value_1): Handle polynomial argument sizes.
1448 Call stack_region_maybe_used_p and mark_stack_region_used.
1449 Maintain stack_usage_watermark.
1450 (store_one_arg): Likewise. Update call to
1451 mem_overlaps_already_clobbered_arg_p.
1452 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1454 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1455 * config/microblaze/microblaze.c (microblaze_function_prologue):
1457 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1458 TARGET_RETURN_POPS_ARGS interface.
1459 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1461 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1462 * config/i386/i386.c (ix86_return_pops_args): Update for new
1463 TARGET_RETURN_POPS_ARGS interface.
1464 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1465 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1466 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1467 TARGET_RETURN_POPS_ARGS interface.
1468 * config/vax/vax.c (vax_return_pops_args): Likewise.
1469 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1470 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1471 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1472 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1473 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1474 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1475 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1476 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1477 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1478 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1479 and x_stack_pointer_delta from int to poly_int64.
1480 * dojump.c (do_pending_stack_adjust): Update accordingly.
1481 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1482 stack_pointer_deltas.
1483 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1484 (pad_to_arg_alignment): Track polynomial offsets.
1485 (assign_parm_find_stack_rtl): Likewise.
1486 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1487 * toplev.c (output_stack_usage): Update reference to
1488 current_function_pushed_stack_size.
1490 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1491 Alan Hayward <alan.hayward@arm.com>
1492 David Sherwood <david.sherwood@arm.com>
1494 * function.c (in_arg_offset, var_offset, dynamic_offset)
1495 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1496 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1497 rather than a HOST_WIDE_INT.
1498 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1499 (instantiate_virtual_regs_in_insn): Likewise.
1501 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1502 Alan Hayward <alan.hayward@arm.com>
1503 David Sherwood <david.sherwood@arm.com>
1505 * rtl.h (get_args_size, add_args_size_note): New functions.
1506 (find_args_size_adjust): Return a poly_int64 rather than a
1508 (fixup_args_size_notes): Likewise. Make the same change to the
1509 end_args_size parameter.
1510 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1511 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1512 * calls.c (emit_call_1): Likewise.
1513 * explow.c (adjust_stack_1): Likewise.
1514 * cfgcleanup.c (old_insns_match_p): Update use of
1515 find_args_size_adjust.
1516 * combine.c (distribute_notes): Track polynomial arg sizes.
1517 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1518 end_true_args_size, beg_delay_args_size and end_delay_args_size
1519 from HOST_WIDE_INT to poly_int64.
1520 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1521 than a HOST_WIDE_INT.
1522 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1523 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1524 polynomial arg sizes.
1525 * emit-rtl.c (try_split): Use get_args_size.
1526 * recog.c (peep2_attempt): Likewise.
1527 * reload1.c (reload_as_needed): Likewise.
1528 * expr.c (find_args_size_adjust): Return the adjustment as a
1529 poly_int64 rather than a HOST_WIDE_INT.
1530 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1531 to a poly_int64 and change the return type in the same way.
1532 (emit_single_push_insn): Track polynomial arg sizes.
1534 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1535 Alan Hayward <alan.hayward@arm.com>
1536 David Sherwood <david.sherwood@arm.com>
1538 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1539 from HOST_WIDE_INT to poly_int64.
1540 * expr.c (push_block, emit_push_insn): Likewise.
1542 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1543 Alan Hayward <alan.hayward@arm.com>
1544 David Sherwood <david.sherwood@arm.com>
1546 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1548 (get_frame_size): Return the size as a poly_int64 rather than a
1550 (frame_offset_overflow): Take the offset as a poly_int64 rather
1551 than a HOST_WIDE_INT.
1552 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1553 (assign_stack_temp): Likewise for the size.
1554 * function.c (get_frame_size): Return a poly_int64 rather than
1556 (frame_offset_overflow): Take the offset as a poly_int64 rather
1557 than a HOST_WIDE_INT.
1558 (try_fit_stack_local): Take the start, length and size as poly_int64s
1559 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1560 rather than a HOST_WIDE_INT.
1561 (add_frame_space): Take the start and end as poly_int64s rather than
1563 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1564 (assign_stack_temp): Likewise for the size.
1565 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1567 (find_temp_slot_from_address): Handle polynomial offsets.
1568 (combine_temp_slots): Likewise.
1569 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1571 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1572 poly_int64 rather than a HOST_WIDE_INT.
1573 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1574 than a HOST_WIDE_INT.
1575 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1576 polynomial frame offsets.
1577 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1578 as a poly_int64 rather than an int.
1579 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1580 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1581 * config/v850/v850.c (compute_frame_size): Likewise.
1582 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1583 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1584 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1585 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1586 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1587 rather than a HOST_WIDE_INT.
1588 * explow.c (get_dynamic_stack_base): Likewise.
1589 * final.c (final_start_function): Use the constant lower bound
1590 of the frame size for -Wframe-larger-than.
1591 * ira.c (do_reload): Adjust for new get_frame_size return type.
1592 * lra.c (lra): Likewise.
1593 * reload1.c (reload): Likewise.
1594 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1595 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1596 * rtlanal.c (get_initial_register_offset): Return the offset as
1597 a poly_int64 rather than a HOST_WIDE_INT.
1599 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1600 Alan Hayward <alan.hayward@arm.com>
1601 David Sherwood <david.sherwood@arm.com>
1603 * reload1.c (elim_table): Change initial_offset, offset and
1604 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1605 (offsets_at): Change the target array's element type from
1606 HOST_WIDE_INT to poly_int64_pod.
1607 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1608 (elimination_costs_in_insn, update_eliminable_offsets)
1609 (verify_initial_elim_offsets, set_offsets_for_label)
1610 (init_eliminable_invariants): Update after above changes.
1612 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1613 Alan Hayward <alan.hayward@arm.com>
1614 David Sherwood <david.sherwood@arm.com>
1616 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1617 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1618 (decomposition): Change start and end from HOST_WIDE_INT
1620 (decompose, immune_p): Update accordingly.
1621 (find_inc_amount): Return a poly_int64 rather than an int.
1622 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1625 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1626 Alan Hayward <alan.hayward@arm.com>
1627 David Sherwood <david.sherwood@arm.com>
1629 * tree.h (get_inner_reference): Return the bitsize and bitpos
1630 as poly_int64_pods rather than HOST_WIDE_INT.
1631 * fold-const.h (ptr_difference_const): Return the pointer difference
1632 as a poly_int64_pod rather than a HOST_WIDE_INT.
1633 * expr.c (get_inner_reference): Return the bitsize and bitpos
1634 as poly_int64_pods rather than HOST_WIDE_INT.
1635 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1637 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1638 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1639 (optimize_bit_field_compare): Update call to get_inner_reference.
1640 (decode_field_reference): Likewise.
1641 (fold_unary_loc): Track polynomial offsets and sizes.
1642 (split_address_to_core_and_offset): Return the bitpos as a
1643 poly_int64_pod rather than a HOST_WIDE_INT.
1644 (ptr_difference_const): Likewise for the pointer difference.
1645 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1646 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1647 * dbxout.c (dbxout_expand_expr): Likewise.
1648 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1649 (loc_list_from_tree_1, fortran_common): Likewise.
1650 * gimple-laddress.c (pass_laddress::execute): Likewise.
1651 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1652 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1653 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1654 * tree-affine.c (tree_to_aff_combination): Likewise.
1655 (get_inner_reference_aff): Likewise.
1656 * tree-data-ref.c (split_constant_offset_1): Likewise.
1657 (dr_analyze_innermost): Likewise.
1658 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1659 * tree-sra.c (ipa_sra_check_caller): Likewise.
1660 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1661 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1662 (instrument_bool_enum_load, instrument_object_size): Likewise.
1663 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1664 to get_inner_reference.
1665 * hsa-gen.c (gen_hsa_addr): Likewise.
1666 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1667 * tsan.c (instrument_expr): Likewise.
1668 * match.pd: Update call to ptr_difference_const.
1670 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1671 Alan Hayward <alan.hayward@arm.com>
1672 David Sherwood <david.sherwood@arm.com>
1674 * fold-const.c (fold_comparison): Track sizes and offsets as
1675 poly_int64s rather than HOST_WIDE_INTs when folding address
1678 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1679 Alan Hayward <alan.hayward@arm.com>
1680 David Sherwood <david.sherwood@arm.com>
1682 * expr.h (get_bit_range): Return the bitstart and bitend as
1683 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1684 as a poly_int64 rather than a HOST_WIDE_INT.
1685 * expr.c (get_bit_range): Likewise.
1686 (expand_assignment): Update call accordingly.
1687 * fold-const.c (optimize_bit_field_compare): Likewise.
1689 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1690 Alan Hayward <alan.hayward@arm.com>
1691 David Sherwood <david.sherwood@arm.com>
1693 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1694 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1695 as an optional final parameter.
1696 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1697 (aarch64_legitimate_address_p): Likewise.
1698 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1699 instead of an rtx code.
1700 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1701 (aarch64_legitimate_address_hook_p): Likewise.
1702 (aarch64_print_ldpstp_address): Likewise.
1703 (aarch64_print_operand_address): Likewise.
1704 (aarch64_address_cost): Likewise.
1705 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1706 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1707 (aarch64_mem_pair_lanes_operand): Likewise.
1709 2017-12-20 Richard Biener <rguenther@suse.de>
1711 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1712 update_stmt_if_modified.
1714 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1716 PR tree-optimization/83491
1717 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1718 before walking uses. Improve coding style and comments.
1720 2017-12-20 Tom de Vries <tom@codesourcery.com>
1722 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1724 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1727 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1729 (pass_ipa_free_fn_summary): Add small_p private data member,
1730 initialize to false in the ctor.
1731 (pass_ipa_free_fn_summary::clone,
1732 pass_ipa_free_fn_summary::set_pass_param,
1733 pass_ipa_free_fn_summary::gate): New methods.
1734 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1735 | TODO_dump_symtab if small_p.
1736 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1737 entry and add another instance of the pass with false parm after
1739 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1740 ipa_free_fn_summary here.
1742 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1744 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1746 2017-12-20 Martin Sebor <msebor@redhat.com>
1749 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1750 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1751 (expand_builtin_strncmp): Same.
1753 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1756 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1757 allowed debug stmts before labels.
1758 (expand_gimple_basic_block): Likewise.
1759 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1760 * gimple-iterator.h (gsi_after_labels): Likewise.
1761 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1762 rename reused variable, and simplify using gsi_move_before.
1763 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1764 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1765 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1766 (gimple_verify_flow_info, gimple_block_label): Likewise.
1767 (make_blocks): Move debug markers after adjacent labels.
1768 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1769 allowed debug insns outside blocks.
1770 * df-scan.c (df_insn_delete): Likewise.
1771 * lra-constraints.c (update_ebb_live_info): Likewise.
1772 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1773 (vt_initialize, delete_vta_debug_insns): Likewise.
1774 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1776 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1777 Alan Hayward <alan.hayward@arm.com>
1778 David Sherwood <david.sherwood@arm.com>
1780 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1781 (round_up_to_byte_boundary): Likewise.
1782 * expr.h (get_bit_range): Add temporary shim.
1783 * gimple-ssa-store-merging.c (store_operand_info): Change the
1784 bitsize, bitpos, bitregion_start and bitregion_end fields from
1785 unsigned HOST_WIDE_INT to poly_uint64.
1786 (merged_store_group): Likewise load_align_base.
1787 (compatible_load_p, compatible_load_p): Update accordingly.
1788 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1789 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1790 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1791 bitregion_start and bitregion_end as poly_uint64s rather than
1792 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1793 (handled_load): Take the bitsize, bitpos,
1794 bitregion_start and bitregion_end as poly_uint64s rather than
1795 unsigned HOST_WIDE_INTs.
1796 (pass_store_merging::process_store): Update call to
1797 mem_valid_for_store_merging.
1799 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1800 Alan Hayward <alan.hayward@arm.com>
1801 David Sherwood <david.sherwood@arm.com>
1803 * builtins.c (get_object_alignment_2): Track polynomial offsets
1804 and sizes. Update the alignment handling.
1806 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1807 Alan Hayward <alan.hayward@arm.com>
1808 David Sherwood <david.sherwood@arm.com>
1810 * tree.h (get_inner_reference): Add a version that returns the
1811 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1812 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1813 the case in which bitpos is not associated with the first byte.
1815 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1816 Alan Hayward <alan.hayward@arm.com>
1817 David Sherwood <david.sherwood@arm.com>
1819 * tree-affine.h (get_inner_reference_aff): Return the size as a
1821 * tree-affine.c (get_inner_reference_aff): Likewise.
1822 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1823 * tree-ssa-loop-im.c (mem_refs_may_alias_p): 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 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1830 HOST_WIDE_INT rather than a poly_int64.
1832 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1833 Alan Hayward <alan.hayward@arm.com>
1834 David Sherwood <david.sherwood@arm.com>
1836 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1837 HOST_WIDE_INT to poly_int64_pod.
1838 (perform_symbolic_merge): Update accordingly.
1839 (bswap_replace): Likewise.
1841 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1842 Alan Hayward <alan.hayward@arm.com>
1843 David Sherwood <david.sherwood@arm.com>
1845 * tree-affine.h (aff_tree::offset): Change from widest_int
1847 (wide_int_ext_for_comb): Delete.
1848 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1849 constants as poly_widest_int rather than widest_int.
1850 (aff_combination_constant_multiple_p): Return the multiplier
1851 as a poly_widest_int.
1852 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1854 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1855 version static and add an overload for poly_widest_int.
1856 (aff_combination_const, aff_combination_add_cst)
1857 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1858 the constants as poly_widest_int rather than widest_int.
1859 (tree_to_aff_combination): Generalize INTEGER_CST case to
1861 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1862 (aff_combination_add_product, aff_combination_mult): Handle
1864 (aff_combination_constant_multiple_p): Return the multiplier
1865 as a poly_widest_int.
1866 * tree-predcom.c (determine_offset): Return the offset as a
1868 (split_data_refs_to_components, suitable_component_p): Update
1870 (valid_initializer_p): Update call to
1871 aff_combination_constant_multiple_p.
1872 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1873 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1874 as a poly_int64 rather than a HOST_WIDE_INT.
1875 (get_address_cost): Handle polynomial offsets.
1876 (iv_elimination_compare_lt): Likewise.
1877 (rewrite_use_nonlinear_expr): Likewise.
1879 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1880 Alan Hayward <alan.hayward@arm.com>
1881 David Sherwood <david.sherwood@arm.com>
1883 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1884 as a poly_int64_pod rather than a HOST_WIDE_INT.
1885 (get_addr_base_and_unit_offset): Likewise.
1886 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1887 (get_addr_base_and_unit_offset): Likewise.
1888 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1889 to poly_int64 in example.
1890 * fold-const.c (fold_binary_loc): Update call to
1891 get_addr_base_and_unit_offset.
1892 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1893 (maybe_canonicalize_mem_ref_addr): Likewise.
1894 (gimple_fold_stmt_to_constant_1): Likewise.
1895 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1897 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1898 * match.pd: Likewise.
1899 * omp-low.c (lower_omp_target): Likewise.
1900 * tree-sra.c (build_ref_for_offset): Likewise.
1901 (build_debug_ref_for_model): Likewise.
1902 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1903 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1904 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1905 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1906 (constant_pointer_difference): Likewise.
1907 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1908 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1909 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1910 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1911 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1912 (set_ssa_val_to): Likewise.
1913 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1914 (maybe_diag_stxncpy_trunc): Likewise.
1915 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1916 * tree.c (build_simple_mem_ref_loc): Likewise.
1917 (array_at_struct_end_p): Likewise.
1919 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1920 Alan Hayward <alan.hayward@arm.com>
1921 David Sherwood <david.sherwood@arm.com>
1923 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1924 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1925 (get_ref_base_and_extent_hwi): Declare.
1926 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1927 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1928 (get_ref_base_and_extent_hwi): New function.
1929 * cfgexpand.c (expand_debug_expr): Update call to
1930 get_ref_base_and_extent.
1931 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1932 * gimple-fold.c (get_base_constructor): Return the offset as a
1933 poly_int64_pod rather than a HOST_WIDE_INT.
1934 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1935 * ipa-polymorphic-call.c
1936 (ipa_polymorphic_call_context::set_by_invariant)
1937 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1938 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1939 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1940 rather than get_ref_base_and_extent.
1941 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1942 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1943 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1945 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1946 call to get_ref_base_and_extent.
1947 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1948 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1949 (stmt_kills_ref_p): Likewise.
1950 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1951 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1953 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1954 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1955 when calling native_encode_expr.
1956 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1957 call to get_ref_base_and_extent.
1958 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1959 get_ref_base_and_extent.
1960 * var-tracking.c (track_expr_p): Likewise.
1962 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1963 Alan Hayward <alan.hayward@arm.com>
1964 David Sherwood <david.sherwood@arm.com>
1966 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1967 HOST_WIDE_INT to poly_int64_pod.
1968 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1969 polynomail parameter offsets.
1971 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1972 Alan Hayward <alan.hayward@arm.com>
1973 David Sherwood <david.sherwood@arm.com>
1975 * gengtype.c (main): Handle poly_int64_pod.
1976 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1977 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1978 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1979 from HOST_WIDE_INT to poly_int64_pod.
1980 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1981 (copy_cfa): New function.
1982 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1983 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1984 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1985 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1986 (initial_return_save): Treat offsets as poly_ints.
1987 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1988 offset is nonconstant.
1989 (reg_save): Take the offset as a poly_int64. Fall back to
1990 DW_CFA_expression for nonconstant offsets.
1991 (queue_reg_save): Take the offset as a poly_int64.
1992 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1994 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1995 Alan Hayward <alan.hayward@arm.com>
1996 David Sherwood <david.sherwood@arm.com>
1998 * rtl.h (operand_subword, operand_subword_force): Take the offset
1999 as a poly_uint64 an unsigned int.
2000 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
2002 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2003 Alan Hayward <alan.hayward@arm.com>
2004 David Sherwood <david.sherwood@arm.com>
2006 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
2007 'p' format code. Use INT_LIST rather than SUBREG as the example of
2008 a code with an XINT and an XEXP. Remove the implication that
2009 accessing an rtx field using XINT is expected to work.
2010 * rtl.def (SUBREG): Change format from "ei" to "ep".
2011 * rtl.h (rtunion::rt_subreg): New field.
2012 (XCSUBREG): New macro.
2013 (SUBREG_BYTE): Use it.
2014 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
2015 Update constructor accordingly.
2016 (subreg_shape::operator ==): Update accordingly.
2017 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
2018 than an unsigned int.
2019 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
2020 a poly_uint64 rather than an unsigned int.
2021 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
2022 than an unsigned int.
2023 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
2024 (subreg_size_highpart_offset): Return a poly_uint64 rather than
2025 an unsigned int. Take the sizes as poly_uint64s.
2026 (subreg_offset_from_lsb): Return a poly_uint64 rather than
2027 an unsigned int. Take the shift as a poly_uint64 rather than
2029 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
2030 as a poly_uint64 rather than an unsigned int.
2031 (simplify_subreg_regno): Likewise.
2032 (byte_lowpart_offset): Return the memory offset as a poly_int64
2034 (subreg_memory_offset): Likewise. Take the subreg offset as a
2035 poly_uint64 rather than an unsigned int.
2036 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
2037 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
2038 poly_uint64 rather than an unsigned int.
2039 * rtl.c (rtx_format): Describe 'p' in comment.
2040 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
2041 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
2042 offset as a poly_uint64 rather than an unsigned int.
2043 (byte_lowpart_offset): Return the memory offset as a poly_int64
2045 (subreg_memory_offset): Likewise. Take the subreg offset as a
2046 poly_uint64 rather than an unsigned int.
2047 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
2048 mode sizes as poly_uint64s rather than unsigned ints. Return a
2049 poly_uint64 rather than an unsigned int.
2050 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
2051 (copy_insn_1): Handle 'p'.
2052 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
2053 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
2054 an unsigned int. Return the shift in the same way.
2055 (subreg_lsb): Return the shift as a poly_uint64 rather than an
2057 (subreg_size_offset_from_lsb): Take the sizes and shift as
2058 poly_uint64s rather than unsigned ints. Return the offset as
2060 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2061 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2063 * rtlhash.c (add_rtx): Handle 'p'.
2064 * genemit.c (gen_exp): Likewise.
2065 * gengenrtl.c (type_from_format, gendef): Likewise.
2066 * gensupport.c (subst_pattern_match, get_alternatives_number)
2067 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2068 (subst_dup): Likewise.
2069 * gengtype.c (adjust_field_rtx_def): Likewise.
2070 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2071 (match_pattern_2): Likewise.
2072 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2073 (rtx_test::subreg_field): New function.
2074 (operator ==, safe_to_hoist_p, transition_parameter_type)
2075 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2076 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2077 * genpeep.c (match_rtx): Likewise.
2078 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2079 (rtx_writer::print_rtx_operand): Handle 'p'.
2080 (print_value): Handle SUBREG.
2081 * read-rtl.c (apply_int_iterator): Likewise.
2082 (rtx_reader::read_rtx_operand): Handle 'p'.
2083 * alias.c (rtx_equal_for_memref_p): Likewise.
2084 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2085 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2087 * calls.c (expand_call): Likewise.
2088 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2089 (make_extraction, gen_lowpart_for_combine): Likewise.
2090 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2092 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2093 rather than an unsigned int. Treat subreg offsets as poly_ints.
2094 (exp_equiv_p): Handle 'p'.
2095 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2096 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2097 * dse.c (find_shift_sequence): Likewise.
2098 * dwarf2out.c (rtl_for_decl_location): Likewise.
2099 * expmed.c (extract_low_bits): Likewise.
2100 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2101 (expand_expr_real_2): Likewise.
2102 * final.c (alter_subreg): Likewise.
2103 (leaf_renumber_regs_insn): Handle 'p'.
2104 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2105 Treat subreg offsets as poly_ints.
2106 * fwprop.c (forward_propagate_and_simplify): Likewise.
2107 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2108 * ira.c (get_subreg_tracking_sizes): Likewise.
2109 * ira-conflicts.c (go_through_subreg): Likewise.
2110 * ira-lives.c (process_single_reg_class_operands): Likewise.
2111 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2112 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2113 as a poly_uint64 rather than an unsigned int.
2114 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2115 subreg offsets as poly_ints.
2116 * lra-constraints.c (operands_match_p): Handle 'p'.
2117 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2118 * lra-spills.c (assign_mem_slot): Likewise.
2119 * postreload.c (move2add_valid_value_p): Likewise.
2120 * recog.c (general_operand, indirect_operand): Likewise.
2121 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2122 (copyprop_hardreg_forward_1): Likewise.
2123 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2124 (record_subregs_of_mode): Likewise.
2125 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2126 * reload.c (operands_match_p): Handle 'p'.
2127 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2128 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2129 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2130 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2131 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2132 (simplify_const_poly_int_tests<N>::run): Likewise.
2133 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2134 a poly_uint64 rather than an unsigned int.
2135 * valtrack.c (debug_lowpart_subreg): Likewise.
2136 * var-tracking.c (var_lowpart): Likewise.
2137 (loc_cmp): Handle 'p'.
2139 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2140 Alan Hayward <alan.hayward@arm.com>
2141 David Sherwood <david.sherwood@arm.com>
2143 * ira.c (get_subreg_tracking_sizes): New function.
2144 (init_live_subregs): Take an integer size rather than a register.
2145 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2146 to init_live_subregs.
2148 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2149 Alan Hayward <alan.hayward@arm.com>
2150 David Sherwood <david.sherwood@arm.com>
2152 * expr.c (store_constructor_field): Change bitsize from a
2153 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2154 HOST_WIDE_INT to a poly_int64.
2155 (store_constructor): Change size from a HOST_WIDE_INT to
2157 (store_field): Likewise bitsize and bitpos.
2159 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2160 Alan Hayward <alan.hayward@arm.com>
2161 David Sherwood <david.sherwood@arm.com>
2163 * expmed.h (store_bit_field): Change bitregion_start and
2164 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2165 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2166 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2167 (store_fixed_bit_field, store_split_bit_field): Likewise.
2168 * expr.c (store_constructor_field, store_field): Likewise.
2169 (optimize_bitfield_assignment_op): Likewise. Make the same change
2170 to bitsize and bitpos.
2171 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2172 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2173 same change in the constructor arguments.
2174 (get_best_mode): Change bitregion_start and bitregion_end from
2175 unsigned HOST_WIDE_INT to poly_uint64.
2176 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2177 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2179 (bit_field_mode_iterator::next_mode): Update for new types
2180 of m_bitregion_start and m_bitregion_end.
2181 (get_best_mode): Change bitregion_start and bitregion_end from
2182 unsigned HOST_WIDE_INT to poly_uint64.
2184 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2185 Alan Hayward <alan.hayward@arm.com>
2186 David Sherwood <david.sherwood@arm.com>
2188 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2189 accepts poly_uint64 offsets.
2190 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2191 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2192 * expmed.c (lowpart_bit_field_p): Likewise.
2193 (extract_bit_field_as_subreg): New function, split out from...
2194 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2195 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2196 extractions, check that BITSIZE matches the size of the extracted
2197 value and that BITNUM is an exact multiple of that size.
2198 If all else fails, try forcing the value into memory if
2199 BITNUM is variable, and adjusting the address so that the
2200 offset is constant. Split the part that can only handle constant
2201 bitsize and bitnum out into...
2202 (extract_integral_bit_field): ...this new function.
2203 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2204 rather than unsigned HOST_WIDE_INTs.
2206 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2207 Alan Hayward <alan.hayward@arm.com>
2208 David Sherwood <david.sherwood@arm.com>
2210 * expmed.h (store_bit_field): Take bitsize and bitnum as
2211 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2212 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2213 that returns the byte size.
2214 (store_bit_field_1): Take bitsize and bitnum as
2215 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2216 to simple_mem_bitfield_p. Split the part that can only handle
2217 constant bitsize and bitnum out into...
2218 (store_integral_bit_field): ...this new function.
2219 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2220 than unsigned HOST_WIDE_INTs.
2221 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2223 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2224 Alan Hayward <alan.hayward@arm.com>
2225 David Sherwood <david.sherwood@arm.com>
2227 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2228 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2230 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2231 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2232 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2233 offset as a poly_int64 rather than an int.
2234 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2235 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2236 * lra-constraints.c (equiv_address_substitution): Track offsets
2238 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2239 (curr_insn_transform): Handle the new form of sp_offset.
2240 * lra-eliminations.c (lra_elim_table): Change previous_offset
2241 and offset from HOST_WIDE_INT to poly_int64.
2242 (print_elim_table, update_reg_eliminate): Update accordingly.
2243 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2244 (get_elimination): Update accordingly.
2245 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2246 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2247 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2248 poly_int64 offsets generally.
2249 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2250 (mark_not_eliminable, init_elimination): Update accordingly.
2251 (remove_reg_equal_offset_note): Return a bool and pass the new
2252 offset back by pointer as a poly_int64.
2253 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2254 rather than a HOST_WIDE_INT.
2255 (do_remat): Track offsets poly_int64s.
2256 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2258 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2259 Alan Hayward <alan.hayward@arm.com>
2260 David Sherwood <david.sherwood@arm.com>
2262 * rtl.h (mem_attrs): Add a default constructor. Change size and
2263 offset from HOST_WIDE_INT to poly_int64.
2264 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2265 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2266 (widen_memory_access): Take the sizes and offsets as poly_int64s
2267 rather than HOST_WIDE_INTs.
2268 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2269 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2271 (adjust_offset_for_component_ref): Change the offset from a
2272 HOST_WIDE_INT to a poly_int64.
2273 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2274 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2275 * dce.c (find_call_stack_args): Likewise.
2276 * dse.c (record_store): Likewise.
2277 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2278 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2279 * read-rtl-function.c (test_loading_mem): Likewise.
2280 * rtlanal.c (may_trap_p_1): Likewise.
2281 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2282 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2283 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2284 (mem_attrs::mem_attrs): New function.
2285 (set_mem_attributes_minus_bitpos): Change bitpos from a
2286 HOST_WIDE_INT to poly_int64.
2287 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2288 (clear_mem_offset, clear_mem_size, change_address)
2289 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2290 initialize mem_attrs.
2291 (set_mem_offset, set_mem_size, adjust_address_1)
2292 (adjust_automodify_address_1, offset_address, widen_memory_access):
2293 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2295 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2296 Alan Hayward <alan.hayward@arm.com>
2297 David Sherwood <david.sherwood@arm.com>
2299 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2300 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2301 rather than 0 to represent an unknown size. Assert that the size
2302 is known when the mode isn't BLKmode.
2303 (may_trap_p_1): Use -1 for unknown sizes.
2304 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2306 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2307 Alan Hayward <alan.hayward@arm.com>
2308 David Sherwood <david.sherwood@arm.com>
2310 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2311 to poly_int64. Update commentary for positions_needed.large.
2312 (read_info_type): Change offset and width from HOST_WIDE_INT
2314 (set_usage_bits): Likewise.
2315 (canon_address): Return the offset as a poly_int64 rather than
2316 a HOST_WIDE_INT. Use strip_offset_and_add.
2317 (set_all_positions_unneeded, any_positions_needed_p): Use
2318 positions_needed.large to track stores with non-constant widths.
2319 (all_positions_needed_p): Likewise. Take the offset and width
2320 as poly_int64s rather than ints. Assert that rhs is nonnull.
2321 (record_store): Cope with non-constant offsets and widths.
2322 Nullify the rhs of an earlier store if we can't tell which bytes
2324 (find_shift_sequence): Take the access_size and shift as poly_int64s
2326 (get_stored_val): Take the read_offset and read_width as poly_int64s
2327 rather than HOST_WIDE_INTs.
2328 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2329 non-constant offsets and widths.
2331 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2332 Alan Hayward <alan.hayward@arm.com>
2333 David Sherwood <david.sherwood@arm.com>
2335 * inchash.h (inchash::hash::add_poly_int): New function.
2336 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2337 Use poly_int64 rather than HOST_WIDE_INT.
2338 (ao_ref::max_size_known_p): New function.
2339 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2340 rather than HOST_WIDE_INT.
2341 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2342 to temporaries until its interface is adjusted to match.
2343 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2344 (aliasing_component_refs_p, decl_refs_may_alias_p)
2345 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2346 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2347 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2349 * alias.c (ao_ref_from_mem): Likewise.
2350 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2351 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2352 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2353 (maybe_trim_complex_store, maybe_trim_constructor_store)
2354 (live_bytes_read, dse_classify_store): Likewise.
2355 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2356 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2357 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2358 (vn_reference_lookup_3): Likewise.
2359 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2361 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2362 Alan Hayward <alan.hayward@arm.com>
2363 David Sherwood <david.sherwood@arm.com>
2365 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2366 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2367 instead of ranges_overlap_p.
2369 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2370 Alan Hayward <alan.hayward@arm.com>
2371 David Sherwood <david.sherwood@arm.com>
2373 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2374 sizes as poly_int64s rather than HOST_WIDE_INTs.
2376 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2377 Alan Hayward <alan.hayward@arm.com>
2378 David Sherwood <david.sherwood@arm.com>
2380 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2381 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2382 * gimple-fold.c (fold_ctor_reference): Likewise.
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 (dwarf_poly_indeterminate_value): New hook.
2389 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2390 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2391 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2392 * doc/tm.texi: Regenerate.
2393 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2394 offset as a poly_int64.
2395 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2396 offset as a poly_int64.
2397 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2398 Take the offset as a poly_int64.
2399 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2400 (frame_pointer_fb_offset): Change to a poly_int64.
2401 (int_loc_descriptor): Take the offset as a poly_int64. Use
2402 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2403 (based_loc_descr): Take the offset as a poly_int64.
2404 Use strip_offset_and_add to handle (plus X (const)).
2405 Use new_reg_loc_descr instead of an open-coded version of the
2406 previous implementation.
2407 (mem_loc_descriptor): Handle CONST_POLY_INT.
2408 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2409 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2411 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2412 Alan Hayward <alan.hayward@arm.com>
2413 David Sherwood <david.sherwood@arm.com>
2415 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2416 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2417 * inchash.h (inchash::hash::add_poly_hwi): New function.
2418 * gengtype.c (main): Register poly_int64.
2419 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2420 offset as a poly_int.
2421 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2422 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2423 offset as a poly_int64.
2424 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2425 * print-rtl.c (print_poly_int): New function.
2426 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2428 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2430 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2431 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2432 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2433 Enforce track_offset_p for parts of a PARALLEL.
2434 (vt_add_function_parameter): Use const_offset for the final
2435 offset to track. Use get_tracked_reg_offset for the parts
2438 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2439 Alan Hayward <alan.hayward@arm.com>
2440 David Sherwood <david.sherwood@arm.com>
2442 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2443 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2444 * doc/tm.texi: Regenerate.
2445 * hooks.h (hook_bool_uint_uint_true): Delete.
2446 (hook_bool_puint64_puint64_true): Declare.
2447 * hooks.c (hook_bool_uint_uint_true): Delete.
2448 (hook_bool_puint64_puint64_true): New function.
2449 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2450 instead of unsigned ints.
2451 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2452 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2454 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2455 Alan Hayward <alan.hayward@arm.com>
2456 David Sherwood <david.sherwood@arm.com>
2458 * optabs.h (expand_operand): Add an int_value field.
2459 (create_expand_operand): Add an int_value parameter and use it
2460 to initialize the new expand_operand field.
2461 (create_integer_operand): Replace with a declaration of a function
2462 that accepts poly_int64s. Move the implementation to...
2463 * optabs.c (create_integer_operand): ...here.
2464 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2465 the mode preserves the value of int_value, instead of calling
2466 const_int_operand on the rtx. Use gen_int_mode to generate
2469 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2470 Alan Hayward <alan.hayward@arm.com>
2471 David Sherwood <david.sherwood@arm.com>
2473 * dumpfile.h (dump_dec): Declare.
2474 * dumpfile.c (dump_dec): New function.
2475 * pretty-print.h (pp_wide_integer): Turn into a function and
2476 declare a poly_int version.
2477 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2479 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2480 Alan Hayward <alan.hayward@arm.com>
2481 David Sherwood <david.sherwood@arm.com>
2483 * doc/generic.texi (POLY_INT_CST): Document.
2484 * tree.def (POLY_INT_CST): New tree code.
2485 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2486 * tree-core.h (tree_poly_int_cst): New struct.
2487 (tree_node): Add a poly_int_cst field.
2488 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2489 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2490 instead of a wide_int_ref.
2491 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2493 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2494 instead of an unsigned HOST_WIDE_INT.
2495 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2496 (ptrdiff_tree_p): Declare.
2497 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2498 extern inline implementations if the target doesn't use POLY_INT_CST.
2499 (poly_int_tree_p): New function.
2500 (wi::unextended_tree): New class.
2501 (wi::int_traits <unextended_tree>): New override.
2502 (wi::extended_tree): Add a default constructor.
2503 (wi::extended_tree::get_tree): New function.
2504 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2505 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2506 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2507 (wi::tree_to_poly_wide_ref): New typedefs.
2508 (wi::ints_for): Provide overloads for extended_tree and
2510 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2511 (wi::to_wide): New functions.
2512 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2513 * tree.c (poly_int_cst_hasher): New struct.
2514 (poly_int_cst_hash_table): New variable.
2515 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2516 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2518 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2519 (init_ttree): Initialize poly_int_cst_hash_table.
2520 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2521 a poly_int64 instead of a HOST_WIDE_INT.
2522 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2523 instead of an unsigned HOST_WIDE_INT.
2524 (wide_int_to_tree): Rename to...
2525 (wide_int_to_tree_1): ...this.
2526 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2527 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2528 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2529 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2530 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2531 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2533 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2534 (streamer_read_tree_body): Likewise.
2535 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2536 (streamer_write_tree_body): Likewise.
2537 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2538 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2539 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2540 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2541 * gimple-expr.h (is_gimple_constant): Likewise.
2542 * gimplify.c (maybe_with_size_expr): Likewise.
2543 * print-tree.c (print_node): Likewise.
2544 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2545 * tree-pretty-print.c (dump_generic_node): Likewise.
2546 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2547 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2548 * tree-vrp.c (compare_values_warnv): Likewise.
2549 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2550 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2551 (force_expr_to_var_cost): Likewise.
2552 * tree-ssa-loop.c (for_each_index): Likewise.
2553 * fold-const.h (build_invariant_address, size_int_kind): Take a
2554 poly_int64 instead of a HOST_WIDE_INT.
2555 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2556 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2558 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2559 (int_const_binop_2): New function, split out from...
2560 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2561 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2562 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2563 * gimple-ssa-strength-reduction.c (slsr_process_add)
2564 (slsr_process_mul): Check for INTEGER_CSTs before using them
2566 * stor-layout.c (bits_from_bytes): New function.
2567 (bit_from_pos): Use it.
2568 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2569 by BITS_PER_UNIT to get the TYPE_SIZE.
2570 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2571 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2573 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2574 Alan Hayward <alan.hayward@arm.com>
2575 David Sherwood <david.sherwood@arm.com>
2577 * doc/rtl.texi (const_poly_int): Document. Also document the
2578 rtl sharing behavior.
2579 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2580 * rtl.h (const_poly_int_def): New struct.
2581 (rtx_def::u): Add a cpi field.
2582 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2583 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2584 (wi::rtx_to_poly_wide_ref): New typedef
2585 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2586 (poly_int_rtx_p): New functions.
2587 (trunc_int_for_mode): Declare a poly_int64 version.
2588 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2589 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2591 (strip_offset): Declare.
2592 (strip_offset_and_add): New function.
2593 * rtl.def (CONST_POLY_INT): New rtx code.
2594 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2595 (shared_const_p): Use poly_int_rtx_p.
2596 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2598 (gen_int_shift_amount): Likewise.
2599 * emit-rtl.c (const_poly_int_hasher): New class.
2600 (const_poly_int_htab): New variable.
2601 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2602 (const_poly_int_hasher::hash): New function.
2603 (const_poly_int_hasher::equal): Likewise.
2604 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2605 (immed_wide_int_const): Rename to...
2606 (immed_wide_int_const_1): ...this and make static.
2607 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2608 instead of a wide_int_ref.
2609 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2610 (gen_lowpart_common): Handle CONST_POLY_INT.
2611 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2612 * cselib.c (cselib_hash_rtx): Likewise.
2613 * dwarf2out.c (const_ok_for_output_1): Likewise.
2614 * expr.c (convert_modes): Likewise.
2615 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2616 * rtlhash.c (add_rtx): Likewise.
2617 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2618 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2619 Handle existing CONST_POLY_INT rtxes.
2620 * expmed.h (expand_shift): Take a poly_int64 instead of a
2622 * expmed.c (expand_shift): Likewise.
2623 * rtlanal.c (strip_offset): New function.
2624 (commutative_operand_precedence): Give CONST_POLY_INT the same
2625 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2627 * rtl-tests.c (const_poly_int_tests): New struct.
2628 (rtl_tests_c_tests): Use it.
2629 * simplify-rtx.c (simplify_const_unary_operation): Handle
2631 (simplify_const_binary_operation): Likewise.
2632 (simplify_binary_operation_1): Fold additions of symbolic constants
2633 and CONST_POLY_INTs.
2634 (simplify_subreg): Handle extensions and truncations of
2636 (simplify_const_poly_int_tests): New struct.
2637 (simplify_rtx_c_tests): Use it.
2638 * wide-int.h (storage_ref): Add default constructor.
2639 (wide_int_ref_storage): Likewise.
2640 (trailing_wide_ints): Use GTY((user)).
2641 (trailing_wide_ints::operator[]): Add a const version.
2642 (trailing_wide_ints::get_precision): New function.
2643 (trailing_wide_ints::extra_size): Likewise.
2645 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2646 Alan Hayward <alan.hayward@arm.com>
2647 David Sherwood <david.sherwood@arm.com>
2649 * emit-rtl.h (gen_int_shift_amount): Declare.
2650 * emit-rtl.c (gen_int_shift_amount): New function.
2651 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2653 * calls.c (shift_return_value): Likewise.
2654 * cse.c (fold_rtx): Likewise.
2655 * dse.c (find_shift_sequence): Likewise.
2656 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2657 (expand_shift, expand_smod_pow2): Likewise.
2658 * lower-subreg.c (shift_cost): Likewise.
2659 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2660 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2661 (expand_vec_perm_var): Likewise.
2662 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2663 (simplify_binary_operation_1): Likewise.
2664 * combine.c (try_combine, find_split_point, force_int_to_mode)
2665 (simplify_shift_const_1, simplify_shift_const): Likewise.
2666 (change_zero_ext): Likewise. Use simplify_gen_binary.
2668 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2670 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2672 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2674 * doc/invoke.texi (ARM Options): Document accepted extension options
2675 for -march=armv8.3-a.
2677 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2680 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2681 or --with-float=softfp, set the default CPU to arm10e.
2683 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2685 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2686 * config/visium/predicates.md (const_shift_operand): Likewise.
2687 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2688 (visium_legitimize_reload_address): Likewise.
2690 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2693 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2695 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2698 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2700 2017-12-20 Julia Koval <julia.koval@intel.com>
2702 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2703 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2704 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2705 * config.gcc: Include vpclmulqdqintrin.h.
2706 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2707 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2709 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2710 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2711 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2712 * config/i386/i386.c (isa2_opts): Add -mcx16.
2713 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2714 (ix86_option_override_internal): Move mcx16 to flags2.
2715 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2716 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2717 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2718 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2719 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2720 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2721 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2722 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2723 * doc/invoke.texi: Add -mvpclmulqdq.
2725 2017-12-20 Tom de Vries <tom@codesourcery.com>
2728 * config/i386/i386.c (ix86_static_chain): Move
2729 DECL_STATIC_CHAIN test ...
2730 * calls.c (rtx_for_static_chain): ... here. New function.
2731 * calls.h (rtx_for_static_chain): Declare.
2732 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2733 instead of targetm.calls.static_chain.
2734 * df-scan.c (df_get_entry_block_def_set): Same.
2736 2017-12-19 Tom de Vries <tom@codesourcery.com>
2738 PR tree-optimization/83493
2739 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2742 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2744 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2745 inform with hardcoded english plural handling.
2747 2017-12-18 Jeff Law <law@redhat.com>
2749 PR tree-optimization/83477
2750 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2751 a non-virtual PHI, always push a new range.
2753 2017-12-19 Martin Sebor <msebor@redhat.com>
2756 * builtins.c (compute_objsize): Handle non-constant offsets.
2758 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2760 PR tree-optimization/83444
2761 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2762 character load case, if get_stridx on MEM_REF's operand doesn't
2763 look usable, retry with get_addr_stridx.
2765 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2768 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2769 (variable_tracking_main_1): Keep markers even when VTA fails.
2772 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2773 even if there are markers before it.
2774 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2776 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2778 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2779 typical order conditions.
2780 * sel-sched.c (extract_new_fences_from): Likewise.
2781 * config/visium/constraints.md (J, K, L): Likewise.
2782 * config/visium/predicates.md (const_shift_operand): Likewise.
2783 * config/visium/visium.c (visium_legitimize_address,
2784 visium_legitimize_reload_address): Likewise.
2785 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2786 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2787 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2788 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2789 SET_DUMP_DETAIL): Likewise.
2790 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2791 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2792 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2793 avr_set_core_architecture, avr_set_current_function,
2794 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2795 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2796 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2797 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2798 avr_map_decompose, avr_fold_builtin): Likewise.
2799 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2800 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2801 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2802 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2803 * config/m32c/m32c.c (m32c_conditional_register_usage,
2804 m32c_address_cost): Likewise.
2805 * config/m32c/predicates.md (shiftcount_operand,
2806 longshiftcount_operand): Likewise.
2807 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2808 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2809 can_use_cdx_ldstw): Likewise.
2810 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2811 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2813 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2814 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2815 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2816 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2817 * config/vax/vax.c (adjacent_operands_p): Likewise.
2818 * config/ft32/constraints.md (L, b, KA): Likewise.
2819 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2821 * cfgexpand.c (expand_stack_alignment): Likewise.
2822 * gcse.c (insert_expr_in_table): Likewise.
2823 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2824 * cgraphunit.c (cgraph_node::expand): Likewise.
2825 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2826 * emit-rtl.c (add_insn): Likewise.
2827 * input.c (dump_location_info): Likewise.
2828 * passes.c (NEXT_PASS): Likewise.
2829 * read-rtl-function.c (parse_note_insn_name,
2830 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2832 * sched-rgn.c (sched_rgn_init): Likewise.
2833 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2834 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2835 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2837 * explow.c (eliminate_constant_term): Likewise.
2838 * final.c (leaf_renumber_regs_insn): Likewise.
2839 * cfgrtl.c (print_rtl_with_bb): Likewise.
2840 * genhooks.c (emit_init_macros): Likewise.
2841 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2842 * tree-data-ref.c (conflict_fn): Likewise.
2843 * selftest.c (assert_streq): Likewise.
2844 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2845 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2846 fold_binary_loc, multiple_of_p): Likewise.
2847 * reload.c (push_reload, find_equiv_reg): Likewise.
2848 * et-forest.c (et_nca, et_below): Likewise.
2849 * dbxout.c (dbxout_symbol_location): Likewise.
2850 * reorg.c (relax_delay_slots): Likewise.
2851 * dojump.c (do_compare_rtx_and_jump): Likewise.
2852 * gengtype-parse.c (type): Likewise.
2853 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2854 simplify_const_relational_operation): Likewise.
2855 * reload1.c (do_output_reload): Likewise.
2856 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2857 * gengtype.c (type_for_name): Likewise.
2858 * gimple-ssa-sprintf.c (format_directive): Likewise.
2860 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2863 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2864 accessing it. Adjust comment.
2866 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2869 * predict.c (zero_one_minusone): New function.
2870 (apply_return_prediction): Avoid return prediction for functions
2871 returning only -1, 0 and 1 values, unless they only return -1 and 0
2874 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2876 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2877 fall-through warning.
2878 (arc_compute_frame_size): Remove unused variables.
2879 (arc_print_operand): Fix fprintif format.
2880 (arc_can_follow_jump): Clean fall-through warning.
2882 2017-12-19 Marek Polacek <polacek@redhat.com>
2885 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2888 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2890 * config/arc/arc.c (overriderregs): New variable.
2891 (arc_override_options): Track fixed/call saved/call options.
2892 (arc_conditional_register_usage): Check against overrideregs
2893 variable whenever we change register properties.
2895 2017-12-19 Nathan Sidwell <nathan@acm.org>
2897 * opts.c (finish_options): Don't prefix dump_base_name if it
2898 already contains directories.
2900 2017-12-19 Martin Liska <mliska@suse.cz>
2902 PR rtl-optimization/82675
2903 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2904 more element in sbitmap.
2906 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2908 * gimplify.c (gimplify_expr): Use error_operand_p.
2910 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2913 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2916 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2918 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2919 the jump is to a label.
2921 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2923 PR tree-optimization/83444
2924 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2929 * ipa-inline.c (flatten_remove_node_hook): New function.
2930 (ipa_inline): Keep only nodes with flatten attribute at the end of
2931 the array in the order from ipa_reverse_postorder, only walk that
2932 portion of array for flattening, if there is more than one such
2933 node, temporarily register a removal hook and ignore removed nodes.
2935 PR tree-optimization/80631
2936 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2937 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2938 IFN_REDUC_MAX or IFN_REDUC_MIN.
2940 2017-12-18 Jeff Law <law@redhat.com>
2942 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2943 record symbolic equivalences from backedges in the CFG.
2946 2017-11-19 Jeff Law <law@redhat.com>
2948 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2949 of degenerates resulting from ignoring an edge.
2951 2017-12-18 Martin Sebor <msebor@redhat.com>
2954 PR tree-optimization/78450
2955 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2956 (handle_builtin_strlen): Call it.
2958 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2960 PR rtl-optimization/83424
2961 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2963 2017-12-18 Marek Polacek <polacek@redhat.com>
2966 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2967 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2970 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2972 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2973 (umaddsidi4, umaddsidi_split): Likewise.
2975 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2977 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2980 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2983 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2985 2017-12-18 Richard Biener <rguenther@suse.de>
2987 PR tree-optimization/81877
2988 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2989 (outermost_indep_loop): Adjust.
2990 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2991 (can_sm_ref_p): Adjust.
2993 2017-12-18 Richard Biener <rguenther@suse.de>
2996 * tree.c (array_at_struct_end_p): Return true if the underlying
2997 object has space for at least one element in excess of what
2998 the array domain specifies.
3000 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3002 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
3003 example to avoid overfull hbox.
3004 * doc/invoke.texi (Option Summary): Add missing @gol.
3005 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
3008 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3009 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3011 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
3012 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
3013 (AArch64 Options): Document -mverbose-cost-dump.
3014 (ARM Options): Likewise, plus -mflip-thumb.
3016 2017-12-17 Martin Sebor <msebor@redhat.com>
3019 * gimple-ssa-warn-restrict.c
3020 (builtin_memref::offset_out_of_bounds): Correct the handling of
3023 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3025 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
3027 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
3029 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
3031 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
3033 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
3036 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
3038 * doc/invoke.texi: Fix some typos.
3040 2017-12-16 Martin Sebor <msebor@redhat.com>
3042 PR tree-optimization/78918
3043 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
3044 * builtins.c (check_sizes): Rename...
3045 (check_access): ...to this. Rename function arguments for clarity.
3046 (check_memop_sizes): Adjust names.
3047 (expand_builtin_memchr, expand_builtin_memcpy): Same.
3048 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
3049 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
3050 (check_strncat_sizes, expand_builtin_strncat): Same.
3051 (expand_builtin_strncpy, expand_builtin_memset): Same.
3052 (expand_builtin_bzero, expand_builtin_memcmp): Same.
3053 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
3054 (maybe_emit_sprintf_chk_warning): Same.
3055 (expand_builtin_strcpy): Adjust.
3056 (expand_builtin_stpcpy): Same.
3057 (expand_builtin_with_bounds): Detect out-of-bounds accesses
3058 in pointer-checking forms of memcpy, memmove, and mempcpy.
3059 (gcall_to_tree_minimal, max_object_size): Define new functions.
3060 * builtins.h (max_object_size): Declare.
3061 * calls.c (alloc_max_size): Call max_object_size instead of
3062 hardcoding ssizetype limit.
3063 (get_size_range): Handle new argument.
3064 * calls.h (get_size_range): Add a new argument.
3065 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3066 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3067 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3069 (gimple_fold_builtin_memory_chk): Same.
3070 (gimple_fold_builtin_stxcpy_chk): New function.
3071 * gimple-ssa-warn-restrict.c: New source.
3072 * gimple-ssa-warn-restrict.h: New header.
3073 * gimple.c (gimple_build_call_from_tree): Propagate location.
3074 * passes.def (pass_warn_restrict): Add new pass.
3075 * tree-pass.h (make_pass_warn_restrict): Declare.
3076 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3078 (handle_builtin_strcat): Same.
3079 (strlen_optimize_stmt): Rename...
3080 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3081 stpncpy, strncpy, and their checking forms.
3083 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3085 PR rtl-optimization/82849
3086 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3087 and get_max_loop_iterations_int.
3089 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3090 Alan Hayward <alan.hayward@arm.com>
3091 David Sherwood <david.sherwood@arm.com>
3093 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3094 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3095 size as a poly_uint64.
3096 (mode_for_vector, mode_for_int_vector): Take the number of vector
3097 elements as a poly_uint64.
3098 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3099 size as a poly_uint64.
3100 (mode_for_vector, mode_for_int_vector): Take the number of vector
3101 elements as a poly_uint64.
3103 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3104 Alan Hayward <alan.hayward@arm.com>
3105 David Sherwood <david.sherwood@arm.com>
3107 * machmode.h (MACRO_MODE): New macro.
3108 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3109 * targhooks.c (default_libcall_value, default_secondary_reload)
3110 (default_memory_move_cost, default_register_move_cost)
3111 (default_class_max_nregs): Likewise.
3113 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3114 Alan Hayward <alan.hayward@arm.com>
3115 David Sherwood <david.sherwood@arm.com>
3117 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3118 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3120 * genautomata.c (main): Likewise.
3121 * genconditions.c (write_header): Likewise.
3122 * genemit.c (main): Likewise.
3123 * genextract.c (print_header): Likewise.
3124 * genopinit.c (main): Likewise.
3125 * genoutput.c (output_prologue): Likewise.
3126 * genpeep.c (main): Likewise.
3127 * genpreds.c (write_insn_preds_c): Likewise.
3128 * genrecog.c (writer_header): Likewise.
3129 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3130 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3131 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3132 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3133 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3134 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3135 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3136 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3137 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3138 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3139 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3140 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3141 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3142 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3143 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3144 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3145 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3146 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3147 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3148 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3149 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3150 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3151 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3152 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3153 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3154 * config/darwin.c (IN_TARGET_CODE): Likewise.
3155 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3156 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3157 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3158 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3159 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3160 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3161 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3162 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3163 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3164 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3165 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3166 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3167 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3168 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3169 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3170 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3171 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3172 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3173 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3174 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3175 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3176 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3177 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3178 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3179 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3180 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3181 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3182 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3183 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3184 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3185 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3186 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3187 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3188 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3189 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3190 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3191 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3192 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3193 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3194 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3195 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3196 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3197 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3198 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3199 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3200 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3201 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3202 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3203 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3204 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3205 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3206 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3207 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3208 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3209 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3210 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3211 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3212 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3213 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3214 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3215 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3216 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3217 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3218 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3219 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3220 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3221 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3222 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3223 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3224 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3225 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3226 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3227 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3228 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3229 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3230 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3231 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3232 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3233 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3234 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3235 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3236 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3237 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3238 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3239 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3240 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3241 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3242 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3243 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3244 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3245 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3246 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3247 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3248 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3249 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3250 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3251 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3252 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3253 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3254 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3255 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3256 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3257 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3258 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3259 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3260 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3262 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3263 Alan Hayward <alan.hayward@arm.com>
3264 David Sherwood <david.sherwood@arm.com>
3266 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3269 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3270 Alan Hayward <alan.hayward@arm.com>
3271 David Sherwood <david.sherwood@arm.com>
3273 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3274 * doc/md.texi (vec_series@var{m}): Document.
3275 * tree.def (VEC_SERIES_EXPR): New tree code.
3276 * tree.h (build_vec_series): Declare.
3277 * tree.c (build_vec_series): New function.
3278 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3279 * tree-pretty-print.c (dump_generic_node): Likewise.
3280 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3281 * tree-inline.c (estimate_operator_cost): Likewise.
3282 * expr.c (expand_expr_real_2): Likewise.
3283 * optabs-tree.c (optab_for_tree_code): Likewise.
3284 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3285 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3286 * expmed.c (make_tree): Handle VEC_SERIES.
3287 * optabs.def (vec_series_optab): New optab.
3288 * optabs.h (expand_vec_series_expr): Declare.
3289 * optabs.c (expand_vec_series_expr): New function.
3290 * tree-vect-generic.c (expand_vector_operations_1): Check that
3291 the operands also have vector type.
3293 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3294 Alan Hayward <alan.hawyard@arm.com>
3295 David Sherwood <david.sherwood@arm.com>
3297 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3298 (VEC_COND_EXPR): Add missing @tindex.
3299 * doc/md.texi (vec_duplicate@var{m}): Document.
3300 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3301 * tree.c (build_vector_from_val): Add stubbed-out handling of
3302 variable-length vectors, using VEC_DUPLICATE_EXPR.
3303 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3304 * cfgexpand.c (expand_debug_expr): Likewise.
3305 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3306 * tree-inline.c (estimate_operator_cost): Likewise.
3307 * tree-pretty-print.c (dump_generic_node): Likewise.
3308 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3309 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3310 (test_vec_duplicate_folding): New function.
3311 (fold_const_c_tests): Call it.
3312 * optabs.def (vec_duplicate_optab): New optab.
3313 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3314 * optabs.h (expand_vector_broadcast): Declare.
3315 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3316 vec_duplicate_optab.
3317 * expr.c (store_constructor): Try using vec_duplicate_optab for
3319 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3321 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3324 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3325 div/mod latencies a bit.
3327 2017-12-15 Jeff Law <law@redhat.com>
3329 PR tree-optimization/36550
3330 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3331 (mark_threaded_blocks): Rewrite code to avoid block copying when
3332 optimizing for size. Don't pessimize blocks which will be
3333 copied, but all the statements will be dead.
3335 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3337 PR tree-optimization/81165
3338 * tree-ssa-threadupdate.c (uses_in_bb): New.
3339 (estimate_threading_killed_stmts): New.
3340 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3341 * tree-ssa-threadedge.c
3342 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3345 2017-12-15 Jeff Law <law@redhat.com>
3347 PR tree-optimization/83410
3348 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3349 threads when parallelizing loops.
3351 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3353 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3355 * config/alpha/alpha.c (vms_attribute_table): Swap
3356 affects_type_identity and handler fields, adjust comments.
3357 * config/mips/mips.c (mips_attribute_table): Likewise.
3358 * config/visium/visium.c (visium_attribute_table): Likewise.
3359 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3360 * config/microblaze/microblaze.c (microblaze_attribute_table):
3362 * config/spu/spu.c (spu_attribute_table): Likewise.
3363 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3364 * config/arc/arc.c (arc_attribute_table): Likewise.
3365 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3366 * config/v850/v850.c (v850_handle_interrupt_attribute,
3367 v850_handle_data_area_attribute): Formatting fixes.
3368 (v850_attribute_table): Swap affects_type_identity and handler
3369 fields, adjust comments.
3370 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3371 * config/arm/arm.c (arm_attribute_table): Likewise.
3372 * config/avr/avr.c (avr_attribute_table): Likewise.
3373 * config/s390/s390.c (s390_attribute_table): Likewise.
3374 * config/sh/sh.c (sh_attribute_table): Likewise.
3375 * config/i386/i386.c (ix86_handle_cconv_attribute,
3376 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3377 (ix86_attribute_table): Swap affects_type_identity and handler
3378 fields, adjust comments.
3379 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3380 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3381 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3382 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3383 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3384 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3385 * config/rx/rx.c (rx_attribute_table): Likewise.
3386 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3387 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3388 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3389 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3390 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3391 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3392 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3393 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3394 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3395 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3396 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3398 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3400 (register_scoped_attributes, decl_attributes): Formatting fixes.
3402 PR tree-optimization/83269
3403 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3404 subtraction in arg0's type if type is signed and arg0 is unsigned.
3408 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3409 convert? on @0 instead of convert. Check type of @1, not @0.
3410 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3411 convert? on @0 instead of convert. Check type of @1, not @0.
3412 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3413 only optimize if either both @1 and @2 types are narrower
3414 precision, or both are wider or equal precision, and in the former
3415 case only if both have undefined overflow.
3417 2017-12-15 Richard Biener <rguenther@suse.de>
3420 * internal-fn.def (IFN_NOP): Add.
3421 * internal-fn.c (expand_NOP): Do nothing.
3422 * lto-streamer-in.c (input_function): Instead of removing
3423 sanitizer calls replace them with IFN_NOP calls.
3425 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3426 Alan Hayward <alan.hayward@arm.com>
3427 David Sherwood <david.sherwood@arm.com>
3429 * dse.c (store_info, read_info_type): Replace begin and end with
3431 (print_range): New function.
3432 (set_all_positions_unneeded, any_positions_needed_p)
3433 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3435 (record_store): Likewise. Optimize the case in which all positions
3437 (get_stored_val): Replace read_begin and read_end with read_offset
3439 (replace_read): Update call accordingly.
3441 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3443 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3444 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3445 (loop_cand::loop_cand): Initialize above members.
3446 (loop_cand::supported_operations): Delete.
3447 (loop_cand::can_interchange_p): Inline above function.
3448 (loop_cand::classify_simple_reduction): Record number of constant
3449 initialized simple reductions.
3450 (should_interchange_loops): New parameters. Check stmt cost of loops
3452 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3453 Update call to should_interchange_loops.
3454 (should_interchange_loop_nest): Update call to
3455 should_interchange_loops.
3457 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3460 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3461 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3462 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3464 2017-12-15 Julia Koval <julia.koval@intel.com>
3466 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3467 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3469 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3470 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3471 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3473 2017-12-15 Julia Koval <julia.koval@intel.com>
3475 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3476 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3477 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3478 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3479 _mm_aesenc_epi128): New intrinsics.
3481 2017-12-15 Julia Koval <julia.koval@intel.com>
3483 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3484 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3486 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3487 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3488 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3490 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3492 * gimple-ssa-strength-reduction.c (analyze_increments):
3493 Distinguish replacement costs for constant strides from those for
3496 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3498 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3500 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3502 * doc/invoke.texi: Document -Wcast-function-type.
3503 * recog.h (stored_funcptr): Change signature.
3504 * tree-dump.c (dump_node): Avoid warning.
3505 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3507 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3510 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3513 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3515 PR tree-optimization/83312
3516 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3517 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3518 "val" param, and to cope with arbitrary basic blocks.
3519 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3520 handle NULL_TREE for "val", dropping "bb" param.
3521 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3522 drop "bb" param. Handle NULL_TREE for "val".
3523 (find_case_label_for_value): Make "switch_stmt" param const.
3524 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3526 (vrp_prop::check_all_array_refs): Reimplement as...
3527 (check_array_bounds_dom_walker::before_dom_children): ...this new
3528 vfunc. Replace linear search through BB block list, excluding
3529 those with non-executable in-edges via dominator walk.
3531 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3533 * config/arm/arm.opt (mverbose-cost-dump): New option.
3534 * config/arm/arm.c (arm_rtx_costs): Use it.
3536 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3539 * reload1.c (emit_input_reload_insns): Skip debug markers.
3541 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3543 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3547 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3548 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3549 * config/sh/sh.c: Skip debug insns besides notes.
3550 * config/sh/sh.md: Likewise.
3551 * config/sh/sh_treg_combine.cc: Likewise.
3552 * config/sh/sync.md: Likewise.
3554 2017-12-14 Tom de Vries <tom@codesourcery.com>
3556 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3557 Add item for weak_undefined.
3559 2017-12-14 Richard Biener <rguenther@suse.de>
3561 PR tree-optimization/67842
3562 * sese.h (bb_in_region): Remove #if 0'ed code.
3564 2017-12-14 Richard Biener <rguenther@suse.de>
3566 PR tree-optimization/83326
3567 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3568 may_be_zero parameter and handle it by not marking the first
3569 peeled copy as not exiting the loop.
3570 (try_peel_loop): Likewise.
3571 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3572 to handle the case of constant or zero iterations and perform
3573 loop header copying on-the-fly.
3575 2017-12-14 Richard Biener <rguenther@suse.de>
3577 PR tree-optimization/83418
3578 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3579 Instead of asserting we don't get unfolded comparisons deal with them.
3581 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3584 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3585 basic blocks. Assert debug bind insns don't appear outside of bbs,
3586 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3589 PR tree-optimization/83198
3590 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3591 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3592 value if arg is a REAL_CST with incompatible type.
3594 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3595 Bin Cheng <bin.cheng@arm.com>
3598 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3600 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3603 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3605 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3606 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3608 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3609 Alan Hayward <alan.hayward@arm.com>
3610 David Sherwood <david.sherwood@arm.com>
3612 * poly-int.h: New file.
3613 * poly-int-types.h: Likewise.
3614 * coretypes.h: Include them.
3615 (POLY_INT_CONVERSION): Define.
3616 * target.def (estimated_poly_value): New hook.
3617 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3618 * doc/tm.texi: Regenerate.
3619 * doc/poly-int.texi: New file.
3620 * doc/gccint.texi: Include it.
3621 * doc/rtl.texi: Describe restrictions on subreg modes.
3622 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3623 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3624 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3625 * targhooks.h (default_estimated_poly_value): Declare.
3626 * targhooks.c (default_estimated_poly_value): New function.
3627 * target.h (estimated_poly_value): Likewise.
3628 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3629 (wi::unary_traits): Delete.
3630 (wi::binary_traits::signed_shift_result_type): Define for
3631 offset_int << HOST_WIDE_INT, etc.
3632 (generic_wide_int::operator <<=): Define for all types and use
3633 wi::lshift instead of <<.
3634 (wi::hwi_with_prec): Add a default constructor.
3635 (wi::ints_for): New class.
3636 (operator <<): Define for all wide-int types.
3637 (operator /): New function.
3638 (operator %): Likewise.
3639 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3640 (ASSERT_MAYBE_NE_AT): New macros.
3642 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3643 Dominik Vogt <vogt@linux.vnet.ibm.com>
3646 * emit-rtl.c (init_emit): Remove ??? comment.
3647 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3648 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3649 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3650 alignment of 3 virtual registers to BITS_PER_WORD.
3652 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3654 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3656 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3657 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3659 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3662 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3664 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3666 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3667 with the exception of debug begin stmt markers appear before
3671 * final.c (rest_of_handle_final): Call variable_tracking_main only
3672 if !flag_var_tracking.
3674 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3675 Jakub Jelinek <jakub@redhat.com>
3679 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3680 labels debug stmts that can only appear after labels.
3682 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3684 PR rtl-optimization/82398
3685 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3686 EXPR_USEFULNESS in priority comparison.
3688 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3690 PR rtl-optimization/83393
3691 * combine.c (move_deaths): If reg_stat points to a too new insn in
3692 last_death, do not use it: find the proper insn instead.
3694 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3697 * config/rs6000/rs6000-string.c (expand_block_compare,
3698 expand_strn_compare): Fix set_mem_size() calls.
3700 2017-12-12 Jeff Law <law@redhat.com>
3702 PR tree-optimization/83298
3703 PR tree-optimization/83362
3704 PR tree-optimization/83383
3705 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3706 push_value_range a public interface. Add new argument to
3707 record_ranges_from_stmt.
3708 * gimple-ssa-evrp-analyze.c
3709 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3710 Update comments. Handle recording temporary equivalences.
3711 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3712 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3713 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3714 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3715 gimple-ssa-evrp-analyze.h.
3716 (record_temporary_equivalences_from_phis): Add new argument. When
3717 the PHI arg is an SSA_NAME, set the result's range to the range
3719 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3720 from statements too.
3721 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3722 Pass it down to children as needed.
3723 (thread_outgoing_edges): Likewise.
3724 (thread_across_edge): Likewise. Push/pop range state as needed.
3725 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3727 2017-12-12 Julia Koval <julia.koval@intel.com>
3729 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3730 (PTA_CANNONLAKE): Remove PTA_CLWB.
3732 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3735 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3736 (vcondv2div2df): Likewise.
3737 (vconduv2dfv2di): Likewise.
3739 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3741 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3742 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3743 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3744 ATTR_MATHFN_FPROUNDING_ERRNO.
3746 2017-12-12 Richard Biener <rguenther@suse.de>
3748 PR tree-optimization/83385
3749 * graphite-scop-detection.c (get_order, order): Remove.
3750 (bb_to_rpo): New global.
3752 (build_scops): Sort pbbs in RPO order.
3754 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3756 * combine.c (simplify_set): Do not transform subregs to zero_extends
3757 if the destination mode is a vector mode.
3759 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3761 PR tree-optimization/83359
3762 * tree-cfg.h (fold_loop_internal_call): Declare.
3763 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3764 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3765 (find_loop_dist_alias): New function.
3766 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3767 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3770 PR tree-optimization/80631
3771 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3772 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3773 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3774 hardcoding zero as the value if COND_EXPR is never true. For
3775 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3776 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3777 hardcoding MAX_EXPR as the reduction operation.
3778 (is_nonwrapping_integer_induction): Allow negative step.
3779 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3780 vect_create_epilog_for_reduction, if no value is suitable, don't
3781 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3783 2017-12-12 Richard Biener <rguenther@suse.de>
3785 PR tree-optimization/81889
3786 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3787 range info from the non-wrapping IV instead of just the range
3790 2017-12-12 Julia Koval <julia.koval@intel.com>
3792 * config.gcc: Add vaesintrin.h.
3793 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3795 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3796 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3798 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3799 * config/i386/immintrin.h: Include vaesintrin.h.
3800 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3801 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3802 _mm_aesdec_epi128): New intrinsics.
3804 2017-12-12 Julia Koval <julia.koval@intel.com>
3806 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3807 OPTION_MASK_ISA_VAES_UNSET): New.
3808 (ix86_handle_option): Handle -mvaes.
3809 * config/i386/cpuid.h: Define bit_VAES.
3810 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3811 * config/i386/i386-c.c (__VAES__): New.
3812 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3813 (ix86_valid_target_attribute_inner_p): Ditto.
3814 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3815 * config/i386/i386.opt: Add -mvaes.
3816 * doc/invoke.texi: Ditto.
3818 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3820 * debug.h (gcc_debug_hooks): Add inline_entry.
3821 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3822 * debug.c (do_nothing_debug_hooks): Likewise.
3823 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3824 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3825 (dwarf2_lineno_debug_hooks): Likewise.
3827 * common.opt (gstatement-frontiers): New, setting
3828 debug_nonbind_markers_p.
3829 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3830 * toplev.c (process_options): Autodetect value for debug statement
3832 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3833 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3835 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3836 markers. Integrate source bind into debug stmt expand loop.
3837 (pass_expand::execute): Check debug marker limit. Avoid deep
3838 TER and expand debug locations for debug bind insns only.
3839 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3841 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3842 * final.c (reemit_insn_block_notes): Take current block from
3843 nonbind markers. Declare note where it's first set.
3844 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3845 begin stmt markers if enabled.
3846 (notice_source_line): Handle nonbind markers. Fail if their
3847 location is unknown or that of builtins.
3848 (rest_of_handle_final): Convert begin stmt markers to notes if
3849 var-tracking didn't run.
3850 (rest_of_clean_state): Skip begin stmt markers.
3851 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3853 * function.c (allocate_struct_function): Set begin_stmt_markers.
3854 * function.h (struct function): Add debug_marker_count counter
3855 and debug_nonbind_markers flag.
3856 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3857 * gimple-low.c (lower_function_body): Adjust
3858 debug_nonbind_markers.
3859 (lower_stmt): Drop or skip gimple debug stmts.
3860 (lower_try_catch): Skip debug stmts.
3861 * gimple.c (gimple_build_debug_begin_stmt): New.
3862 (gimple_copy): Increment debug_marker_count if copying one.
3863 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3864 * gimplify.c (rexpr_location): New.
3865 (rexpr_has_location): New.
3866 (warn_switch_unreachable_r): Handle gimple debug stmts.
3867 (shortcut_cond_r): Call expr_location.
3869 (find_goto_label): New.
3870 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3872 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3874 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3875 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3876 (LANG_HOOKS_INITIALIZER): ... this.
3877 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3878 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3880 (update_ebb_live_info): Skip debug insn markers.
3881 * lra.c (debug_insn_static_data): Rename to...
3882 (debug_bind_static_data): ... this.
3883 (debug_marker_static_data): New.
3884 (lra_set_insn_recog_data): Select one of the above depending
3886 (lra_update_isn_regno_info): Don't assume debug insns have
3888 (push_insns): Skip debug insns.
3889 * lto-streamer-in.c (input_function): Drop debug stmts
3890 depending on active options. Adjust debug_nonbind_markers.
3891 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3892 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3893 begin stmt marker notes.
3894 (print_insn): Likewise.
3895 * recog.c (extract_insn): Recognize rtl for debug markers.
3896 * rtl.def (DEBUG_MARKER): New.
3897 * tree-inline.c: Include params.h.
3898 (remap_gimple_stmt): Handle nonbind markers.
3899 (maybe_move_debug_stmts_to_successors): Likewise.
3900 (copy_debug_stmt): Likewise.
3901 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3902 markers regardless of no side effects.
3903 (tsi_link_before): Don't update container's side effects when adding
3904 a begin stmt marker.
3905 (tsi_link_after): Likewise.
3906 (expr_first): Skip begin stmt markers.
3907 (expr_last): Likewise.
3908 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3909 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3910 Disregard nonbind markers.
3911 * tree.c (make_node_stat): Don't set side effects for begin stmt
3913 (build1_stat): Likewise.
3914 * tree.def (DEBUG_BEGIN_STMT): New.
3915 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3916 * var-tracking.c (delete_debug_insns): Renamed to...
3917 (delete_vta_debug_insns): ... this.
3918 (reemit_marker_as_note): New.
3919 (vt_initialize): Reemit markers.
3920 (delete_vta_debug_insns): Likewise.
3921 (vt_debug_insns_local): Reemit or delete markers.
3922 (variable_tracking_main_1): Likewise.
3923 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3924 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3925 (gimple_debug_nonbind_marker_p): New.
3926 (gimple_build_debug_bind): Adjust.
3927 (gimple_build_debug_begin_stmt): New.
3928 * doc/invoke.texi (max-debug-marker-count): New param.
3929 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3930 (debug_parameter_ref, debug_marker): New.
3931 (NOTE_INSN_BEGIN_STMT): New.
3932 (DEBUG_INSN): Describe begin stmt markers.
3934 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3935 without debug insns, we wouldn't, but clean up debug insns
3936 after a control flow insn nevertheless.
3938 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3939 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3940 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3941 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3942 (rtl_tidy_fallthru_edge): Likewise.
3943 (rtl_verify_fallthru): Likewise.
3944 (rtl_verify_bb_layout): Likewise.
3945 (skip_insns_after_block): Likewise.
3946 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3947 * dwarf2out.c: Include print-rtl.h.
3948 (dwarf2out_next_real_insn): New.
3949 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3950 Dump debug binds in asm comments.
3951 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3952 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3953 callers to use gsi_start_nondebug_bb instead.
3954 (gsi_after_labels): Skip gimple debug stmts.
3955 (gsi_start_nondebug): New.
3956 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3957 (proper_loop_form_for_interchange): Adjust.
3958 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3959 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3960 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3961 (collect_fallthrough_labels): Likewise.
3962 (should_warn_for_implicit_fallthrough): Likewise.
3963 (warn_implicit_fallthrough_r): Likewise.
3964 (expand_FALLTHROUGH_r): Likewise.
3965 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3966 (graphite_copy_stmts_from_block): Skip nonbind markers.
3967 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3968 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3969 * jump.c (clean_barriers): Skip debug insns.
3970 * omp-expand.c (expand_parallel_call): Skip debug insns.
3971 (expand_task_call): Likewise.
3972 (remove_exit_barrier): Likewise.
3973 (expand_omp_taskreg): Likewise.
3974 (expand_omp_for_init_counts): Likewise.
3975 (expand_omp_for_generic): Likewise.
3976 (expand_omp_for_static_nochunk): Likewise.
3977 (expand_omp_for_static_chunk): Likewise.
3978 (expand_omp_simd): Likewise.
3979 (expand_omp_taskloop_for_outer): Likewise.
3980 (expand_omp_taskloop_for_inner): Likewise.
3981 (expand_oacc_for): Likewise.
3982 (expand_omp_sections): Likewise.
3983 (expand_omp_single): Likewise.
3984 (expand_omp_synch): Likewise.
3985 (expand_omp_atomic_load): Likewise.
3986 (expand_omp_atomic_store): Likewise.
3987 (expand_omp_atomic_fetch_op): Likewise.
3988 (expand_omp_atomic_pipeline): Likewise.
3989 (expand_omp_atomic_mutex): Likewise.
3990 (expand_omp_target): Likewise.
3991 (grid_expand_omp_for_loop): Likewise.
3992 (grid_expand_target_grid_body): Likewise.
3993 (build_omp_regions_1): Likewise.
3994 * omp-low.c (check_combined_parallel): Skip debug stmts.
3995 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3996 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3997 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3999 * tree-cfg.c (make_blobs_1): Skip debug stmts.
4000 (make_edges): Likewise.
4001 (cleanup_dead_labels): Likewise.
4002 (gimple_can_merge_blocks_p): Likewise.
4003 (stmt_starts_bb_p): Likewise.
4004 (gimple_block_label): Likewise.
4005 (gimple_redirect_edge_and_branch): Likewise.
4006 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
4008 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
4010 * tree-pretty-print (print_declaration): Omit initializer in slim
4012 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
4014 (eliminate_unnecessary_stmts): Stabilize block removal order.
4015 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
4016 * var-tracking.c (get_first_insn): New.
4017 (vt_emit_notes): Call it.
4018 (vt_initialize): Walk any insns before the first BB.
4019 (delete_debug_insns): Likewise.
4021 * gimple.h (enum gimple_debug_subcode): Add
4022 GIMPLE_DEBUG_BEGIN_STMT.
4023 (gimple_debug_begin_stmt_p): New.
4024 (gimple_debug_nonbind_marker_p): New.
4025 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
4026 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
4027 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
4028 * insn-notes.def (BEGIN_STMT): New.
4029 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
4030 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
4031 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
4032 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
4033 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
4034 (INSN_DEBUG_MARKER_KIND): New.
4035 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
4036 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
4037 (INSN_VAR_LOCATION_PTR): New.
4038 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
4040 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
4041 deps only if debug bind insns are enabled.
4042 (pass_expand::execute): Avoid deep TER and expand
4043 debug locations for debug bind insns only.
4044 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
4045 debug stmts special handling down to debug bind stmts.
4046 * combine.c (try_combine): Narrow debug insns special handling
4047 down to debug bind insns.
4048 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
4049 Narrow debug insns preexisting special handling down to debug
4051 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
4052 handling down to debug bind insns.
4053 * function.c (instantiate_virtual_regs): Skip debug markers,
4054 adjust handling of debug binds.
4055 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
4056 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
4057 * haifa-sched.c (schedule_insn): Narrow special handling of debug
4058 insns to debug bind insns.
4059 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
4060 special handling of debug stmts to debug bind stmts.
4061 * ipa-split.c (split_function): Likewise.
4062 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4063 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4065 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4066 * regrename.c (build_def_use): Likewise.
4067 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4068 (pass_cprop_hardreg): Narrow special casing of debug insns to
4070 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4071 * reload1.c (reload): Likewise.
4072 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4073 casing of debug stmts to debug bind stmts.
4074 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4075 * ssa-iterators.h (num_imm_uses): Likewise.
4076 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4077 debug stmts to debug bind stmts.
4078 * tree-inline.c (tree_function_versioning): Narrow special casing
4079 of debug stmts to debug bind stmts.
4080 * tree-loop-distribution.c (generate_loops_for_partition):
4081 Narrow special casing of debug stmts to debug bind stmts.
4082 * tree-sra.c (analyze_access_subtree): Narrow special casing
4083 of debug stmts to debug bind stmts.
4084 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4085 stmts to debug bind stmts.
4086 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4087 casing of debug stmts to debug bind stmts.
4088 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4089 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4090 casing of debug stmts to debug bind stmts.
4091 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4093 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4094 debug stmts to debug bind stmts.
4095 (gimple_replace_ssa_lhs): Likewise.
4096 (insert_debug_temp_for_var_def): Likewise.
4097 (insert_debug_temps_for_defs): Likewise.
4098 (reset_debug_uses): Likewise.
4099 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4100 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4101 (adjust_debug_stmts): Likewise.
4102 (adjust_phi_and_debug_stmts): Likewise.
4103 (vect_do_peeling): Likewise.
4104 * tree-vect-loop.c (vect_transform_loop): Likewise.
4105 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4106 * var-tracking.c (adjust_mems): Narrow special casing of debug
4107 insns to debug bind insns.
4108 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4109 (compute_bb_dataflow, vt_find_locations): Likewise.
4110 (vt_expand_loc, emit_notes_for_changes): Likewise.
4111 (vt_init_cfa_base): Likewise.
4112 (vt_emit_notes): Likewise.
4113 (vt_initialize): Likewise.
4114 (vt_finalize): Likewise.
4116 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4117 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4118 (next_nonnote_nondebug_insn_bb): New.
4119 (prev_nonnote_nondebug_insn_bb): New.
4120 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4121 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4122 (prev_nonnote_nondebug_insn_bb): Declare.
4123 (next_nonnote_nondebug_insn_bb): Declare.
4124 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4125 * cfgrtl.c (get_last_bb_insn): Likewise.
4126 * lra.c (push_insns): Likewise.
4128 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4131 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4133 * selftest.c (selftest::test_runner::test_runner): New ctor.
4134 (selftest::test_runner::~test_runner): New dtor.
4135 * selftest.h (class selftest::test_runner): New class.
4137 2017-12-11 Carl Love <cel@us.ibm.com>
4139 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4140 vec_extract_fp32_from_shortl]): Add #defines.
4141 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4142 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4143 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4144 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4145 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4146 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4147 * doc/extend.texi: Add documentation for the added builtins.
4149 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4151 PR rtl-optimization/80693
4152 PR rtl-optimization/81019
4153 PR rtl-optimization/81020
4154 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4155 are not mentioned in i3. Place the REG_UNUSED note on i2,
4156 possibly modified to REG_DEAD, if it did not originate in i3.
4158 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4160 * recog.c (store_data_bypass_p_1): New function.
4161 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4162 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4164 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4166 PR rtl-optimization/83361
4167 * ifcvt.c (if_convert): Call fixup_partitions.
4169 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4171 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4172 early folding of splat_u{8,16,32}.
4174 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4176 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4177 output_operand_lossage first argument with capital letter.
4178 (aarch64_override_options): Don't start error and sorry first argument
4179 with capital letter.
4181 2017-12-11 Andi Kleen <ak@linux.intel.com>
4183 PR gcov-profile/83355
4184 * auto-profile.c (string_table::get_index_by_decl): Don't
4185 recurse when abstract origin points to itself.
4187 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4189 PR tree-optimization/83320
4190 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4191 (prune_datarefs_not_in_loop): Ditto.
4193 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4195 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4197 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4199 PR tree-optimization/83337
4200 * gimple-loop-interchange.cc (compute_access_stride): Handle
4201 bitfield DRs properly.
4203 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4205 PR tree-optimization/83338
4206 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4207 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4208 vector of unsigned integers to vector of signed integers.
4210 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4212 PR rtl-optimization/83317
4213 * lra-constraints.c (process_address_1): Add insn code check.
4215 2017-12-08 Michael Matz <matz@suse.de>
4217 Fix PR tree-optimization/83323
4218 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4219 head-controlled loops and loop BBs.
4220 * common.opt (funroll-and-jam): Remove, instead ...
4221 (floop-unroll-and-jam): ... reuse this option.
4222 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4223 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4224 (-floop-unroll-and-jam): ... this option.
4226 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4228 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4229 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4230 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4232 2017-12-08 Richard Biener <rguenther@suse.de>
4235 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4236 handle accesses outside of zero-sized vars.
4238 2017-12-08 Martin Jambor <mjambor@suse.cz>
4240 PR tree-optimization/83141
4241 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4242 test for MEM_REFs implicitely changing types with padding. Remove
4244 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4246 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4248 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4250 * config/avr/avr.c (avr_attribute_table): Likewise.
4251 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4252 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4253 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4254 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4255 of struct attribute_spec.
4256 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4258 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4261 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4263 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4266 * profile-count.c (profile_count::from_gcov_type): Move from
4267 profile-count.h; handle overflow.
4268 * profile-count.h (profile_count::from_gcov_type): Move offline.
4270 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4272 PR rtl-optimization/83304
4273 * combine.c (move_deaths): If we do not know where a register died,
4276 2017-12-08 Richard Biener <rguenther@suse.de>
4278 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4279 Provide -fopt-info-loop feedback when we interchange in a nest.
4281 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4283 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4284 for armv6 ARM CPU IDs.
4286 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4288 * common/config/arm/arm-common.c: Include <algorithm>.
4289 (INCLUDE_VECTOR): Define.
4290 (compare_opt_names): New function.
4291 (arm_rewrite_selected_arch): Only strip out extensions that can be
4292 expressed through -mfpu. Sort the remaining extensions
4295 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4297 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4298 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4299 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4300 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4302 2017-06-08 Tristan Gingold <gindold@adacore.com>
4305 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4306 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4308 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4310 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4311 of struct attribute_spec.
4313 2017-12-08 Julia Koval <julia.koval@intel.com>
4315 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4316 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4317 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4318 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4319 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4320 _mm_maskz_dpwssds_epi32): Ditto.
4322 2017-12-08 Richard Biener <rguenther@suse.de>
4324 PR tree-optimization/81303
4325 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4326 conditions try to create a comparison vector type matching
4327 the data vector type.
4328 (vectorizable_condition): Adjust.
4329 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4330 Leave invariant conditions alone in case we can vectorize those.
4332 2017-12-08 Julia Koval <julia.koval@intel.com>
4334 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4335 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4336 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4337 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4338 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4340 2017-12-08 Julia Koval <julia.koval@intel.com>
4342 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4343 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4344 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4345 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4346 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4347 _mm_maskz_dpbusds_epi32): New intrinsics.
4349 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4351 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4353 (ld<bh>io_signed): Likewise.
4354 (st<bhw>io): Likewise.
4355 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4356 SMALL_INT12 constant integer operand.
4358 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4360 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4362 (nios2_large_constant_allowed): ...this. Adjust uses.
4363 (nios2_plus_symbolic_constant_p): Rename to...
4364 (nios2_plus_large_constant_p): ...this. Adjust uses.
4365 (nios2_legitimate_address_p): Correct CONST_INT handling.
4366 (nios2_symbolic_memory_operand_p): Rename to...
4367 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4368 (nios2_large_constant_p): Check for large constant integers too.
4369 (nios2_split_large_constant): Handle constant integers.
4370 (nios2_split_symbolic_memory_operand): Rename to...
4371 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4372 (nios2_legitimize_constant_address): Handle constant integers.
4373 (r0rel_constant_p): Handle small constant integers.
4374 (nios2_print_operand_address): Handle r0-relative integer addresses.
4375 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4376 * config/nios2/nios2.md: Adjust for renamed functions.
4378 2017-12-07 Andrew Waterman <andrew@sifive.com>
4380 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4381 (TARGET_HAVE_SRODATA_SECTION): New define.
4382 (riscv_select_section): New function.
4384 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4385 Alexander Monakov <amonakov@ispras.ru>
4386 Jakub Jelinek <jakub@redhat.com>
4389 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4391 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4393 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4394 a bool success value. Don't call output_operand_lossage here.
4395 (aarch64_print_ldpstp_address): Return a bool success value.
4396 (aarch64_print_operand_address): Call output_addr_const if
4397 aarch64_print_address_internal fails.
4398 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4399 'y'; call output_operand_lossage instead. Call output_operand_lossage
4400 if aarch64_print_ldpstp_address fails.
4402 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4404 * tree-vector-builder.h
4405 (tree_vector_builder::binary_encoded_nelts): Declare.
4406 * tree-vector-builder.c
4407 (tree_vector_builder::binary_encoded_nelts): New function.
4408 * fold-const.c (negate_expr_p): Likewise.
4409 (operand_equal_p, fold_checksum_tree): Likewise.
4410 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4411 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4412 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4413 (uniform_vector_p): Likewise.
4414 * varasm.c (const_hash_1, compare_constant): Likewise.
4415 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4416 (valid_lattice_transition): Operate directly on the VECTOR_CST
4418 * ipa-icf.c: Include tree-vector-builder.h.
4419 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4420 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4422 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4424 * tree.c (build_vector): Delete.
4425 * tree.h (build_vector): Make static and move into the self-testing
4428 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4430 * vector-builder.h (vector_builder::derived): New const overload.
4431 (vector_builder::elt): New function.
4432 * tree-vector-builder.h (tree_vector_builder::type): New function.
4433 (tree_vector_builder::apply_step): Declare.
4434 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4436 * gimple-fold.h (tree_vector_builder): Declare.
4437 (gimple_build_vector): Take a tree_vector_builder instead of a
4438 type and vector of elements.
4439 * gimple-fold.c (gimple_build_vector): Likewise.
4440 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4442 (get_initial_defs_for_reduction): Likewise.
4443 (vectorizable_induction): Likewise.
4445 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4447 * tree-vector-builder.h
4448 (tree_vector_builder::new_binary_operation): Declare.
4449 * tree-vector-builder.c
4450 (tree_vector_builder::new_binary_operation): New function.
4451 * fold-const.c (fold_relational_const): Use it.
4452 (const_binop): Likewise. Check that both input vectors have
4453 the same number of elements, thus excluding things like WIDEN_SUM.
4454 Check whether it is possible to operate directly on the encodings
4457 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4459 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4460 new_unary_operation, operating only on the encoded elements.
4461 (const_unop): Likewise.
4462 (exact_inverse): Likewise.
4463 (distributes_over_addition_p): New function.
4464 (const_binop): Use tree_vector_builder and new_unary_operation
4465 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4466 on the encoded elements unless the encoding is strided and the
4467 operation does not distribute over addition.
4468 (fold_convert_const): Use tree_vector_builder and
4469 new_unary_operation. Operate only on the encoded elements
4470 for truncating integer conversions, or for non-stepped encodings.
4472 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4474 * config/sparc/sparc.c: Include tree-vector-builder.h.
4475 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4476 * expmed.c: Include tree-vector-builder.h.
4477 (make_tree): Use tree_vector_builder instead of build_vector.
4478 * fold-const.c: Include tree-vector-builder.h.
4479 (const_binop): Use tree_vector_builder instead of build_vector.
4480 (const_unop): Likewise.
4481 (native_interpret_vector): Likewise.
4482 (fold_vec_perm): Likewise.
4483 (fold_ternary_loc): Likewise.
4484 * gimple-fold.c: Include tree-vector-builder.h.
4485 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4487 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4488 (simplify_vector_constructor): Use tree_vector_builder instead
4490 * tree-vect-generic.c: Include tree-vector-builder.h.
4491 (add_rshift): Use tree_vector_builder instead of build_vector.
4492 (expand_vector_divmod): Likewise.
4493 (optimize_vector_constructor): Likewise.
4494 * tree-vect-loop.c: Include tree-vector-builder.h.
4495 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4496 of build_vector. Explicitly use a stepped encoding for
4498 * tree-vect-slp.c: Include tree-vector-builder.h.
4499 (vect_get_constant_vectors): Use tree_vector_builder instead
4501 (vect_transform_slp_perm_load): Likewise.
4502 (vect_schedule_slp_instance): Likewise.
4503 * tree-vect-stmts.c: Include tree-vector-builder.h.
4504 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4505 (vect_gen_perm_mask_any): Likewise.
4506 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4507 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4509 (build_vector_from_val): Likewise. Explicitly use a duplicate
4512 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4514 * doc/generic.texi (VECTOR_CST): Describe new representation of
4516 * vector-builder.h: New file.
4517 * tree-vector-builder.h: Likewise.
4518 * tree-vector-builder.c: Likewise.
4519 * Makefile.in (OBJS): Add tree-vector-builder.o.
4520 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4521 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4522 (tree_vector): Change the number of elements to
4523 vector_cst_encoded_nelts.
4524 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4525 (VECTOR_CST_ELTS): Delete.
4526 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4527 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4528 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4529 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4530 (VECTOR_CST_ENCODED_ELT): Likewise.
4531 (vector_cst_encoded_nelts): New function.
4532 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4533 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4534 (vector_cst_int_elt, vector_cst_elt): Declare.
4535 * tree.c: Include tree-vector-builder.h.
4536 (tree_code_size): Abort if passed VECTOR_CST.
4537 (tree_size): Update for new VECTOR_CST layout.
4538 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4539 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4540 (build_vector): Use tree_vector_builder.
4541 (vector_cst_int_elt, vector_cst_elt): New functions.
4542 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4543 encoded elements and then create the vector in the canonical form.
4544 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4545 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4546 (tree_c_tests): Call test_vector_cst_patterns.
4547 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4549 (hash_tree): Likewise.
4550 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4551 (streamer_write_tree_header): Likewise.
4552 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4553 (streamer_alloc_tree): Likewise. Update call to make_vector.
4554 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4556 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4558 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4559 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4560 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4562 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4563 Richard Biener <rguenther@suse.de>
4565 PR tree-optimization/81303
4566 * Makefile.in (gimple-loop-interchange.o): New object file.
4567 * common.opt (floop-interchange): Reuse the option from graphite.
4568 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4569 -floop-interchange and mention it for -O3.
4570 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4571 * gimple-loop-interchange.cc: New file.
4572 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4573 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4574 * passes.def (pass_linterchange): New pass.
4575 * timevar.def (TV_LINTERCHANGE): New time var.
4576 * tree-pass.h (make_pass_linterchange): New declaration.
4577 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4578 interchange. Record IV before/after increment in new parameters.
4579 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4580 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4582 (check_reduction_path): ...New function here.
4583 * tree-vectorizer.h (check_reduction_path): New declaration.
4585 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4588 PR rtl-optimization/80818
4589 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4590 always early clobbered.
4591 * lra-lives.c (process_bb_lives): Check input hard regs for early
4592 clobbered non-operand hard reg.
4594 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4597 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4598 types_compatible_p, just that TYPE_MODE is the same.
4600 2017-12-07 Martin Sebor <msebor@redhat.com>
4603 * attribs.c (empty_attribute_table): Initialize new member of
4604 struct attribute_spec.
4605 (decl_attributes): Add argument. Handle mutually exclusive
4606 combinations of attributes.
4607 (selftests::test_attribute_exclusions): New function.
4608 (selftests::attribute_c_tests): Ditto.
4609 * attribs.h (decl_attributes): Add default argument.
4610 * selftest.h (attribute_c_tests): Declare.
4611 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4612 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4614 * doc/extend.texi (Common Function Attributes): Update const and pure.
4615 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4616 of struct attribute_spec.
4617 * config/arc/arc.c (arc_attribute_table): Same.
4618 * config/arm/arm.c (arm_attribute_table): Same.
4619 * config/avr/avr.c ( avr_attribute_table): Same.
4620 * config/bfin/bfin.c (bfin_attribute_table): Same.
4621 * config/cr16/cr16.c (cr16_attribute_table): Same.
4622 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4623 * config/h8300/h8300.c (h8300_attribute_table): Same.
4624 * config/i386/i386.c (ix86_attribute_table): Same.
4625 * config/ia64/ia64.c (ia64_attribute_table): Same.
4626 * config/m32c/m32c.c (m32c_attribute_table): Same.
4627 * config/m32r/m32r.c (m32r_attribute_table): Same.
4628 * config/m68k/m68k.c (m68k_attribute_table): Same.
4629 * config/mcore/mcore.c (mcore_attribute_table): Same.
4630 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4631 * config/mips/mips.c (mips_attribute_table): Same.
4632 * config/msp430/msp430.c (msp430_attribute_table): Same.
4633 * config/nds32/nds32.c (nds32_attribute_table): Same.
4634 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4635 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4636 * config/rl78/rl78.c (rl78__attribute_table): Same.
4637 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4638 * onfig/rx/rx.c (rx_attribute_table): Same.
4639 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4640 * config/sh/sh.c (sh_attribute_table): Same.
4641 * config/sparc/sparc.c (sparc_attribute_table): Same.
4642 * config/spu/spu.c (spu_attribute_table): Same.
4643 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4644 * config/v850/v850.c (v850_attribute_table): Same.
4645 * config/visium/visium.c (visium_attribute_table): Same.
4647 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4650 * config/arm/arm.c (INCLUDE_STRING): Define.
4651 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4652 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4655 2017-12-07 Michael Matz <matz@suse.de>
4657 Add unroll and jam pass
4659 * gimple-loop-jam.c: New file.
4660 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4661 * common.opt (funroll-and-jam): New option.
4662 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4663 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4664 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4665 * passes.def: Add pass_loop_jam.
4666 * timevar.def (TV_LOOP_JAM): Add.
4667 * tree-pass.h (make_pass_loop_jam): Declare.
4668 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4669 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4670 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4671 to flow_loop_tree_node_add.
4672 (duplicate_subloops, copy_loops_to): Append to sibling list.
4673 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4674 * doc/invoke.texi (-funroll-and-jam): Document new option.
4675 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4677 2017-12-07 Richard Biener <rguenther@suse.de>
4679 PR tree-optimization/83296
4680 PR tree-optimization/67769
4681 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4682 flow sensitive info in an unrelated BB.
4683 (value_replacement): Use reset_flow_sensitive_info.
4684 (minmax_replacement): Reset flow sensitive info on the def
4685 we move. Do not reset flow sensitive info in the whole BB
4686 we move the stmt to.
4687 (abs_replacement): Likewise.
4689 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4692 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4693 rs6000_cpu to the given -mcpu=, or to the default processor.
4695 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4697 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4698 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4699 instead of rs6000_cpu_attr.
4700 (rs6000_adjust_cost): Ditto.
4701 (is_microcoded_insn): Ditto.
4702 (rs6000_adjust_priority): Ditto.
4703 (rs6000_issue_rate): Ditto.
4704 (rs6000_use_sched_lookahead): Ditto.
4705 (rs6000_use_sched_lookahead_guard): Ditto.
4706 (rs6000_sched_reorder): Ditto.
4707 (force_new_group): Ditto.
4708 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4709 (group_ending_nop): Ditto.
4711 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4713 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4714 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4715 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4716 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4718 (rs6000_reassociation_width): Ditto.
4719 (rs6000_emit_epilogue): Ditto.
4720 (rs6000_adjust_cost): Ditto.
4721 (is_microcoded_insn): Ditto.
4722 (is_cracked_insn): Ditto.
4723 (rs6000_adjust_priority): Ditto.
4724 (rs6000_sched_reorder): Ditto.
4725 (rs6000_sched_reorder2): Ditto.
4726 (insn_must_be_first_in_group): Ditto.
4727 (insn_must_be_last_in_group): Ditto.
4728 (rs6000_register_move_cost): Ditto.
4729 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4732 2017-12-07 Julia Koval <julia.koval@intel.com>
4734 * config.gcc: Add vaesintrin.h.
4735 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4737 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4738 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4740 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4741 * config/i386/immintrin.h: Include vaesintrin.h.
4742 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4743 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4744 _mm_aesdec_epi128): New intrinsics.
4746 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4748 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4749 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4750 c-family/c-spellcheck.cc.
4751 (best_macro_match::best_macro_match): Likewise.
4753 (struct edit_distance_traits<cpp_hashnode *>): Move to
4754 c-family/c-spellcheck.h.
4755 (class best_macro_match): Likewise.
4757 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4759 PR tree-optimization/83293
4760 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4761 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4762 might insert into empty bb.
4765 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4766 simplify for plus with :c added, and pointer_plus without that.
4767 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4768 with undefined overflow and the conversion is not widening,
4769 perform negation in utype and only convert to type afterwards.
4770 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4771 simplify for plus with :c added, and pointer_plus without that.
4772 If type is integral with undefined overflow and the conversion is
4773 not widening, perform minus in utype and only convert to type
4774 afterwards. Move the last pointer_diff_expr simplify into the
4777 2017-12-06 Martin Sebor <msebor@redhat.com>
4779 PR tree-optimization/82646
4780 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4781 strncpy, not maxlen.
4783 2017-12-06 Martin Sebor <msebor@redhat.com>
4785 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4788 PR tree-optimization/83075
4789 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4790 strncat/strncpy don't change length of source string.
4792 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4795 2017-11-29 Martin Aberg <maberg@gaisler.com>
4797 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4798 to prevent b2bst errata sequence.
4799 (sqrtdf2_fix): Likewise.
4801 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4803 PR tree-optimization/81945
4804 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4805 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4806 to dest_cfun has orig_loop_num set, either remap it to the new
4807 loop number if the loop got moved too, or clear it.
4809 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4811 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4813 (thunderx2t99_nothing): New insn reservation.
4814 (thunderx2t99_mrs): New insn reservation.
4815 (thunderx2t99_multiple): New insn reservation.
4816 (thunderx2t99_alu_basi): Add bfx to reservation.
4817 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4819 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4822 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4824 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4826 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4827 tree-ssa-pre.c::remove_dead_inserted_code.
4828 * tree-ssa-dce.h: New file.
4829 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4830 (remove_dead_inserted_code): Move and rename to function
4831 tree-ssa-dce.c::simple_dce_from_worklist.
4832 (pass_pre::execute): Update use.
4834 2017-12-05 Richard Biener <rguenther@suse.de>
4836 PR tree-optimization/83277
4837 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4838 to code-gen liveout vars.
4840 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4842 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4843 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4844 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4847 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4849 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4851 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4852 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4855 2017-12-05 Richard Biener <rguenther@suse.de>
4857 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4858 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4859 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4860 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4862 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4864 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4866 2017-12-05 Martin Liska <mliska@suse.cz>
4867 Jakub Jelinek <jakub@redhat.com>
4869 * doc/invoke.texi: Document the options.
4870 * flag-types.h (enum sanitize_code): Add
4871 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4872 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4873 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4874 * opts.c: Define new sanitizer options.
4875 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4876 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4878 2017-12-05 Julia Koval <julia.koval@intel.com>
4880 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4881 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4882 (ix86_handle_option): Handle -mavx512vnni.
4883 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4884 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4885 * config/i386/i386-c (__AVX512VNNI__): New.
4886 * config/i386/i386.c (ix86_target_string): Handle new option.
4887 (ix86_valid_target_attribute_inner_p): Handle new option.
4888 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4889 * config/i386/i386.opt (mavx512vnni): New option.
4891 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4894 * athlon.md: Disable for generic.
4895 * haswell.md: Enable for generic.
4896 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4897 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4899 (ix86_adjust_cost): Move generic to haswell path.
4901 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4903 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4904 instead of MEM_P in a couple more places. Fix formatting issues.
4906 2017-12-04 Jim Wilson <jimw@sifive.com>
4908 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4909 instead of GP_REG_LAST-1.
4910 (riscv_adjust_libcall_cfi_prologue): Likewise.
4911 (riscv_adjust_libcall_cri_epilogue): Likewise.
4912 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4915 2017-12-04 Luis Machado <luis.machado@linaro.org>
4917 * ipa-pure-const.c (check_decl): Add missing newline.
4918 (state_from_flags): Likewise.
4920 2017-12-04 Jeff Law <law@redhat.com>
4922 PR tree-optimizatin/78496
4923 * gimple-ssa-evrp-analyze.h
4924 (evrp_range_analyzer::get_vr_values): Simplify.
4925 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4926 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4927 and gimple-ssa-evrp-analyze.h.
4928 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4929 (simplify_stmt_for_jump_threading): Copy a blob of code from
4930 tree-vrp.c to use ranges to simplify statements.
4931 (dom_opt_dom_walker::before_dom_children): Call
4932 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4933 (dom_opt_dom_walker::after_dom_children): Similarly for
4934 evrp_range_analyzer::leave.
4935 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4938 * gimple-ssa-evrp-analyze.c
4939 (evrp_range_analyzer::extract_range_from_stmt): Always use
4940 vr_values::update_value_range so preexisting range info is
4941 medged with new range info, even if the new range is VR_VARYING.
4943 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4945 * combine.c: Adjust comment.
4946 (use_crosses_set_p): Delete.
4947 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4948 (try_combine): Ditto.
4950 2017-12-04 Richard Biener <rguenther@suse.de>
4952 PR tree-optimization/83255
4953 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4954 Re-add zero-iteration check.
4956 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4958 PR rtl-optimization/83245
4959 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4960 hard registers as earlyclobber, also if not in an asm.
4962 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4966 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4969 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4970 rs6000_cpu based on cpu_index, not tune_index.
4972 2017-12-04 Richard Biener <rguenther@suse.de>
4974 PR tree-optimization/83238
4975 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4976 code match comment, rejecting invalid SESE regions.
4978 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4980 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4981 require base operand is a REG_POINTER prior to reload on targets
4982 with non-equivalent space registers.
4984 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4986 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4987 (update_specialized_profile): Fix updating of counts.
4988 (perhaps_add_new_callers): Likewise.
4990 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4993 * x86-tune.def: Remove obsolette FIXMEs.
4994 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4995 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4996 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4998 (X86_TUNE_PAD_RETURNS): Disable for generic.
5000 2017-12-02 Jakub Jelinek <jakub@redhat.com>
5002 PR tree-optimization/83170
5003 PR tree-optimization/83241
5004 * gimple-ssa-store-merging.c
5005 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
5006 gimple_vuse (ins_stmt) in case it has changed.
5007 (imm_store_chain_info::output_merged_store): Likewise.
5009 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
5013 * tree-cfg.c (pass_warn_function_return::execute): Handle
5014 __builtin_ubsan_handle_missing_return like __builtin_unreachable
5015 with BUILTINS_LOCATION.
5019 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
5020 is BLKmode for vector field with vector raw mode, use TYPE_MODE
5021 instead of DECL_MODE.
5023 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
5024 last argument to rtx pointer.
5025 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
5026 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
5027 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
5028 depending on the chosen ISAs.
5029 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
5030 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
5031 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
5033 * config/i386/sse.md (mov<mode>_internal): Likewise.
5034 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
5036 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5038 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
5040 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5042 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
5043 parameters from prototype.
5044 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
5045 parameters. Don't print a comment.
5046 (emit_fusion_gpr_load): Adjust.
5047 (emit_fusion_load_store): Adjust.
5048 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
5049 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
5050 comment on the second line.
5052 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5055 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5056 rs6000_cpu based on cpu_index, not tune_index.
5058 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5060 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5061 which_alternative instead of which_alternative + 1.
5062 (output_asm_insn): Print an extra tab if the template is short.
5064 2017-12-01 Jim Wilson <jimw@sifive.com>
5066 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5068 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5069 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5071 * doc/tm.texi: Regenerate.
5073 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5076 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5077 whether we can allocate pseudos before trying to fix an address.
5078 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5079 memory address is indexed or indirect.
5080 (floatuns_<mode>si2_hw2): Likewise.
5082 2017-12-01 Jason Merrill <jason@redhat.com>
5084 * Makefile.in (TAGS): Add c-family/*.cc.
5086 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5088 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5089 (call_value_insn): Likewise.
5090 (sibcall_insn): Likewise.
5091 (sibcall_value_insn): Likewise.
5092 (movsi_aarch64): Likewise.
5093 (movdi_aarch64): Likewise.
5094 (add_losym_): Likewise.
5095 (ldr_got_small_): Likewise.
5096 (ldr_got_small_sidi): Likewise.
5097 (ldr_got_small_28k_): Likewise.
5098 (ldr_got_small_28k_sidi): Likewise.
5099 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5100 Move output_addr_const to symbolic case. Add error check.
5102 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5104 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5105 * config/s390/s390.md: Disable MVC merging peephole if it would
5106 disable operand forwarding.
5107 (new peephole2): Split MVCs if it would turn them into up to 2
5110 2017-12-01 Richard Biener <rguenther@suse.de>
5112 PR tree-optimization/83232
5113 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5114 detection of same access. Instead of breaking the group here
5115 do not consider the duplicate. Add comment explaining real fix.
5117 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5119 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5121 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5123 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5125 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5127 * function.h (struct function): Remove cilk_frame_decl,
5128 is_cilk_function and calls_cilk_spawn fields.
5129 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5131 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5132 cilk_elemental field.
5133 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5134 * target.def: Adjust comment.
5135 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5136 Don't test cilk_elemental.
5138 PR tree-optimization/83233
5139 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5140 bswap_stat name for the struct.
5143 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5144 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5145 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5146 SWITCH_BREAK_LABEL_P set on the label.
5147 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5148 added for default case if it was missing and not all cases covered.
5149 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5150 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5152 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5153 to avoid -Wimplicit-fallthrough warning.
5154 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5155 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5157 PR tree-optimization/83221
5158 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5160 (init_reassoc): Formatting fix.
5163 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5164 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5168 * tree-cfg.c: Include asan.h.
5169 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5171 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5173 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5175 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5177 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5178 vpcmpeqd instruction.
5180 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5182 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5183 registers type for 128bit mode.
5185 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5187 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5189 * gengtype-state.c (read_state_pair): Likewise.
5190 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5191 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5192 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5193 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5194 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5195 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5196 * builtins.c (fold_builtin_3): Likewise.
5197 * graphite-scop-detection.c
5198 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5199 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5202 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5205 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5206 parameter. Return true if ignore_decl_rtl_set_p is true and other
5207 conditions are satisfied.
5208 * asan.h (asan_protect_global): Add new parameter.
5209 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5210 to asan_protect_global calls.
5212 2017-11-30 Jim Wilson <jimw@sifive.com>
5214 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5215 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5216 -mstrict-align, add info on default value. Delete redundant lines for
5217 -mabi. Add missing -mexplicit-relocs docs.
5219 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5221 * config/arc/arc.md (trap): New pattern.
5223 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5225 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5226 instruction to end into a delay slot.
5227 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5228 can be placed into a delay slot against reg_note.
5230 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5232 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5233 labels number of usages.
5235 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5237 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5239 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5241 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5244 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5245 multiplication by power of 2 constant into two shifts + comparison.
5247 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5250 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5252 2017-11-30 Richard Biener <rguenther@suse.de>
5254 PR tree-optimization/83202
5255 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5256 allow_peel argument and guard peeling.
5257 (canonicalize_loop_induction_variables): Likewise.
5258 (canonicalize_induction_variables): Pass false.
5259 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5260 peeling from cunrolli.
5262 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5264 * combine.c (try_combine): Print a message to dump file whenever
5265 I0, I1, or I2 cannot be combined into I3.
5267 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5269 PR rtl-optimization/83156
5270 PR rtl-optimization/82621
5271 * combine.c (try_combine): Don't split an I2 if one of the dests is
5272 set again before I3. Allow unused dests.
5274 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5276 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5278 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5280 PR rtl-optimization/80818
5281 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5282 recursively. Use insn code for clobber.
5283 (lra_set_insn_recog_data): Pass the new arg to
5284 collect_non_operand_hard_regs.
5285 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5287 (lra_update_insn_regno_info): Pass insn to
5288 add_regs_to_insn_regno_info.
5290 2017-11-29 Jim Wilson <jimw@sifive.com>
5291 Andrew Waterman <andrew@sifive.com>
5293 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5294 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5295 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5296 use SINGLE_SHIFT_COST.
5297 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5298 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5300 2017-11-29 Julia Koval <julia.koval@intel.com>
5302 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5303 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5304 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5305 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5306 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5307 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5308 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5309 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5310 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5311 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5312 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5313 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5314 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5315 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5316 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5317 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5318 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5319 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5320 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5321 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5322 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5323 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5324 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5325 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5326 __builtin_ia32_vpshldv_v2di_mask,
5327 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5328 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5329 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5331 2017-11-29 Julia Koval <julia.koval@intel.com>
5333 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5334 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5335 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5336 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5337 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5338 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5339 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5340 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5341 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5342 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5343 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5344 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5345 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5346 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5347 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5348 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5349 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5350 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5351 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5352 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5353 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5355 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5357 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5358 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5361 2017-11-29 Martin Aberg <maberg@gaisler.com>
5363 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5364 to prevent b2bst errata sequence.
5365 (sqrtdf2_fix): Likewise.
5367 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5369 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5370 (div_sqrt_insn_p): New function.
5371 (sparc_do_work_around_errata): Insert NOP instructions to
5372 prevent sequences that could trigger the TN-0013 errata for
5373 certain LEON3 processors.
5374 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5375 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5376 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5377 (in_branch_delay): Prevent div and sqrt in delay slot if
5379 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5381 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5383 * config/sparc/sparc.c (atomic_insn_p): New function.
5384 (sparc_do_work_around_errata): Insert NOP instructions to
5385 prevent sequences that could trigger the TN-0010 errata for
5387 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5388 instruction referable in atomic_insns_p.
5390 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5392 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5393 (atomic_compare_and_swap_leon3_1): Likewise.
5396 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5398 * config/sparc/sparc.c (fpop_insn_p): New function.
5399 (sparc_do_work_around_errata): Insert NOP instructions to
5400 prevent sequences that could trigger the TN-0012 errata for
5402 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5403 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5404 (in_branch_annul_delay): Prevent floating-point instructions
5405 in delay slot of annulled integer branch.
5407 2017-11-29 Richard Biener <rguenther@suse.de>
5409 PR tree-optimization/83202
5410 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5412 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5413 nodes vectorized to the same stmts multiple times.
5414 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5415 (vect_analyze_slp_instance): Adjust.
5416 (scalar_stmts_to_slp_tree_map_t): New typedef.
5417 (vect_schedule_slp_instance): Add a map recording the SLP node
5418 representing the vectorized stmts for a set of scalar stmts.
5419 Avoid code-generating redundancies.
5420 (vect_schedule_slp): Allocate map and pass it down.
5422 2017-11-29 Nathan Sidwell <nathan@acm.org>
5425 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5428 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5430 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5431 cases for printing LDP/STP memory addresses.
5432 (aarch64_print_address_internal): Renamed from
5433 aarch64_print_operand_address, added parameter, add Pmode check.
5434 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5435 (aarch64_print_operand_address): Indirect to
5436 aarch64_print_address_internal.
5437 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5438 'y' operand output specifier.
5440 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5443 * tree.c (build_simple_mem_ref_loc): Handle
5444 get_addr_base_and_unit_offset returning a MEM_REF.
5447 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5450 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5453 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5455 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5457 (layout::move_to_column): Likewise.
5459 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5462 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5463 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5465 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5466 Martin Jambor <mjambor@suse.cz>
5469 * tree.h (expr_type_first_operand_type_p): Declare
5470 * tree.c (expr_type_first_operand_type_p): New function.
5471 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5472 (ipa_value_from_jfunc): Adjust declaration.
5473 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5474 Use it as result type for arithmetics, unless it is NULL in which case
5475 be more conservative.
5476 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5477 ipa_get_jf_pass_through_result.
5478 (propagate_vals_across_pass_through): Likewise.
5479 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5480 is to propagate_vals_across_pass_through.
5481 (propagate_constants_across_call): Pass PARM_TYPE to
5482 propagate_scalar_across_jump_function.
5483 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5484 ipa_value_from_jfunc.
5485 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5486 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5487 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5488 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5489 target_type, pass it to ipa_value_from_jfunc.
5490 (update_indirect_edges_after_inlining): Pass parameter type to
5491 try_make_edge_direct_simple_call.
5493 2017-11-28 Jeff Law <law@redhat.com>
5495 * gimple-ssa-evrp-analyze.c
5496 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5497 refine ranges if scev_initialized_p returns true.
5498 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5500 2017-11-28 Julia Koval <julia.koval@intel.com>
5502 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5503 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5504 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5505 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5506 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5507 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5508 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5509 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5510 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5511 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5512 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5513 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5514 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5515 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5516 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5517 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5518 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5519 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5520 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5521 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5522 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5523 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5525 2017-11-28 Julia Koval <julia.koval@intel.com>
5527 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5528 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5529 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5530 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5531 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5532 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5533 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5534 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5535 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5536 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5537 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5538 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5539 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5540 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5541 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5542 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5543 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5544 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5545 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5546 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5547 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5548 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5549 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5550 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5551 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5552 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5553 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5554 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5555 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5556 __builtin_ia32_vpshld_v2di_mask): New builtins.
5557 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5558 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5560 2017-11-28 Richard Biener <rguenther@suse.de>
5562 PR tree-optimization/80776
5563 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5565 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5567 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5568 If the incoming edge is an effective fallthru because the other
5569 edge only reaches a __builtin_unreachable () then record ranges
5570 derived from the controlling condition in SSA info.
5571 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5572 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5574 2017-11-28 Olivier Hainque <hainque@adacore.com>
5576 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5577 on mingw build hosts.
5579 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5582 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5584 2017-11-28 Richard Biener <rguenther@suse.de>
5587 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5588 copies generated from memcpy use a character array as reference
5591 2017-11-28 Julia Koval <julia.koval@intel.com>
5592 Sebastian Peryt <sebastian.peryt@intel.com>
5594 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5595 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5596 cilk-common.o, cilk.h, cilk-common.c): Remove.
5598 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5599 * builtins.c (is_builtin_name): Remove cilkplus condition.
5600 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5601 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5602 cilk-builtins.def, cilkplus.def): Remove.
5603 * cif-code.def (CILK_SPAWN): Remove.
5604 * cilk-builtins.def: Delete.
5605 * cilk-common.c: Ditto.
5607 * cilkplus.def: Ditto.
5608 * config/darwin.h (fcilkplus): Delete.
5609 * cppbuiltin.c: Ditto.
5610 * doc/extend.texi: Remove cilkplus doc.
5611 * doc/generic.texi: Ditto.
5612 * doc/invoke.texi: Ditto.
5613 * doc/passes.texi: Ditto.
5614 * gcc.c (fcilkplus): Remove.
5615 * gengtype.c (cilk.h): Remove.
5616 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5618 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5620 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5622 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5623 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5624 cilkplus conditions.
5625 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5626 inline_read_section): Ditto.
5627 * ipa-inline-analysis.c (cilk.h): Remove.
5628 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5629 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5630 append_linker_options): Remove condition for fcilkplus.
5631 * lto/lto-lang.c (cilk.h): Remove.
5632 (lto_init): Remove condition for fcilkplus.
5633 * omp-expand.c (expand_cilk_for_call): Delete.
5634 (expand_omp_taskreg, expand_omp_for_static_chunk,
5635 expand_omp_for): Remove cilkplus
5637 (expand_cilk_for): Delete.
5638 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5639 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5640 execute_lower_omp, diagnose_sb_0): Ditto.
5641 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5642 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5643 * tree-nested.c: Ditto.
5644 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5645 (dump_generic_node): Ditto.
5646 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5647 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5649 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5651 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5653 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5654 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5655 for vector mode and !TARGET_SIMD.
5657 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5659 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5661 * tree.h (SWITCH_LABELS): Remove.
5662 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5663 assert SWITCH_BODY is non-NULL.
5664 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5666 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5668 PR tree-optimization/80788
5669 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5670 has TREE_OVERFLOW set, call drop_tree_overflow.
5672 2017-11-28 Richard Biener <rguenther@suse.de>
5674 PR tree-optimization/83158
5675 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5677 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5680 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5681 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5683 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5685 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5686 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5688 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5691 * varasm.c (compare_constant): Take the mode of the constants into
5692 account when comparing floating point constants.
5694 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5696 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5697 from explicit instantiation of debug_helper.
5698 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5700 2017-11-27 Richard Biener <rguenther@suse.de>
5702 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5705 2017-11-27 Richard Biener <rguenther@suse.de>
5707 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5708 (type_hash_canon): Also clear int_cst_hash_table entry for
5710 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5712 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5714 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5715 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5716 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5718 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5720 * hash-map.h (gt_cleare_cache): Avoid UB.
5722 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5724 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5725 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5726 (decide_unroll_runtime_iterations): Likewise.
5727 (decide_unroll_stupid): Likewise.
5729 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5732 * config/i386/i386.md: Add a loop with incssp.
5734 2017-11-27 Martin Jambor <mjambor@suse.cz>
5736 PR tree-optimization/81248
5737 * tree-sra.c (splice_param_accesses): Remove size check.
5738 (decide_one_param_reduction): Fix size check.
5739 * gimple-pretty-print.c (dump_profile): Silence warning.
5740 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5742 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5745 * dbxout.c (lastlineno): New variable.
5746 (dbx_debug_hooks): Use dbxout_switch_text_section as
5747 switch_text_section debug hook.
5748 (dbxout_function_end): Switch to current_function_section
5749 rather than function_section. If crtl->has_bb_partition,
5750 output just one N_FUN, depending on in_cold_section_p.
5751 (dbxout_source_line): Remember last lineno in lastlineno.
5752 (dbxout_switch_text_section): New function.
5753 (dbxout_function_decl): Adjust dbxout_block caller.
5754 (dbx_block_with_cold_children): New function.
5755 (dbxout_block): Return true if any LBRAC/RBRAC have been
5756 emitted. Use dbx_block_with_cold_children at depth == 0
5757 in second partition. Add PARENT_BLOCKNUM argument, pass
5758 it optionally adjusted to children. Output LBRAC/RBRAC
5759 around recursive call only if the block is in the current
5760 partition, if not and anything was output, emit empty
5762 * final.c (final_scan_insn): Compute cold_function_name
5763 before calling switch_text_section debug hook. Call
5764 that hook even if dwarf2out_do_frame if not emitting
5768 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5769 TREE_READONLY decls.
5771 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5773 PR rtl-optimization/82488
5774 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5776 2017-11-26 Julia Koval <julia.koval@intel.com>
5778 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5780 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5783 2017-11-26 Julia Koval <julia.koval@intel.com>
5785 * config/i386/driver-i386.c (host_detect_local_cpu):
5786 Detect skylake-avx512.
5788 2017-11-26 Julia Koval <julia.koval@intel.com>
5790 * config.gcc: Add -march=cannonlake.
5791 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5792 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5793 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5794 (PTA_CANNONLAKE): New.
5795 (processor_target_table): Add cannonlake.
5796 (ix86_option_override_internal): Ditto.
5797 (fold_builtin_cpu): Ditto.
5798 (get_builtin_code_for_version): Handle cannonlake.
5799 (M_INTEL_COREI7_CANNONLAKE): New.
5800 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5801 * doc/invoke.texi: Add -march=cannonlake.
5803 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5805 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5806 (try_init_one_plugin): Alternative implementation for MinGW.
5807 * Makefile.in (plugin_implib): New.
5808 (gengtype-lex.c): Fix broken AIX workaround.
5809 * configure: Regenerate.
5810 * doc/plugins.texi: Document support for MinGW.
5812 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5814 PR rtl-optimization/81553
5815 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5816 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5817 is a shift where C1 has different mode than the whole shift, use C1's
5818 mode for MULT rather than the shift's mode.
5821 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5822 builtins not enabled in the currently selected ISA.
5824 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5826 PR tree-optimization/71026
5827 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5828 (insert_reciprocals): Change to insert reciprocals before a division
5829 by a square and to insert the square of a reciprocal.
5830 (execute_cse_reciprocals_1): Change to consider division by a square.
5831 (register_division_in): Add importance parameter.
5833 2017-11-24 Richard Biener <rguenther@suse.de>
5835 PR tree-optimization/82402
5836 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5837 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5839 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5841 * match.pd (0-ptr): New transformation.
5843 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5846 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5847 when skipping checking.
5849 2017-11-24 Richard Biener <rguenther@suse.de>
5851 PR tree-optimization/83128
5852 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5853 (vn_reference_lookup_3): Likewise.
5855 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5858 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5859 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5862 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5863 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5864 check for the latter with gimple_call_builtin_p. Do not handle
5865 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5867 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5869 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5872 2017-11-23 Julia Koval <julia.koval@intel.com>
5874 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5875 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5876 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5877 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5878 _mm512_maskz_expandloadu_epi16): New intrinsics.
5879 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5880 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5881 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5882 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5883 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5884 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5885 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5886 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5887 _mm256_maskz_expandloadu_epi8): New intrinsics.
5888 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5889 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5890 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5891 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5892 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5893 * config/i386/sse.md (VI248_VLBW): New iterator.
5894 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5896 2017-11-23 Julia Koval <julia.koval@intel.com>
5898 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5899 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5900 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5901 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5902 _mm512_mask_compressstoreu_epi16): New.
5903 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5904 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5905 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5906 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5907 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5908 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5909 _mm256_mask_compressstoreu_epi8): New.
5910 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5911 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5912 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5913 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5914 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5915 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5916 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5917 __builtin_ia32_compresshi128_mask,
5918 __builtin_ia32_compressstoreuqi512_mask,
5919 __builtin_ia32_compressstoreuhi512_mask,
5920 __builtin_ia32_compressstoreuqi256_mask,
5921 __builtin_ia32_compressstoreuqi128_mask,
5922 __builtin_ia32_compressstoreuhi256_mask,
5923 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5924 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5926 (ix86_expand_special_args_builtin): Handle new types.
5927 (s4fma_expand): Handle new builtin array.
5928 * config/i386/immintrin.h: Include new headers.
5929 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5930 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5932 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5934 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5935 fixes. Declare temp and g variables at the top in order to avoid
5936 {} in most of the cases.
5938 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5940 * match.pd (ptr-0): New transformation.
5942 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5944 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5945 (struct addr_mode_cost_table): New.
5946 (struct tune_params): Add field addr_mode_costs.
5947 * config/arm/arm.c (generic_addr_mode_costs): New.
5948 (arm_slowmul_tune): Initialise addr_mode_costs field.
5949 (arm_fastmul_tune): Likewise.
5950 (arm_strongarm_tune): Likewise.
5951 (arm_xscale_tune): Likewise.
5952 (arm_9e_tune): Likewise.
5953 (arm_marvell_pj4_tune): Likewise.
5954 (arm_v6t2_tune): Likewise.
5955 (arm_cortex_tune): Likewise.
5956 (arm_cortex_a8_tune): Likewise.
5957 (arm_cortex_a7_tune): Likewise.
5958 (arm_cortex_a15_tune): Likewise.
5959 (arm_cortex_a35_tune): Likewise.
5960 (arm_cortex_a53_tune): Likewise.
5961 (arm_cortex_a57_tune): Likewise.
5962 (arm_exynosm1_tune): Likewise.
5963 (arm_xgene1_tune): Likewise.
5964 (arm_cortex_a5_tune): Likewise.
5965 (arm_cortex_a9_tune): Likewise.
5966 (arm_cortex_a12_tune): Likewise.
5967 (arm_cortex_a73_tune): Likewise.
5968 (arm_v7m_tune): Likewise.
5969 (arm_cortex_m7_tune): Likewise.
5970 (arm_v6m_tune): Likewise.
5971 (arm_fa726te_tune): Likewise.
5972 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5974 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5976 * config/arm/arm.c (arm_mem_costs): New function.
5977 (arm_rtx_costs_internal): Use arm_mem_costs.
5979 2017-11-23 Mark Wielaard <mark@klomp.org>
5981 * dwarf2out.c (init_sections_and_labels): Use generation to create
5982 unique ranges_section_label and ranges_base_label. Return generation.
5983 (output_rnglists): Add generation argument. Use generation to create
5984 unique ranges labels.
5985 (dwarf2out_finish): Get generation from init_sections_and_labels
5986 and pass generation to output_rnglists.
5988 2017-11-23 Mike Stump <mikestump@comcast.net>
5989 Eric Botcazou <ebotcazou@adacore.com>
5991 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5992 * cfgloop.h (struct loop): Add unroll field.
5993 * function.h (struct function): Add has_unroll bitfield.
5994 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5995 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5996 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5997 (pass_rtl_unroll_loops::gate): Likewise.
5998 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5999 for which loop->unroll==1.
6000 (decide_unroll_constant_iterations): Use note for consistency and
6001 take loop->unroll into account. Return early if loop->unroll is set.
6002 Fix thinko in existing test.
6003 (decide_unroll_runtime_iterations): Use note for consistency and
6004 take loop->unroll into account.
6005 (decide_unroll_stupid): Likewise.
6006 * lto-streamer-in.c (input_cfg): Read loop->unroll.
6007 * lto-streamer-out.c (output_cfg): Write loop->unroll.
6008 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
6010 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
6011 (print_loop): Print loop->unroll if set.
6012 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
6013 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
6014 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
6016 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
6017 loop->unroll is set and smaller than the trip count. Otherwise bypass
6018 entirely the heuristics if loop->unroll is set. Remove dead note.
6019 Fix off-by-one bug in other note.
6020 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
6021 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
6023 (tree_unroll_loops_completely): Make static.
6024 (pass_complete_unroll::execute): Use correct type for variable.
6025 (pass_complete_unrolli::execute): Fix formatting.
6026 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
6028 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
6030 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
6031 enable when TARGET_PREFER_AVX128 is set.
6033 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6035 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
6037 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
6038 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
6040 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6042 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
6043 frequencies. Fix estimation of aggregate parameters.
6045 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6047 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
6050 2017-11-23 Tom de Vries <tom@codesourcery.com>
6052 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
6054 2017-11-23 Tom de Vries <tom@codesourcery.com>
6056 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
6059 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6062 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6063 sibcall_value_pcrel_fdpic): Use local variable instead of
6065 (calli_tbr_rel): Add missing operand 2.
6066 (call_valuei_tbr_rel): Add missing operand 3.
6068 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6071 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6072 bitpos/bitsize covering the whole destination, use store_expr only if
6073 the complex mode is the same. Otherwise, use expand_normal and if
6074 it returns CONCAT, subreg each part separately instead of trying to
6075 subreg the whole result.
6077 2017-11-23 Richard Biener <rguenther@suse.de>
6079 PR tree-optimization/23094
6080 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6081 come here from walking over backedges in the first iteration.
6082 (vn_reference_lookup_3): Skip clobbers that store the same value.
6084 2017-11-23 Richard Biener <rguenther@suse.de>
6086 PR tree-optimization/81403
6087 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6088 a block we need a leader relative to.
6089 (phi_translate_1): For nary processing require a leader from
6090 get_representative_for given we run expression simplification
6091 using match-and-simplify. Remove previous fix.
6093 2017-11-22 Jeff Law <law@redhat.com>
6095 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6096 Use new method allocate_value_range rather than accessing the
6097 vrp_value_range_pool data member directly.
6098 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6099 to use extract_range_from_stmt method to avoid need for
6100 extract_range_from_assignment method.
6101 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6102 method rather than setting values_propgated data member directly.
6103 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6104 and values propagated data members and extract_range_from_assignment
6105 method. Reorder private data members to conform to standards.
6106 Add new methods set_lattice_propagation_complete and
6107 allocate_value_range.
6109 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6111 PR rtl-optimization/83030
6112 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6113 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6114 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6115 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6116 CROSSING_JUMP_P flag is already set before setting it.
6117 * cfgrtl.c (fixup_partition_crossing): Likewise.
6118 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6121 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6123 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6124 Handle the case where both arguments are using gen_const_vec_series.
6126 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6129 * pretty-print.c (pp_format): Move quoting implementation to
6130 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6131 to pass address of "quote" local.
6132 (pp_begin_quote): New function.
6133 (pp_end_quote): New function.
6134 * pretty-print.h (printer_fn): Convert penultimate param from bool
6136 (pp_begin_quote): New decl.
6137 (pp_end_quote): New decl.
6138 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6139 param from bool to bool *.
6140 * tree-diagnostic.h (default_tree_printer): Likewise.
6142 2017-11-22 Jeff Law <law@redhat.com>
6144 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6145 Initialize vr_values.
6146 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6147 vr_values via vr_values class instance rather than delegators.
6148 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6149 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6150 (evrp_range_analyzer::push_value_range): Likewise.
6151 (evrp_range_analyzer::pop_value_range): Likewise.
6152 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6153 most delegators. Those remaining are exposed as public interfaces.
6154 Make vr_values a pointer and private.
6155 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6157 (evrp_range_analyzer::get_vr_value): New method.
6158 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6159 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6160 (class evrp_dom_walker): Attach evrp_folder class, initialize
6161 it in the ctor. Remove temporary delegators.
6162 (evrp_dom_walker::before_dom_children): Call methods in attached
6163 evrp_range_analyzer class via class instance pointer. Use
6164 free value_range_constant_singleton to remove need for
6165 op_with_constant_singleton_value delegator method. Do not
6166 create a vrp_prop class instance for every call! Narrow
6167 scope of a couple variables.
6168 (evrp_dom_walker::cleanup): Call methods in attached
6169 evrp_range_analyzer class via class instance pointer.
6170 * vr-values.h (class vr_values): Privatize many methods and
6173 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6175 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6176 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6177 * expr.c (expand_expr_real_2): Likewise.
6178 * fold-const.c (const_unop): Likewise.
6179 * optabs-tree.c (optab_for_tree_code): Likewise.
6180 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6181 * tree-inline.c (estimate_operator_cost): Likewise.
6182 * tree-pretty-print.c (dump_generic_node): Likewise.
6183 (op_code_prio): Likewise.
6184 (op_symbol_code): Likewise.
6185 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6186 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6187 * internal-fn.c (direct_internal_fn_optab): New function.
6188 (direct_internal_fn_array, direct_internal_fn_supported_p
6189 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6190 * fold-const-call.c (fold_const_reduction): New function.
6191 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6193 * tree-vect-loop.c: Include internal-fn.h.
6194 (reduction_code_for_scalar_code): Rename to...
6195 (reduction_fn_for_scalar_code): ...this and return an internal
6197 (vect_model_reduction_cost): Take an internal_fn rather than
6199 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6201 (vectorizable_reduction): Use internal functions rather than tree
6202 codes for the reduction operation. Update calls to the functions
6204 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6205 Use calls to internal functions rather than REDUC tree codes.
6206 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6208 2017-11-22 Olivier Hainque <hainque@adacore.com>
6210 * config/vxworks.c (vxworks_override_options): Pick default
6211 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6212 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6213 DWARF_GNAT_ENCODINGS_DEFAULT.
6214 * config/vxworksae.h: Likewise.
6216 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6218 PR tree-optimization/83104
6219 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6222 2017-11-22 Marek Polacek <polacek@redhat.com>
6223 H.J. Lu <hongjiu.lu@intel.com>
6224 Jason Merrill <jason@redhat.com>
6229 * calls.c (initialize_argument_information): Call
6230 warn_parameter_passing_abi target hook.
6231 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6232 argument onto stack.
6233 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6234 * common.opt: Update -fabi-version description.
6235 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6236 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6238 (ix86_is_empty_record): New function.
6239 (ix86_warn_parameter_passing_abi): New function.
6240 (TARGET_EMPTY_RECORD_P): Redefine.
6241 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6242 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6243 * doc/tm.texi: Regenerated.
6244 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6245 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6246 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6247 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6248 instead of int_size_in_bytes.
6249 * expr.c (copy_blkmode_to_reg): Likewise.
6250 * function.c (aggregate_value_p): Return 0 for empty types.
6251 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6253 (locate_and_pad_parm): Call arg size_in_bytes instead
6255 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6256 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6257 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6259 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6260 (std_gimplify_va_arg_expr): Skip empty records. Call
6261 arg_size_in_bytes instead size_in_bytes.
6262 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6263 * tree-core.h (tree_type_common): Add empty_flag.
6264 (tree_decl_common): Update comments.
6265 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6267 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6268 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6270 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6271 * tree.c (default_is_empty_type): New function.
6272 (default_is_empty_record): New function.
6273 (arg_int_size_in_bytes): New function.
6274 (arg_size_in_bytes): New function.
6275 (get_ultimate_context): New function.
6276 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6277 TRANSLATION_UNIT_WARN_EMPTY_P.
6278 (default_is_empty_record, arg_int_size_in_bytes,
6279 arg_size_in_bytes, get_ultimate_context): Declare.
6281 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6283 * config/arm/arm.c (cmse_clear_registers): New function.
6284 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6285 code by call to cmse_clear_registers.
6286 (cmse_nonsecure_entry_clear_before_return): Likewise.
6288 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6290 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6291 (vdot_s32, vdotq_s32): New.
6292 (vdot_lane_u32, vdotq_lane_u32): New.
6293 (vdot_lane_s32, vdotq_lane_s32): New.
6296 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6299 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6300 for unsigned values with fewer HWIs than the precision.
6301 (test_overflow): New function.
6302 (wide_int_cc_tests): Call it.
6304 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6306 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6307 has the same mode class as Pmode.
6309 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6311 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6312 padding_bits_to_clear_ptr.
6313 (cmse_nonsecure_entry_clear_before_return): Likewise.
6315 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6317 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6318 auto_sbitap instead of integer bitfield to control register needing
6321 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6323 PR tree-optimization/83044
6324 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6325 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6326 that need the upper bound. Subtract offset from
6327 get_addr_base_and_unit_offset only if positive and subtract it
6328 before division by eltsize rather than after it.
6331 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6332 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6333 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6337 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6338 expand_binop with *mul_widen_optab, make sure at least one of the
6339 operands doesn't have VOIDmode.
6342 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6344 PR rtl-optimization/82044
6345 PR tree-optimization/82042
6346 * dse.c (record_store): Check for overflow.
6347 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6348 add_wild_read instead of clear_rhs_from_active_local_stores on
6349 overflow and log it into dump_file.
6351 2017-11-22 Richard Biener <rguenther@suse.de>
6353 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6354 fake edges to exit when looking for a place to insert.
6355 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6357 (insert_into_preds_of_block): Commit edge insertion immediately,
6358 assert that doesn't require new BBs.
6359 (fini_pre): Release expressions.
6360 (pass_pre::execute): Shuffle things around a bit, if the fn
6361 is too large do not compute AVAIL either as this is really the
6364 2017-11-22 Richard Biener <rguenther@suse.de>
6366 PR tree-optimization/83089
6367 * tree-if-conv.c (pass_if_conversion::execute): If anything
6368 changed reset SCEV and free the number of iteration estimates.
6370 2017-11-21 Martin Sebor <msebor@redhat.com>
6372 PR tree-optimization/82945
6373 * calls.h (warn_nonstring_bound): Remove unused function.
6375 2017-11-21 Martin Sebor <msebor@redhat.com>
6377 PR tree-optimization/82945
6378 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6379 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6380 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6382 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6383 * calls.h (get_attr_nonstring_decl): Declare new function.
6384 * doc/extend.texi (attribute nonstring): Update.
6385 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6386 get_attr_nonstring_decl and handle it.
6387 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6388 detection of nul-termination.
6389 (strlen_to_stridx): Change to a pointer.
6390 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6391 (pass_strlen::execute): Same.
6393 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6395 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6396 for the new option -mprefer-vector-width=[none|128|256|512].
6397 * config/i386/i386.c (ix86_target_string): remove old style options
6398 -mprefer-avx256 and make -mprefer-avx128 as alias.
6399 (ix86_option_override_internal): Apply defaults for the
6400 -mprefer-vector-width=[128|256] option.
6401 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6402 Implement macros to work with -mprefer-vector-width=.
6403 * config/i386/i386.opt: Implemented option
6404 -mprefer-vector-width=[none|128|256|512].
6405 * doc/invoke.texi: Documentation for
6406 -mprefer-vector-width=[none|128|256|512].
6408 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6410 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6412 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6414 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6416 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6418 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6420 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6421 * expr.c (expand_expr_real_2): Likewise.
6422 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6423 fold_binary_loc): Likewise.
6424 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6425 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6426 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6427 MINUS_EXPR transformations.
6428 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6429 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6430 * tree-inline.c (estimate_operator_cost): Likewise.
6431 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6432 op_symbol_code): Likewise.
6433 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6434 * vr-values.c (extract_range_from_binary_expr): Likewise.
6435 * varasm.c (initializer_constant_valid_p_1): Likewise.
6436 * tree.def: New tree code POINTER_DIFF_EXPR.
6438 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6440 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6441 integer suffix to movbe mnemonic.
6442 (*bswaphi2_movbe): Ditto.
6443 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6445 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6448 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6449 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6450 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6452 2017-11-21 Martin Liska <mliska@suse.cz>
6454 * tree-inline.c (expand_call_inline): Remove not needed
6457 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6458 Jakub Jelinek <jakub@redhat.com>
6461 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6462 Remove static keyword from f variable.
6464 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6466 PR tree-optimization/83086
6467 * gimple-ssa-store-merging.c
6468 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6469 rather than n.base_addr.
6471 2017-11-21 Martin Liska <mliska@suse.cz>
6473 PR rtl-optimization/82044
6474 PR tree-optimization/82042
6475 * dse.c (check_mem_read_rtx): Check for overflow.
6477 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6479 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6482 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6484 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6485 and its corresponding call arg location note.
6487 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6488 Andrew Burgess <andrew.burgess@embecosm.com>
6490 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6492 (arc_return_slot_offset): Likewise.
6493 (arc_eh_return_address_location): New declaration.
6494 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6495 (MUST_SAVE_REGISTER): Add exception handler case.
6496 (MUST_SAVE_RETURN_ADDR): Likewise.
6497 (arc_frame_pointer_required): Likewise.
6498 (arc_frame_pointer_needed): New function.
6499 (arc_compute_frame_size): Changed.
6500 (arc_expand_prologue): Likewise.
6501 (arc_expand_epilogue): Likewise.
6502 (arc_initial_elimination_offset): Likewise.
6503 (arc_return_slot_offset): Delete.
6504 (arc_eh_return_address_location): New function.
6505 (arc_builtin_setjmp_frame_value): Likewise.
6506 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6507 (EH_RETURN_STACKADJ_RTX): Define.
6508 (EH_RETURN_HANDLER_RTX): Likewise.
6509 * config/arc/arc.md (eh_return): Delete.
6511 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6513 * print-tree.h (debug_vec_tree): Remove prototype.
6514 * gdbinit.in (pvt): Remove macro.
6516 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6518 PR tree-optimization/83047
6519 * gimple-ssa-store-merging.c
6520 (imm_store_chain_info::output_merged_store): If the loads with the
6521 same vuse are in different basic blocks, for load_gsi pick a load
6522 location that is dominated by the other loads.
6525 * config/i386/i386.c (ix86_memmodel_check): Start
6526 -Winvalid-memory-model diagnostics with lowercase letter.
6529 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6530 set in_cold_section_p to the partition containing loc_list->first.
6531 When seeing loc_list->last_before_switch node, update secname and
6532 perform range_across_switch second partition handling only after that.
6535 * run-rtl-passes.c: Include debug.h.
6536 (run_rtl_passes): Call debug_hooks->assembly_start.
6537 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6541 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6542 OPTAB_DIRECT in calls to expand_simple_binop.
6544 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6547 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6549 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6552 * spellcheck.h (best_match::blithely_get_best_candidate): New
6555 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6557 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6558 with lower case letter.
6560 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6562 * config/i386/i386.md (bswaphi2): New expander.
6563 (*bswaphi2_movbe): New insn pattern.
6564 (bswaphi -> rorhi pepehole2): New peephole pattern.
6566 2017-11-20 Jeff Law <law@redhat.com>
6568 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6569 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6570 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6571 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6572 gimple-ssa-evrp-analyze.h.
6574 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6575 set BB_VISITED here.
6576 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6578 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6580 * tree-predcom.c: Add general comment on Store-Store chains.
6581 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6583 (get_chain_last_ref_at): Rename into...
6584 (get_chain_last_write_at): ...this.
6585 (get_chain_last_write_before_load): New function.
6586 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6587 CT_STORE_STORE when write reference is added.
6588 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6589 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6590 (initialize_root_vars_store_elim_1): Ditto.
6591 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6592 definition is created.
6593 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6594 chain by replacing it with dominant stored value.
6596 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6598 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6600 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6602 * vr-values.c (extract_range_from_binary_expr): Use a full range
6605 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6607 * config/arm/arm.md (R4_REGNUM): Define constant.
6608 (nonsecure_call_internal): Remove r4 clobber.
6609 (nonsecure_call_value_internal): Likewise.
6610 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6611 clobber and resequence match_operands.
6612 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6613 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6614 (nonsecure_call_value_reg_thumb2): Likewise.
6616 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6618 PR tree-optimization/78821
6619 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6620 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6621 to the address of the base rather than the base itself.
6622 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6623 (find_bswap_or_nop_finalize): New function.
6624 (find_bswap_or_nop): Use it.
6625 (bswap_replace): Return a tree rather than bool, change first
6626 argument from gimple * to gimple_stmt_iterator, allow inserting
6627 into an empty sequence, allow ins_stmt to be NULL - then emit
6628 all stmts into gsi. Fix up MEM_REF address gimplification.
6629 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6630 (struct store_immediate_info): Add N and INS_STMT non-static
6632 (store_immediate_info::store_immediate_info): Initialize them
6633 from newly added ctor args.
6634 (merged_store_group::apply_stores): Formatting fixes. Sort by
6636 (stmts_may_clobber_ref_p): For stores call also
6637 refs_anti_dependent_p.
6638 (gather_bswap_load_refs): New function.
6639 (imm_store_chain_info::try_coalesce_bswap): New method.
6640 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6641 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6642 (imm_store_chain_info::output_merged_store): Fail if number of
6643 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6644 and NOP_EXPR rhs_code.
6645 (pass_store_merging::process_store): Compute n and ins_stmt, if
6646 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6647 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6649 (pass_store_merging::execute): Calculate dominators.
6651 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6652 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6653 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6654 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6655 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6656 class pass_optimize_bswap, bswap_replace,
6657 pass_optimize_bswap::execute): Moved to ...
6658 * gimple-ssa-store-merging.c: ... this file.
6659 Include optabs-tree.h.
6660 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6661 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6662 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6663 anonymous namespace, remove static keywords.
6664 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6665 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6667 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6670 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6672 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6674 * vec.h (debug_helper): New function.
6675 (DEFINE_DEBUG_VEC): New macro.
6676 * hash-set.h (debug_helper): New function.
6677 (DEFINE_DEBUG_HASH_SET): New macro.
6678 * cfg.c (debug_slim (edge)): New function.
6679 Call DEFINE_DEBUG_VEC for edges.
6680 Call DEFINE_DEBUG_HASH_SET for edges.
6681 * cfghooks.c (debug_slim (basic_block)): New function.
6682 Call DEFINE_DEBUG_VEC for basic blocks.
6683 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6684 * print-tree.c (debug_slim): New function to handle trees.
6685 Call DEFINE_DEBUG_VEC for trees.
6686 Call DEFINE_DEBUG_HASH_SET for trees.
6687 (debug (vec<tree, va_gc>) &): Remove.
6688 (debug (<vec<tree, va_gc>) *): Remove.
6689 * print-rtl.c (debug_slim): New function to handle const_rtx.
6690 Call DEFINE_DEBUG_VEC for rtx_def.
6691 Call DEFINE_DEBUG_VEC for rtx_insn.
6692 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6693 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6694 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6695 (debug (vec<rtx_insn *> *ptr): Remove.
6696 (debug_insn_vector): Remove.
6697 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6699 2017-11-20 Tom de Vries <tom@codesourcery.com>
6701 PR rtl-optimization/82020
6702 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6703 IF_THEN_ELSE condition.
6705 2017-11-19 Jeff Law <law@redhat.com>
6707 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6708 of degenerates resulting from ignoring an edge.
6710 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6713 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6715 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6718 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6719 for uninitialied counts.
6721 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6724 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6726 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6729 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6731 2017-11-19 Tom de Vries <tom@codesourcery.com>
6733 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6736 2017-11-19 Tom de Vries <tom@codesourcery.com>
6738 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6741 2017-11-19 Tom de Vries <tom@codesourcery.com>
6743 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6746 2017-11-19 Tom de Vries <tom@codesourcery.com>
6748 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6750 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6752 2017-11-19 Tom de Vries <tom@codesourcery.com>
6754 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6756 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6757 semicolon after MCORE_EXPORT_NAME call.
6759 2017-11-19 Tom de Vries <tom@codesourcery.com>
6762 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6763 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6765 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6767 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6768 (cgraph_edge::clone): Cleanup updating of profile.
6769 * ipa-cp.c (update_profiling_info): Likewise.
6770 * ipa-inline-transform.c (inline_transform): Likewise.
6771 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6772 * ipa-split.c (execute_split_functions): Do not split when function
6774 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6775 * profile-count.c (profile_count::dump): Add global0.
6776 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6778 (profile_count::to_sreal_scale): Likewise.
6779 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6780 (profile_count::combine_with_ipa_count): New function.
6781 * profile-count.h (profile_guessed_global0adjusted): New.
6782 (profile_count::adjusted_zero): New.
6783 (profile_count::global0adjusted): New.
6784 (profile_count::combine_with_ipa_count): New.
6785 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6786 correct profile of return block of split functions.
6787 (copy_cfg_body): Remove unused profile_count.
6788 (copy_body): Likewise.
6789 (expand_call_inline): Update.
6790 (tree_function_versioning): Update.
6792 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6794 * hash-set.h (hash_set::empty): New.
6795 * tree-ssa-threadbackward.h: Delete.
6796 * tree-ssa-threadbackward.c (class thread_jumps): New.
6797 Move max_threaded_paths into class.
6798 (fsm_find_thread_path): Remove arguments that are now in class.
6799 (profitable_jump_thread_path): Rename to...
6800 (thread_jumps::profitable_jump_thread_path): ...this.
6801 (convert_and_register_jump_thread_path): Rename to...
6802 (thread_jumps::convert_and_register_current_path): ...this.
6803 (check_subpath_and_update_thread_path): Rename to...
6804 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6805 (register_jump_thread_path_if_profitable): Rename to...
6806 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6807 (handle_phi): Rename to...
6808 (thread_jumps::handle_phi): ...this.
6809 (handle_assignment): Rename to...
6810 (thread_jumps::handle_assignment): ...this.
6811 (fsm_find_control_statement_thread_paths): Rename to...
6812 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6813 (find_jump_threads_backwards): Rename to...
6814 (thread_jumps::find_jump_threads_backwards): ...this.
6815 Initialize path local data.
6816 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6817 from within thread_jumps class.
6818 (pass_early_thread_jumps::execute): Same.
6820 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6822 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6824 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6826 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6827 to canonical location.
6829 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6832 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6834 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6836 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6838 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6839 rcrt1.o%s/grcrt1.o%s for -static-pie.
6841 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6843 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6844 ix86_shift_rotate_cost): Break out from ...
6845 (ix86_rtx_costs): ... here.
6846 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6849 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6851 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6852 when entry block was promoted unlikely.
6853 (estimate_bb_frequencies): Increase frequency scale.
6854 * profile-count.h (profile_count): Export precision info.
6856 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6858 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6859 disturbin profile of entry block.
6861 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6863 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6866 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6868 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6869 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6871 2017-11-17 Jeff Law <law@redhat.com>
6873 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6874 from evrp_dom_walker class. Various methods moved into new class.
6875 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6876 (evrp_range_analyzer::enter): New method.
6877 (evrp_range_analyzer::leave): New method.
6878 (evrp_dom_walker): Remove delegators no longer needed by this class.
6879 Replace vr_values data member with evrp_range_analyzer
6881 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6882 method extracted from evrp_dom_walker::before_dom_children.
6883 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6884 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6886 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6887 Add private copy constructor and move assignment operators.
6888 Privatize methods and class data where trivially possible.
6889 (evrp_dom_walker::cleanup): New function, extracted from
6890 execute_early_vrp. Simplify access to class data.
6892 * vr-values.h (get_output_for_vrp): Prototype.
6893 * vr-values.c (get_output_for_vrp): New function extracted from
6894 vrp_visit_assignment_or_call and extract_range_from_stmt.
6895 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6897 2017-11-17 Luis Machado <luis.machado@linaro.org>
6899 * config/aarch64/aarch64.c
6900 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6901 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6902 tune_params::AUTOPREFETCHER_WEAK.
6904 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6907 * config/arm/arm.c (arm_valid_target_attribute_rec):
6908 Parse "arch=" and "+<ext>".
6909 (arm_valid_target_attribute_tree): Re-init global options.
6910 (arm_option_override): Make non-static.
6911 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6912 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6913 (__ARM_FEATURE_CRC32): Support undef.
6914 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6915 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6917 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6919 * gdbinit.in (break-on-diagnostic): New command.
6921 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6923 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6924 return to indirect jump for EH return if control-flow
6925 protection is enabled. Change explicit 'false' argument in
6926 pro_epilogue_adjust_stack with a value of
6928 * config/i386/i386.md (simple_return_indirect_internal):
6929 Remove SImode restriction to support 64-bit.
6931 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6933 * combine.c (added_notes_insn): New.
6934 (try_combine): Handle added_notes_insn like added_links_insn.
6935 Rewrite return value code.
6936 (distribute_notes): Set added_notes_insn to the earliest insn we added
6939 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6941 PR rtl-optimization/82621
6942 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6943 dest of one of those SETs is unused.
6945 2017-11-17 Richard Biener <rguenther@suse.de>
6948 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6949 * tree-pretty-print.c (dump_generic_node): Handle
6950 annot_expr_parallel_kind.
6951 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6952 * gimplify.c (gimple_boolify): Likewise.
6954 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6956 * config.gcc (extra_headers): Add cet.h for x86 targets.
6957 * config/i386/cet.h: New file.
6958 * doc/install.texi: Add --enable-cet/--disable-cet.
6960 2017-11-17 Richard Biener <rguenther@suse.de>
6962 PR tree-optimization/83017
6963 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6964 (gen_parallel_loop): Properly count iterations.
6965 (parallelize_loops): Handle loop->can_be_parallel independent
6966 of flag_loop_parallelize_all. Make static profitability test match
6968 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6969 * invoke.texi (parloops-min-per-thread): Document.
6971 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6973 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6974 upstreaming review comments.
6976 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6978 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6979 with fast unaligned access.
6980 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6982 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6984 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6985 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6987 2017-11-17 Richard Biener <rguenther@suse.de>
6989 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6990 folding of references.
6992 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6995 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6996 of replacing call to strncmp with corresponding call to strcmp when
6999 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
7001 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
7002 option prefer-avx256 for skylake-avx512 configuration.
7003 * config/i386/i386.c (ix86_option_override_internal): Ditto.
7004 (get_builtin_code_for_version): Ditto.
7006 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7007 Monk Chiang <sh.chiang04@gmail.com>
7009 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
7010 (FIXED_REGISTERS): Reserve more register numbers.
7011 (CALL_USED_REGISTERS): Likewise.
7012 (REG_ALLOC_ORDER): Likewise.
7013 (REG_CLASS_CONTENTS): Likewise.
7014 (REGISTER_NAMES): Likewise.
7016 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7017 Kito Cheng <kito.cheng@gmail.com>
7019 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
7021 * config/nds32/iterators.md: Add vector mode iterators and attributes.
7023 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
7025 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
7026 parameter list for vec_splats.
7028 2017-11-16 Joseph Myers <joseph@codesourcery.com>
7030 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
7032 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
7034 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
7037 * doc/install.texi: Document incompatibility of Perl >=5.6.26
7038 with the required version of automake 1.11.6.
7040 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
7042 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
7044 (DU_C2_3_power9): Correct reservation combinations.
7045 (FP_DIV_power9, VEC_DIV_power9): New.
7046 (power9-alu): Split out rotate/shift...
7047 (power9-rot): ...to here, correct dispatch resource.
7048 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
7050 (power9-fp): Correct latency.
7051 (power9-sdiv): Add div/sqrt resource.
7052 (power9-ddiv): Correct latency, add div/sqrt resource.
7053 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
7054 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
7056 (power9-qpdiv, power9-qpmul): Adjust resource usage.
7058 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
7060 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7061 switch statement mapping KF built-ins to TF built-ins if we don't
7062 have the proper ISA 3.0 assembler support.
7064 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7066 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7067 (lower_emutls_function_body): Do not compute it.
7069 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7071 * ipa-split.c (split_bb_info): Turn time to sreal.
7072 (split_point): Likewise.
7073 (dump_split_point): Likewise.
7074 (fine_split_points): Likewise.
7075 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7077 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7079 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7082 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7084 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7086 (expensive_function_p): Remove useless assert.
7087 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7089 2017-11-16 Martin Sebor <msebor@redhat.com>
7091 PR tree-optimization/82588
7092 PR tree-optimization/82583
7093 * tree-vrp.c (check_array_ref): Handle flexible array members,
7094 string literals, and inner indices.
7095 (search_for_addr_array): Add detail to diagnostics.
7097 2017-11-16 Nathan Sidwell <nathan@acm.org>
7101 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7102 (SET_DECL_ASSEMBLER_NAME): Forward to
7103 overwrite_decl_assembler_name.
7104 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7105 (overwrite_decl_assembler_name): Declare.
7106 * tree.c (overwrite_decl_assembler_name): New.
7107 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7108 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7109 (LANG_HOOKS_INITIALIZER): Add it.
7110 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7111 * langhooks.c (lhd_set_decl_assembler_name): Use
7112 SET_DECL_ASSEMBLER_NAME.
7113 (lhd_overwrite_decl_assembler_name): Default implementation.
7115 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7116 Jackson Woodruff <jackson.woodruff@arm.com>
7118 PR tree-optimization/71026
7119 * match.pd: Canonicalize constant multiplies in division.
7121 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7123 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7124 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7125 Remove OPT_fomit_frame_pointer entry.
7126 * common/config/arc/arc-common.c: Likewise.
7127 * common/config/arm/arm-common.c: Likewise.
7128 * common/config/avr/avr-common.c: Likewise.
7129 * common/config/c6x/c6x-common.c: Likewise.
7130 * common/config/cr16/cr16-common.c: Likewise.
7131 * common/config/cris/cris-common.c: Likewise.
7132 * common/config/epiphany/epiphany-common.c: Likewise.
7133 * common/config/fr30/fr30-common.c: Likewise.
7134 * common/config/frv/frv-common.c: Likewise.
7135 * common/config/ia64/ia64-common.c: Likewise.
7136 * common/config/iq2000/iq2000-common.c: Likewise.
7137 * common/config/lm32/lm32-common.c: Likewise.
7138 * common/config/m32r/m32r-common.c: Likewise.
7139 * common/config/mcore/mcore-common.c: Likewise.
7140 * common/config/microblaze/microblaze-common.c: Likewise.
7141 * common/config/mips/mips-common.c: Likewise.
7142 * common/config/mmix/mmix-common.c: Likewise.
7143 * common/config/mn10300/mn10300-common.c: Likewise.
7144 * common/config/nios2/nios2-common.c: Likewise.
7145 * common/config/pa/pa-common.c: Likewise.
7146 * common/config/pdp11/pdp11-common.c: Likewise.
7147 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7148 * common/config/riscv/riscv-common.c: Likewise.
7149 * common/config/rs6000/rs6000-common.c: Likewise.
7150 * common/config/rx/rx-common.c: Likewise.
7151 * common/config/s390/s390-common.c: Likewise.
7152 * common/config/sh/sh-common.c: Likewise.
7153 * common/config/sparc/sparc-common.c: Likewise.
7154 * common/config/tilegx/tilegx-common.c: Likewise.
7155 * common/config/tilepro/tilepro-common.c: Likewise.
7156 * common/config/v850/v850-common.c: Likewise.
7157 * common/config/visium/visium-common.c: Likewise.
7158 * common/config/xstormy16/xstormy16-common.c: Likewise.
7159 * common/config/xtensa/xtensa-common.c: Likewise.
7160 * invoke.texi (-fomit-frame-pointer): Update documentation.
7162 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7164 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7166 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7168 * cfg.c (scale_bbs_frequencies_int,
7169 cale_bbs_frequencies_gcov_type): Remove.
7170 * cfg.h (scale_bbs_frequencies_int,
7171 cale_bbs_frequencies_gcov_type): Remove.
7173 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7175 * tree-ssa-loop-manip.c
7176 (scale_dominated_blocks_in_loop): Update to profile counts.
7177 (tree_transform_and_unroll_loop): Likewise.
7179 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7181 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7182 scale_bbs_frequencies_int.
7184 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7186 * final.c (compute_alignments): Use counts rather than frequencies.
7188 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7190 * cfgloopanal.c: Include sreal.h
7191 (average_num_loop_insns): Use counts and sreal for accounting.
7193 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7195 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7198 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7199 Kito Cheng <kito.cheng@gmail.com>
7201 * config/nds32/constraints.md: Provide more constraints.
7202 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7203 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7204 support constraints usage.
7206 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7208 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7210 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7211 Kito Cheng <kito.cheng@gmail.com>
7213 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7214 * config/nds32/nds32.opt: Refine the layout.
7215 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7216 TARGET_EXT_STRING): Support new options.
7217 * config/nds32/nds32.h: Likewise.
7218 * config/nds32/nds32.md: Likewise.
7219 * config/nds32/nds32-predicates.c: Likewise.
7220 * config/nds32/constraints.md: Likewise.
7221 * common/config/nds32/nds32-common.c: Likewise.
7223 2017-11-16 Julia Koval <julia.koval@intel.com>
7226 * config/i386/gfniintrin.h: Add sse check.
7227 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7229 2017-11-16 Julia Koval <julia.koval@intel.com>
7231 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7232 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7233 (ix86_handle_option): Handle -mavx512vbmi2.
7234 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7235 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7236 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7237 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7238 (ix86_valid_target_attribute_inner_p): Ditto.
7239 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7240 * config/i386/i386.opt (mavx512vbmi2): New option.
7241 * doc/invoke.texi: Add new option.
7243 2017-11-16 Julia Koval <julia.koval@intel.com>
7245 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7246 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7247 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7248 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7249 _mm512_gf2p8mul_epi8): New intrinsics.
7250 * config/i386/i386-builtin-types.def
7251 (V64QI_FTYPE_V64QI_V64QI): New type.
7252 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7253 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7254 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7255 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7256 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7257 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7259 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7261 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7262 explicitly as a stream of bytes.
7264 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7266 * config/rs6000/altivec.h (vec_xst_be): New #define.
7267 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7268 and externalize from *altivec_vperm_<mode>_internal.
7269 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7271 (XL_BE_V8HI): Likewise.
7272 (XL_BE_V4SI): Likewise.
7273 (XL_BE_V4SI): Likewise.
7274 (XL_BE_V2DI): Likewise.
7275 (XL_BE_V4SF): Likewise.
7276 (XL_BE_V2DF): Likewise.
7277 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7278 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7279 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7280 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7281 VSX_BUILTIN_VEC_XST_BE.
7282 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7283 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7285 (altivec_init_builtins): Replace conditional calls to def_builtin
7286 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7287 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7288 with unconditional calls. Remove calls to def_builtin for
7289 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7290 __builtin_vec_xst_be.
7291 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7292 to define_expand, and add alternate RTL generation for P8.
7293 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7294 vsx_ld_elemrev_v8hi.
7295 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7296 add alternate RTL generation for P8.
7297 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7298 vsx_ld_elemrev_v16qi.
7299 (vsx_st_elemrev_v8hi): Convert define_insn
7300 to define_expand, and add alternate RTL generation for P8.
7301 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7302 vsx_st_elemrev_v8hi.
7303 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7304 add alternate RTL generation for P8.
7305 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7306 vsx_st_elemrev_v16qi.
7308 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7311 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7312 TARGET_AVX512ER check.
7313 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7314 neither -mzeroupper nor -mno-zeroupper is used and
7315 TARGET_EMIT_VZEROUPPER is set.
7316 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7317 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7319 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7321 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7322 folding of vector compares.
7323 (fold_build_vec_cmp): New helper function.
7324 (fold_compare_helper): New helper function.
7325 (builtin_function_type): Add compare builtins to the list of functions
7326 having unsigned arguments. Cosmetic updates to comment indentation.
7327 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7328 the not+eq combination.
7330 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7332 PR tree-optimization/82726
7333 PR tree-optimization/70754
7334 * tree-predcom.c (order_drefs_by_pos): New function.
7335 (combine_chains): Move code setting has_max_use_after to...
7336 (try_combine_chains): ...here. New parameter. Sort combined chains
7337 according to position information.
7338 (tree_predictive_commoning_loop): Update call to above function.
7339 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7341 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7343 PR tree-optimization/82726
7345 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7347 PR tree-optimization/70754
7348 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7349 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7350 combined stmt before it if not NULL.
7351 (combine_chains): Process refs reversely and compute dominance point
7355 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7357 PR tree-optimization/79663
7358 * tree-predcom.c (combine_chains): Process refs in reverse order
7359 only for ZERO length chains, and add explaining comment.
7361 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7363 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7364 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7365 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7367 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7369 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7371 2017-11-15 Martin Liska <mliska@suse.cz>
7373 * tree-cfg.c (pass_warn_function_return::execute):
7374 Compare warn_return_type for greater than zero.
7376 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7380 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7381 to return true on Xeon and not on Xeon Phi.
7382 (ix86_check_avx256_register): Changed to ...
7383 (ix86_check_avx_upper_register): ... this. Add extra check for
7384 VALID_AVX512F_REG_OR_XI_MODE.
7385 (ix86_avx_u128_mode_needed): Changed
7386 ix86_check_avx256_register to ix86_check_avx_upper_register.
7387 (ix86_check_avx256_stores): Changed to ...
7388 (ix86_check_avx_upper_stores): ... this. Changed
7389 ix86_check_avx256_register to ix86_check_avx_upper_register.
7390 (ix86_avx_u128_mode_after): Changed
7391 avx_reg256_found to avx_upper_reg_found. Changed
7392 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7393 (ix86_avx_u128_mode_entry): Changed
7394 ix86_check_avx256_register to ix86_check_avx_upper_register.
7395 (ix86_avx_u128_mode_exit): Ditto.
7396 * config/i386/i386.h: (host_detect_local_cpu): New define.
7398 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7400 * config/arm/xgene1.md (xgene1): Split into automatons
7401 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7402 (xgene1_f_load): Adjust reservations and/or types.
7403 (xgene1_f_store): Likewise.
7404 (xgene1_load_pair): Likewise.
7405 (xgene1_store_pair): Likewise.
7406 (xgene1_fp_load1): Likewise.
7407 (xgene1_load1): Likewise.
7408 (xgene1_store1): Likewise.
7409 (xgene1_move): Likewise.
7410 (xgene1_alu): Likewise.
7411 (xgene1_simd): Likewise.
7412 (xgene1_bfm): Likewise.
7413 (xgene1_neon_load1): Likewise.
7414 (xgene1_neon_store1): Likewise.
7415 (xgene1_neon_logic): Likewise.
7416 (xgene1_neon_st1): Likewise.
7417 (xgene1_neon_ld1r): Likewise.
7418 (xgene1_alu_cond): Added.
7419 (xgene1_shift_reg): Likwise.
7420 (xgene1_bfx): Likewise.
7421 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7423 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7426 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7428 (can_widen_mult_without_libcall): New function.
7429 (expand_mul_overflow): If only checking unsigned mul overflow,
7430 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7431 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7432 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7433 (expand_DIVMOD): Formatting fix.
7434 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7435 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7436 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7438 PR tree-optimization/82977
7439 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7440 constructed temporary to strlen_to_stridx.put.
7442 2017-11-15 Martin Liska <mliska@suse.cz>
7444 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7445 * configure: Regenerate.
7447 2017-11-15 Martin Liska <mliska@suse.cz>
7450 * config/sh/sh-mem.cc: Use proper probability for
7453 2017-11-14 Jeff Law <law@redhat.com>
7455 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7456 the red zone for stack_clash_protection_final_dynamic_probe targets
7457 when the total dynamic stack size is zero bytes.
7459 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7460 blocks is post order.
7462 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7464 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7465 * final.c (rest_of_clean_state): Set it for the
7466 -fcompare-debug dump.
7467 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7468 class when TDF_COMPARE_DEBUG is set.
7470 * dwarf2out.c (gen_producer_string): Discard
7473 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7476 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7477 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7478 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7479 (__TGMATH_REAL_2_3): Remove macros.
7480 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7481 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7483 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7485 (remquo): Define using __TGMATH_REAL_3.
7487 2017-11-14 Jeff Law <law@redhat.com>
7489 * vr-values.c: New file with contents extracted from tree-vrp.c.
7490 * Makefile.in (OBJS): Add vr-values.o
7491 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7492 (set_value_range, set_and_canonicalize_value_range): Likewise.
7493 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7494 (value_range_constant_singleton, symbolic_range_p): Likewise.
7495 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7496 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7497 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7498 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7499 (find_case_label_range, find_case_label_index): Likewise.
7500 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7501 (range_int_cst_singleton_p, value_inside_range): Likewise.
7502 (get_single_symbol): Likewise.
7503 (switch_update): Move structure definition here.
7504 (to_remove_edges, to_update_switch_stmts): Provide externs.
7505 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7506 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7507 (vrp_val_is_min, set_value_range): Likewise.
7508 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7509 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7510 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7511 (range_is_nonnull, range_int_cst_p): Likewwise.
7512 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7513 (get_single_symbol, operand_less_p): Likewise
7514 (compare_values_warnv, compare_values): Likewise.
7515 (value_inside_range, value_range_constant_singleton): Likewise.
7516 (zero_nonzero_bitgs_from_vr): Likewise.
7517 (extract_range_from_binary_expr_1): Likewise.
7518 (overflow_comparison_p): Likewise.
7519 (to_remove_edges, to_update_switch_stmts): Likewise.
7520 (find_case_label-index, find_case_label_range): Likewise.
7521 (switch_update, set_value_range_to_nonnegative): Remove.
7522 (set_value_range_to_truthvalue): Likewise.
7523 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7524 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7525 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7526 (find_case_label_ranges, test_for_singularity): Likewise.
7527 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7528 (x_vr_values): Move to its remaining use site.
7530 2017-11-10 Jeff Law <law@redhat.com>
7532 * vr-values.h (VR_INITIALIZER): Move #define here.
7533 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7534 * Makefile.in (OBJS): Add tree-evrp.o
7535 * tree-vrp.h (assert_info): Move structure definition here.
7536 (set_value_range_to_varying): Prototype.
7537 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7538 (infer_value_range, register_edge_assert_for): Likewise.
7539 (stmt_interesting_for_vrp): Likewise.
7540 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7541 (set_value_range_to_varying): No longer static.
7542 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7543 (infer_value_range, register_edge_assert_for): Likewise.
7545 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7547 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7548 generate the XXBRD instruction.
7550 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7551 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7552 as being compatible if -mabi=ieeelongdouble.
7553 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7554 to setup float128 built-ins with hardware support.
7555 (BU_FLOAT128_HW_2): Likewise.
7556 (BU_FLOAT128_HW_3): Likewise.
7557 (BU_FLOAT128_HW_VSX_1): Likewise.
7558 (BU_FLOAT128_HW_VSX_2): Likewise.
7559 (scalar_extract_expq): Change float128 built-in functions to
7560 accommodate having both KFmode and TFmode functions. Use the
7561 KFmode variant as the default.
7562 (scalar_extract_sigq): Likewise.
7563 (scalar_test_neg_qp): Likewise.
7564 (scalar_insert_exp_q): Likewise.
7565 (scalar_insert_exp_qp): Likewise.
7566 (scalar_test_data_class_qp): Likewise.
7567 (sqrtf128_round_to_odd): Delete processing the round to odd
7568 built-in functions as special built-in functions, and define them
7569 as float128 built-ins. Use the KFmode variant as the default.
7570 (truncf128_round_to_odd): Likewise.
7571 (addf128_round_to_odd): Likewise.
7572 (subf128_round_to_odd): Likewise.
7573 (mulf128_round_to_odd): Likewise.
7574 (divf128_round_to_odd): Likewise.
7575 (fmaf128_round_to_odd): Likewise.
7576 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7577 support for KFmode and TFmode xststdcqp calls.
7578 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7579 point, switch the built-in handlers for the get/set float128
7580 exponent, get float128 mantissa, float128 test built-ins, and the
7581 float128 round to odd built-in functions. Eliminate creating the
7582 float128 round to odd built-in functions as special built-ins.
7583 (rs6000_init_builtins): Eliminate special creation of the float128
7584 round to odd built-in functions.
7585 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7586 function insns to support both TFmode and KFmode variants.
7587 (xsxsigqp_<mode>): Likewise.
7588 (xsiexpqpf_<mode>): Likewise.
7589 (xsiexpqp_<mode>): Likewise.
7590 (xststdcqp_<mode>): Likewise.
7591 (xststdcnegqp_<mode>): Likewise.
7592 (xststdcqp_<mode>): Likewise.
7594 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7596 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7597 unused path_in_freq_ptr parameter.
7598 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7600 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7602 * ipa-inline.c (edge_badness): Dump sreal frequency.
7603 (compute_inlined_call_time): Match natural implementaiton ...
7604 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7605 forgotten division by CGRAPH_FREQ_BASE.
7607 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7609 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7610 Solaris 11. Update comment.
7611 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7613 * config/sol2.h (STARTFILE_SPEC): Likewise.
7614 * configure: Regenerate.
7616 2017-11-14 Carl Love <cel@us.ibm.com>
7618 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7619 le_ and be_ prefixes to swap* variables. Remove
7620 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7623 2017-11-14 Jason Merrill <jason@redhat.com>
7625 Support GTY((cache)) on hash_map.
7626 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7627 (ggc_cache_remove): Override it instead of ggc_mx.
7628 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7629 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7630 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7631 (simple_cache_map_traits): Override maybe_mx.
7632 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7633 (hash_map): Friend gt_cleare_cache.
7634 (gt_cleare_cache): New.
7635 * tree.h (tree_cache_traits): New hash_map traits class.
7636 (tree_cache_map): New typedef.
7638 2017-11-14 Richard Biener <rguenther@suse.de>
7640 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7641 paramter and handling.
7642 (cleanup_control_flow_bb): Likewise.
7643 (cleanup_control_flow_pre): New helper performing a DFS walk
7644 to call cleanup_control_flow_bb in PRE order.
7645 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7646 via cleanup_control_flow_pre.
7648 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7650 * config/aarch64/aarch64-simd.md
7651 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7652 (*aarch64_simd_bsl<mode>_alt): Likewise.
7653 (aarch64_simd_bsldi_internal): New.
7654 (aarch64_simd_bsldi_alt): Likewise.
7656 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7658 * tracer.c (better_p): Do not compare frequencies.
7659 * reg-stack.c (better_edge): Likewise.
7660 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7663 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7665 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7666 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7667 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7668 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7669 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7670 * predict.c (maybe_hot_frequency_p): Inline to ...
7671 (maybe_hot_count_p): ... here; rewrite to counts.
7672 (counts_to_freqs): Rename to ...
7673 (update_max_bb_count): ... this one.
7674 (expensive_function_p): Use counts.
7675 (estimate_bb_frequencies): Update.
7676 (rebuild_frequencies): Update.
7677 * predict.h (counts_to_freqs): Rename to ...
7678 (update_max_bb_count): ... this one.
7679 * profile.c (compute_branch_probabilities): Add debug info
7680 * tree-inline.c (expand_call_inline): Update debug info.
7681 (optimize_inline_calls): Use update_max_bb_count..
7682 (tree_function_versioning): Use update_max_bb_count..
7683 * value-prof.c (gimple_value_profile_transformations):
7684 Do not use update_max_bb_count.
7686 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7688 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7689 always use frequencies.
7691 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7693 * bb-reorder.c: Remove frequencies from comments.
7694 (better_edge_p): Use profile counts.
7695 (find_traces): Dump profile counts.
7696 (rotate_loop): Use profile counts.
7697 (find_traces_1_round): Likewise.
7698 (connect_better_edge_p): Use counts instead of probabilities for
7700 (copy_bb_p): Drop early check for non-0 frequency.
7701 (sanitize_hot_paths): Update comments.
7703 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7705 * ipa-split.c (struct split_point): Add count.
7706 (consider_split): Do not compute incoming frequency; compute incoming
7707 count and store it to split_point.
7708 (split_function): Set count of the call to split part correctly.
7710 2017-11-13 Carl Love <cel@us.ibm.com>
7712 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7714 2017-11-13 Tom Tromey <tom@tromey.com>
7716 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7718 2017-11-13 Carl Love <cel@us.ibm.com>
7720 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7721 Add support for builtins:
7722 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7723 vector {un,}signed {char,int,short},
7724 vector {un,}signed {char,int,short}) arguments.
7725 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7726 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7727 Add BU_P9V_AV_2 expansions for the builtins.
7728 * config/rs6000/altivec.h (vec_first_match_index,
7729 vec_first_mismatch_index, vec_first_match_or_eos_index,
7730 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7731 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7732 new extern declaration.
7733 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7734 * config/rs6000/vsx.md (first_match_index_<mode>,
7735 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7736 first_mismatch_or_eos_index_<mode>): Add define expand.
7737 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7738 * doc/extend.texi: Update the built-in documenation file for the new
7741 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7743 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7744 into the min/max operations for _Float<N> and _Float<N>X types.
7746 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7749 * dwarf2out.c (do_eh_frame): New static variable.
7750 (dwarf2out_begin_prologue): Set it.
7751 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7753 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7755 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7757 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7759 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7762 * debug.h (dwarf2out_do_eh_frame): Declare.
7763 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7764 (dwarf2out_do_frame): Use it.
7765 (dwarf2out_do_cfi_asm): Likewise.
7766 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7767 (dwarf2out_assembly_start): Likewise.
7768 (dwarf2out_begin_prologue): Fix comment.
7769 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7770 if the target needs either debug or unwind DWARF2 info.
7771 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7772 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7773 -ftrapping-math, -ftrapv and -fwrapv.
7775 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7777 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7778 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7779 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7780 (estimate_edge_size_and_time): Likewise.
7781 (ipa_merge_fn_summary_after_inlining): Likewise.
7782 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7783 (compute_uninlined_call_time): Use sreal_frequency.
7784 (compute_inlined_call_time): Likewise.
7785 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7786 * profile-count.c: Include sreal.h.
7787 (profile_count::to_sreal_scale): New.
7788 * profile-count.h: Forward declare sreal.
7789 (profile_count::to_sreal_scale): Declare.
7791 2017-11-13 Nathan Sidwell <nathan@acm.org>
7793 * diagnostic.c (maybe_line_and_column): New.
7794 (diagnostic_get_location_text): Use it.
7795 (diagnostic_report_current_module): Likewise.
7796 (test_diagnostic_get_location_text): Add tests.
7798 2017-11-13 Luis Machado <luis.machado@linaro.org>
7800 * doc/md.texi (Specifying processor pipeline description): Fix
7801 incorrect latency for the div instruction example.
7803 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7805 PR tree-optimization/78821
7806 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7807 that bit_not_p is the same.
7808 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7809 (split_group): Count precisely bit_not_p bits in each statement.
7810 (invert_op): New function.
7811 (imm_store_chain_info::output_merged_store): Use invert_op to
7812 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7813 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7815 2017-11-13 Martin Liska <mliska@suse.cz>
7817 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7818 (struct source_info): Likewise.
7819 (add_branch_counts): Likewise.
7820 (add_line_counts): Likewise.
7821 (function_summary): Likewise.
7822 (output_intermediate_line): Likewise.
7823 (generate_results): Likewise.
7825 2017-11-13 Martin Liska <mliska@suse.cz>
7827 * gcov.c (struct block_info): Remove typedef for block_t.
7828 (struct line_info): Likewise.
7829 (line_info::has_block): Likewise.
7830 (EXIT_BLOCK): Likewise.
7831 (unblock): Likewise.
7832 (circuit): Likewise.
7833 (get_cycles_count): Likewise.
7834 (process_file): Likewise.
7835 (read_graph_file): Likewise.
7836 (solve_flow_graph): Likewise.
7837 (find_exception_blocks): Likewise.
7838 (add_line_counts): Likewise.
7839 (accumulate_line_info): Likewise.
7840 (output_line_details): Likewise.
7842 2017-11-13 Martin Liska <mliska@suse.cz>
7844 * gcov.c (struct arc_info): Remove typedef for arc_t.
7845 (struct line_info): Likewise.
7846 (add_branch_counts): Likewise.
7847 (output_branch_count): Likewise.
7848 (function_info::~function_info): Likewise.
7849 (circuit): Likewise.
7850 (output_intermediate_line): Likewise.
7851 (read_graph_file): Likewise.
7852 (solve_flow_graph): Likewise.
7853 (find_exception_blocks): Likewise.
7854 (add_line_counts): Likewise.
7855 (accumulate_line_info): Likewise.
7856 (output_line_details): Likewise.
7857 (output_function_details): Likewise.
7859 2017-11-13 Martin Liska <mliska@suse.cz>
7861 * gcov.c (struct function_info): Remove typedef for function_t.
7862 (struct source_info): Likewise.
7863 (source_info::get_functions_at_location): Likewise.
7864 (solve_flow_graph): Likewise.
7865 (find_exception_blocks): Likewise.
7866 (add_line_counts): Likewise.
7867 (output_intermediate_file): Likewise.
7868 (process_file): Likewise.
7869 (generate_results): Likewise.
7870 (release_structures): Likewise.
7871 (read_graph_file): Likewise.
7872 (read_count_file): Likewise.
7873 (accumulate_line_counts): Likewise.
7874 (output_lines): Likewise.
7876 2017-11-13 Martin Liska <mliska@suse.cz>
7878 * gcov.c (function_info::function_info): Remove num_counts
7879 and add vector<gcov_type>.
7880 (function_info::~function_info): Use the vector.
7881 (process_file): Likewise.
7882 (read_graph_file): Likewise.
7883 (read_count_file): Likewise.
7884 (solve_flow_graph): Likewise.
7886 2017-11-13 Martin Liska <mliska@suse.cz>
7888 * gcov.c (function_info::is_artificial): New function.
7889 (process_file): Erase all artificial early.
7890 (generate_results): Skip as all artificial are already
7893 2017-11-13 Martin Liska <mliska@suse.cz>
7895 * gcov.c (read_graph_file): Store to global vector of functions.
7896 (read_count_file): Iterate the vector.
7897 (process_file): Likewise.
7898 (generate_results): Likewise.
7899 (release_structures): Likewise.
7901 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7903 PR tree-optimization/82954
7904 * gimple-ssa-store-merging.c
7905 (imm_store_chain_info::coalesce_immediate_stores): If
7906 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7908 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7910 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7911 Upddate call to ENDIAN_LANE_N.
7912 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7913 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7914 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7915 and use aarch64_endian_lane_rtx.
7916 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7918 2017-11-12 Tom de Vries <tom@codesourcery.com>
7920 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7922 2017-11-12 Tom de Vries <tom@codesourcery.com>
7924 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7926 2017-11-12 Tom de Vries <tom@codesourcery.com>
7928 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7930 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7931 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7932 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7934 2017-11-11 Martin Sebor <msebor@redhat.com>
7937 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7940 * prefic.c (translate_name): Replace strncpy with memcpy to
7941 avoid -Wstringop-truncation.
7943 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7945 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7947 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7949 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7951 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7953 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7956 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7958 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7959 enable generating XXBRH if the value is in a vector register.
7960 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7961 value is in a vector register.
7962 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7963 register to register bswap64's instead of doing the GPR sequence
7964 used on previous machines.
7965 (bswapdi2_xxbrd): New insn.
7966 (bswapdi2_reg): Disallow on ISA 3.0.
7967 (register to register bswap64 splitter): Do not split the insn on
7968 ISA 3.0 systems that use XXBRD.
7970 2017-11-10 Martin Sebor <msebor@redhat.com>
7973 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7975 (find_subframework_file): Same.
7977 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7979 * auto-profile.c (afdo_indirect_call): Drop frequency.
7980 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7981 (cgraph_node::create_edge): Drop frequency argument.
7982 (cgraph_node::create_indirect_edge): Drop frequency argument.
7983 (cgraph_edge::make_speculative): Drop frequency arguments.
7984 (cgraph_edge::resolve_speculation): Do not update frequencies
7985 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7986 (cgraph_node::dump): Check consistency in IPA mode.
7987 (cgraph_edge::maybe_hot_p): Use IPA counter.
7988 (cgraph_edge::verify_count_and_frequency): Rename to ...
7989 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7990 (cgraph_node::verify_node): Update.
7991 * cgraph.h (struct cgraph_edge): Drop frequency.
7992 (cgraph_edge::frequency): New function.
7993 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7995 (cgraph_edge::rebuild_edges): Likewise.
7996 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7997 (duplicate_thunk_for_node): Do not pass frequency.
7998 (cgraph_node::create_clone): Scale only counts.
7999 (cgraph_node::create_virtual_clone): Do not pass frequency.
8000 (cgraph_node::create_edge_including_clones): Do not pass frequency.
8001 (cgraph_node::create_version_clone): Do not pass frequency.
8002 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
8003 (cgraph_node::expand_thunk): Do not pass frequency.
8004 (cgraph_node::create_wrapper): Do not pass frequency.
8005 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
8007 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
8008 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
8009 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
8010 (gather_caller_stats): Use frequency function.
8011 (ipcp_cloning_candidate_p): Use frequency function.
8012 (ipcp_propagate_stage): Use frequency function.
8013 (get_info_about_necessary_edges): Use frequency function.
8014 (update_profiling_info): Update only IPA profile.
8015 (update_specialized_profile): Use frequency functoin.
8016 (perhaps_add_new_callers): Update only IPA profile.
8017 * ipa-devirt.c (ipa_devirt): Use IPA profile.
8018 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
8019 (dump_ipa_call_summary): Use frequency function.
8020 (estimate_edge_size_and_time): Use frequency function.
8021 (ipa_merge_fn_summary_after_inlining): Use frequency function.
8022 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
8023 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
8024 (update_noncloned_counts): ... ths one; scale counts only.
8025 (clone_inlined_nodes): Do not scale frequency.
8026 (inline_call): Do not pass frequency.
8027 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
8028 (compute_inlined_call_time): Use IPA profile.
8029 (want_inline_small_function_p): Use IPA profile.
8030 (want_inline_self_recursive_call_p): Use IPA profile.
8031 (edge_badness): Use IPA profile.
8032 (lookup_recursive_calls): Use IPA profile.
8033 (recursive_inlining): Do not pass frequency.
8034 (resolve_noninline_speculation): Do not update frequency.
8035 (inline_small_functions): Collect max of IPA profile.
8036 (dump_overall_stats): Dump IPA porfile.
8037 (dump_inline_stats): Dump IPA porfile.
8038 (ipa_inline): Collect IPA stats.
8039 * ipa-inline.h (clone_inlined_nodes): Update prototype.
8040 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
8041 (ipa_propagate_frequency): Use frequency function.
8042 (ipa_profile): Cleanup.
8043 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
8044 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
8045 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
8046 (input_node): Do not stream frequency.
8047 (input_edge): Do not stream frequency.
8048 (merge_profile_summaries): Scale only IPA profiles.
8049 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
8050 * predict.c (drop_profile): Do not recompute frequency.
8051 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
8052 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
8053 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
8054 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
8055 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
8056 * tree-inline.c (copy_bb): Do not scale frequency.
8057 (expand_call_inline): Do not scale frequency.
8058 (tree_function_versioning): Do not scale frequency.
8059 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8061 2017-11-10 Julia Koval <julia.koval@intel.com>
8063 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8064 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8065 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8066 (_mm256_maskz_gf2p8affine_epi64_epi8)
8067 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8068 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8069 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8070 (__builtin_ia32_vgf2p8affineqb_v32qi)
8071 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8072 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8074 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8078 (arm_option_override): Refactor.
8079 (arm_option_reconfigure_globals): New.
8080 (arm_options_perform_arch_sanity_checks): New.
8081 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8083 (arm_options_perform_arch_sanity_checks): Likewise
8085 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8087 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8088 (power9-qpmul): New.
8089 * rs6000/rs6000.md ("type" attr): Add qmul.
8090 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8091 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8092 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8094 2017-11-10 Martin Sebor <msebor@redhat.com>
8097 * builtins.c (compute_objsize): Handle arrays that
8098 compute_builtin_object_size likes to fail for. Make extern.
8099 * builtins.h (compute_objsize): Declare.
8100 (check_strncpy_sizes): New function.
8101 (expand_builtin_strncpy): Call check_strncpy_sizes.
8102 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8103 -Wstringop-truncation.
8104 (gimple_fold_builtin_strncat): Same.
8105 * gimple.c (gimple_build_call_from_tree): Set call location.
8106 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8107 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8108 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8109 (handle_builtin_strlen): Use strlen_to_stridx.
8110 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8112 Use strlen_to_stridx.
8113 (pass_strlen::execute): Release strlen_to_stridx.
8114 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8115 (-Wstringop-truncation): Document new option.
8117 2017-11-10 Martin Liska <mliska@suse.cz>
8119 PR gcov-profile/82702
8120 * gcov.c (main): Handle intermediate files in a different
8122 (get_gcov_intermediate_filename): New function.
8123 (output_gcov_file): Remove support of intermediate files.
8124 (generate_results): Allocate intermediate file.
8125 (release_structures): Clean-up properly fn_end.
8126 (output_intermediate_file): Start iterating with line 1.
8128 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8130 PR tree-optimization/82929
8131 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8132 ops_swapped_p non-static data member.
8133 (store_immediate_info::store_immediate_info): Clear it.
8134 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8135 ops set ops_swapped_p.
8136 (count_multiple_uses): Handle ops_swapped_p.
8138 2017-11-10 Martin Liska <mliska@suse.cz>
8140 * coverage.c (coverage_init): Stream information about
8141 support of has_unexecuted_blocks.
8142 * doc/gcov.texi: Document that.
8143 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8144 * gcov.c (read_graph_file): Likewise.
8145 (output_line_beginning): Fix a small issue with
8148 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8150 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8151 reference of trivial component.
8153 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8156 * gimple-ssa-store-merging.c
8157 (pass_store_merging::terminate_all_aliasing_chains): For
8158 gimple_store_p stmts also call refs_output_dependent_p.
8160 PR rtl-optimization/82913
8161 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8164 2017-11-09 Jeff Law <law@redhat.com>
8166 * vr-values.h: New file with vr_values class.
8167 * tree-vrp.c: Include vr-values.h
8168 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8169 data objects into the vr_values class.
8170 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8171 (get_value_range): Make it a member function within vr_values class.
8172 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8173 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8174 (op_with_constant_singleton_value_range): Likewise.
8175 (extract_range_for_var_from_comparison_expr): Likewise.
8176 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8177 (extract_range_from_binary_expr): Likewise.
8178 (extract_range_from_unary_expr): Likewise.
8179 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8180 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8181 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8182 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8183 (compare_name_with_value, compare_names): Likewise.
8184 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8185 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8186 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8187 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8188 (extract_range_from_phi_node): Likewise.
8189 (simplify_truth_ops_using_ranges): Likewise.
8190 (simplify_div_or_mod_using_ranges): Likewise.
8191 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8192 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8193 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8194 (simplify_float_conversion_using_ranges): Likewise.
8195 (simplify_internal_call_using_ranges): Likewise.
8196 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8197 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8198 poitner via x_vr_values for calls into gimple folder.
8199 (vrp_initialize_lattice): Make this the vr_values ctor.
8200 (vrp_free_lattice): Make this the vr_values dtor.
8201 (set_vr_value): New function.
8202 (class vrp_prop): Add vr_values data member. Add various member
8203 functions as well as member functions that delegate to vr_values.
8204 (check_array_ref): Make a member function within vrp_prop class.
8205 (search_for_addr_array, vrp_initialize): Likewise.
8206 (vrp_finalize): Likewise. Revamp to avoid direct access to
8207 vr_value, values_propagated, etc.
8208 (check_array_bounds): Extract vrp_prop class instance pointer from
8209 walk info structure. Use it to call member functions.
8210 (check_all_array_refs): Make a member function within vrp_prop class.
8211 Smuggle class instance pointer via walk info structure.
8212 (x_vr_values): New local static.
8213 (vrp_valueize): Use x_vr_values to get class instance.
8214 (vr_valueize_1): Likewise.
8215 (class vrp_folder): Add vr_values data member. Add various member
8216 functions as well as member functions that delegate to vr_values.
8217 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8218 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8219 class instance from vr_values. Use it to call member functions.
8220 (vrp_dom_walker): Add vr_values data member.
8221 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8222 instance via x_vr_values.
8223 (identify_jump_threads): Accept vr_values as argument. Store
8224 it into the walker structure.
8225 (evrp_dom_walker): Add vr_values class data member. Add various
8227 (evrp_dom_walker::try_find_new_range): Use vr_values data
8228 member to access the memory allocator.
8229 (evrp_dom_walker::before_dom_children): Store vr_values class
8230 instance into the vrp_folder class.
8231 (evrp_dom_walker::push_value_range): Rework to avoid direct
8232 access to num_vr_values and vr_value.
8233 (evrp_dom_walker::pop_value_range): Likewise.
8234 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8235 Use vr_values to get to dump_all_value_ranges member function.
8236 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8237 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8238 Pass vr_values class instance down to identify_jump_threads.
8239 Remove call to vrp_free_lattice.
8240 (debug_all_value_ranges): Remove.
8242 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8243 (vrp_folder): Likewise.
8245 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8246 Get it from the existing bitmap instead.
8247 (vrp_intersect_ranges_1): Likewise.
8249 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8251 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8253 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8254 set bit_not_p to it.
8255 (imm_store_chain_info::coalesce_immediate_stores): Break group
8256 if bit_not_p is different.
8257 (count_multiple_uses, split_group,
8258 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8259 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8260 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8261 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8262 to store_immediate_info ctor.
8264 2017-11-09 Jim Wilson <jimw@sifive.com>
8266 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8267 (scan_prog_file): Likewise.
8269 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8271 * bb-reorder.c (max_entry_frequency): Remove.
8272 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8273 connect_traces, push_to_next_round_p): Remove prototypes.
8274 (find_traces_1_round): Use counts only.
8275 (push_to_next_round_p): Likewise.
8276 (find_traces): Likewise.
8277 (rotate_loop): Likewise.
8278 (find_traces_1_round): Likewise.
8279 (connect_traces): Likewise.
8280 (edge_order): Likewise.
8282 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8284 * config/arm/arm.c (output_return_instruction): Add comments to
8285 indicate requirement for cmse_nonsecure_entry return to account
8286 for the size of clearing instruction output here.
8287 (thumb_exit): Likewise.
8288 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8289 return in hardfloat mode.
8291 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8293 * config/rs6000/rs6000.c (machine_function): Add a bool,
8294 "toc_is_wrapped_separately".
8295 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8296 if it wasn't explicitly set or unset, we are optimizing for speed, and
8297 doing separate shrink-wrapping.
8298 (rs6000_get_separate_components): Enable the TOC component if
8299 saving the TOC register in the prologue.
8300 (rs6000_components_for_bb): Handle the TOC component.
8301 (rs6000_emit_prologue_components): Store the TOC register where needed.
8302 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8303 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8305 2017-11-09 Martin Jambor <mjambor@suse.cz>
8307 * ipa-param-manipulation.c: New file.
8308 * ipa-param-manipulation.h: Likewise.
8309 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8310 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8311 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8312 (ipa_parm_adjustment): Likewise.
8313 (ipa_parm_adjustment_vec): Likewise.
8314 (ipa_get_vector_of_formal_parms): Moved declaration to
8315 ipa-param-manipulation.h.
8316 (ipa_get_vector_of_formal_parm_types): Likewise.
8317 (ipa_modify_formal_parameters): Likewise.
8318 (ipa_modify_call_arguments): Likewise.
8319 (ipa_combine_adjustments): Likewise.
8320 (ipa_dump_param_adjustments): Likewise.
8321 (ipa_modify_expr): Likewise.
8322 (ipa_get_adjustment_candidate): Likewise.
8323 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8324 ipa-param-manipulation.c.
8325 (ipa_get_vector_of_formal_parm_types): Likewise.
8326 (ipa_modify_formal_parameters): Likewise.
8327 (ipa_modify_call_arguments): Likewise.
8328 (ipa_modify_expr): Likewise.
8329 (get_ssa_base_param): Likewise.
8330 (ipa_get_adjustment_candidate): Likewise.
8331 (index_in_adjustments_multiple_times_p): Likewise.
8332 (ipa_combine_adjustments): Likewise.
8333 (ipa_dump_param_adjustments): Likewise.
8334 * tree-sra.c: Also include ipa-param-manipulation.h
8335 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8338 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8339 Alan Hayward <alan.hayward@arm.com>
8340 David Sherwood <david.sherwood@arm.com>
8342 * doc/sourcebuild.texi (vect_masked_store): Document.
8344 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8345 Alan Hayward <alan.hayward@arm.com>
8346 David Sherwood <david.sherwood@arm.com>
8348 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8350 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8351 Alan Hayward <alan.hayward@arm.com>
8352 David Sherwood <david.sherwood@arm.com>
8354 * doc/sourcebuild.texi (vect_variable_length): Document.
8356 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8357 Alan Hayward <alan.hayward@arm.com>
8358 David Sherwood <david.sherwood@arm.com>
8360 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8362 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8363 Alan Hayward <alan.hayward@arm.com>
8364 David Sherwood <david.sherwood@arm.com>
8366 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8368 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8369 Alan Hayward <alan.hayward@arm.com>
8370 David Sherwood <david.sherwood@arm.com>
8372 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8373 previously undocumented selectors.
8374 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8376 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8378 * doc/rtl.texi (const_vector): Say that elements can be
8379 const_wide_ints too.
8380 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8381 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8382 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8383 * optabs.c (expand_vector_broadcast): Likewise.
8385 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8386 Alan Hayward <alan.hayward@arm.com>
8387 David Sherwood <david.sherwood@arm.com>
8389 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8390 scaled index even if the unscaled address was invalid.
8391 Don't increase the complexity of using a scale in that case.
8393 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8394 Alan Hayward <alan.hayward@arm.com>
8395 David Sherwood <david.sherwood@arm.com>
8397 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8398 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8399 UNITS_PER_WORD bytes.
8400 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8401 based on REGMODE_NATURAL_SIZE of the inner mode.
8402 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8403 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8404 * expmed.c (lowpart_bit_field_p): Divide the value up into
8405 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8406 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8407 whether something is likely to occupy more than one register.
8409 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8412 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8413 profile_count::adjust_for_ipa_scaling.
8414 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8415 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8417 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8419 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8421 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8422 (split_group): Add total_orig and total_new arguments, estimate the
8423 number of statements related to the store group without store merging
8424 and with store merging.
8425 (imm_store_chain_info::output_merged_store): Adjust split_group
8426 callers, punt if estimated number of statements with store merging
8427 is not smaller than estimated number of statements without it.
8429 (handled_load): Remove has_single_use checks.
8430 (pass_store_merging::process_store): Likewise.
8432 2017-11-09 Richard Biener <rguenther@suse.de>
8434 PR tree-optimization/82902
8435 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8437 2017-11-09 Martin Liska <mliska@suse.cz>
8440 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8443 2017-11-09 Martin Liska <mliska@suse.cz>
8445 PR tree-optimization/82669
8446 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8448 2017-11-09 Martin Liska <mliska@suse.cz>
8450 PR gcov-profile/48463
8451 * coverage.c (coverage_begin_function): Output also end locus
8452 of a function and information whether the function is
8454 * gcov-dump.c (tag_function): Parse and print the information.
8455 * gcov.c (INCLUDE_MAP): Add include.
8456 (INCLUDE_SET): Likewise.
8457 (struct line_info): Move earlier in the source file because
8458 of vector<line_info> in function_info structure.
8459 (line_info::line_info): Likewise.
8460 (line_info::has_block): Likewise.
8461 (struct source_info): Add new member index.
8462 (source_info::get_functions_at_location): New function.
8463 (function_info::group_line_p): New function.
8464 (output_intermediate_line): New function.
8465 (output_intermediate_file): Use the mentioned function.
8466 (struct function_start): New.
8467 (struct function_start_pair_hash): Likewise.
8468 (process_file): Add code that identifies group functions.
8469 Assign lines either to global or function scope.
8470 (generate_results): Skip artificial functions.
8471 (find_source): Assign index for each source file.
8472 (read_graph_file): Read new flag artificial and end_line.
8473 (add_line_counts): Assign it either to global of function scope.
8474 (accumulate_line_counts): Isolate core of the function to
8475 accumulate_line_info and call it for both function and global
8477 (accumulate_line_info): New function.
8478 (output_line_beginning): Fix GNU coding style.
8479 (print_source_line): New function.
8480 (output_line_details): Likewise.
8481 (output_function_details): Likewise.
8482 (output_lines): Iterate both source (global) scope and function
8484 (struct function_line_start_cmp): New class.
8485 * doc/gcov.texi: Reflect changes in documentation.
8487 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8490 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8491 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8492 and similarly for not instead of neg.
8494 2017-11-08 Andi Kleen <ak@linux.intel.com>
8496 * config/i386/i386.opt: Add -mforce-indirect-call.
8497 * config/i386/predicates.md: Check for flag_force_indirect_call.
8498 * doc/invoke.texi: Document -mforce-indirect-call
8500 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8502 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8504 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8505 config/riscv/riscv.c (predict.h): New include.
8506 (riscv_slow_unaligned_access_p): No longer static.
8507 (riscv_block_move_straight): Add require.
8508 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8510 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8513 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8514 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8515 nonimmediate_operand predicate for operand 1 instead of
8518 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8520 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8522 * config/aarch64/constraints.md (Uml): New constraint.
8523 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8526 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8528 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8529 of two vec_duplicates into a vec_concat.
8531 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8533 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8534 Simplify vec_merge of vec_duplicate and vec_concat.
8535 * config/aarch64/constraints.md (Utq): New constraint.
8536 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8539 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8541 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8542 Simplify vec_merge of vec_duplicate and const_vector.
8543 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8545 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8546 mode iterator. Update predicate on operand 1 to
8547 handle non-const_vec constants. Delete constraints.
8548 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8550 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8552 PR tree-optimization/78821
8553 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8555 (store_operand_info::store_operand_info): Initialize it to false.
8556 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8557 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8558 store in the group, and if chain_info is non-NULL, to ignore altogether
8560 (compatible_load_p): Fail if bit_not_p does not match.
8561 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8562 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8563 (pass_store_merging::process_store): Adjust
8564 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8565 call terminate_all_aliasing_chains newly when adding a store into
8566 a chain with non-NULL chain_info.
8568 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8570 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8572 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8574 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8576 (aarch64_layout_frame): Initialise emit_frame_chain.
8577 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8578 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8580 2017-11-08 Martin Liska <mliska@suse.cz>
8582 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8583 of gimple_call_internal_p.
8585 2017-11-07 Tom de Vries <tom@codesourcery.com>
8587 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8589 2017-11-07 Tom de Vries <tom@codesourcery.com>
8591 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8594 2017-11-08 Martin Liska <mliska@suse.cz>
8597 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8599 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8601 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8603 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8606 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8607 (*cmp<mode>_ccz_1): New insn with $k alternative.
8610 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8611 RTX_COMM_COMPARE as commutative as well.
8612 (ix86_binary_operator_ok): Formatting fix.
8613 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8614 *<code><mode>3<mask_name><round_saeonly_name>,
8615 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8616 *<s>mul<mode>3_highpart<mask_name>,
8617 *vec_widen_umult_even_v16si<mask_name>,
8618 *vec_widen_umult_even_v8si<mask_name>,
8619 *vec_widen_umult_even_v4si<mask_name>,
8620 *vec_widen_smult_even_v16si<mask_name>,
8621 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8622 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8623 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8624 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8625 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8626 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8627 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8628 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8629 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8630 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8631 ix86_binary_operator_ok. Formatting fixes.
8632 (*<plusminus_insn><mode>3<mask_name><round_name>,
8633 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8636 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8638 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8639 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8640 eq and ne if TARGET_ISEL.
8641 (cmp): New code_iterator.
8642 (UNS, UNSU_, UNSIK): New code_attrs.
8643 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8644 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8646 ("eq<mode>3"): ... this.
8647 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8649 ("ne<mode>3"): ... this.
8651 2017-11-07 Julia Koval <julia.koval@intel.com>
8654 * common/config/i386/i386-common.c
8655 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8656 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8657 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8658 * config/i386/i386.opt: Ditto.
8659 * config/i386/i386.c (ix86_target_string): Ditto.
8660 (ix86_option_override_internal): Ditto.
8661 (ix86_init_mpx_builtins): Move MPX to args2.
8662 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8663 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8664 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8665 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8666 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8667 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8668 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8670 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8673 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8674 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8675 (zero-extendsidi peephole2): Remove peephole.
8677 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8680 * stor-layout.c: Include attribs.h.
8681 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8682 explicit lookup of "aligned" attribute.
8684 2017-11-07 Andrew Waterman <andrew@sifive.com>
8686 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8687 (riscv_expand_block_move): Likewise.
8688 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8690 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8691 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8692 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8693 (riscv_adjust_block_mem): Likewise.
8694 (riscv_block_move_loop): Likewise.
8695 (riscv_expand_block_move): Likewise.
8696 * config/riscv/riscv.md (movmemsi): New pattern.
8698 2017-11-07 Michael Clark <michaeljclark@mac.com>
8700 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8701 (MUSL_DYNAMIC_LINKER): Likewise.
8703 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8705 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8706 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8709 2017-11-07 Richard Biener <rguenther@suse.de>
8711 * match.pd: Fix build.
8713 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8714 Jackson Woodruff <jackson.woodruff@arm.com>
8716 PR tree-optimization/71026
8717 * match.pd: Canonicalize negate in division.
8719 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8722 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8724 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8726 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8727 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8729 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8731 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8732 non-scalar integral types.
8733 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8734 (-(A-B), -(~A)): New transformations.
8736 2017-11-07 Tom de Vries <tom@codesourcery.com>
8738 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8739 semicolon after "do {} while (0)".
8740 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8741 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8742 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8743 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8744 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8746 2017-11-07 Tom de Vries <tom@codesourcery.com>
8748 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8749 after "do {} while (0)".
8750 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8751 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8752 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8753 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8754 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8756 2017-11-07 Tom de Vries <tom@codesourcery.com>
8759 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8761 (arm_rtx_costs_internal): Add missing semicolon after
8762 HANDLE_NARROW_SHIFT_ARITH call.
8764 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8766 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8767 disable isel if it was not set explicitly.
8769 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8771 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8772 (add_type_attribute) likewise.
8774 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8776 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8777 of struct ix86_frame.
8778 (ix86_initial_elimination_offset): Likewise.
8779 (ix86_expand_split_stack_prologue): Likewise.
8781 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8783 * tree-vrp.h (enum value_range_type): Update stale comment.
8785 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8786 Alan Hayward <alan.hayward@arm.com>
8787 David Sherwood <david.sherwood@arm.com>
8789 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8790 (aarch64_expand_vec_perm_const): Take the number of units too.
8791 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8792 (aarch64_expand_vec_perm_const): Likewise.
8793 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8794 (vec_perm<mode>): Update accordingly.
8796 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8797 Alan Hayward <alan.hayward@arm.com>
8798 David Sherwood <david.sherwood@arm.com>
8800 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8801 Take the number of units too.
8802 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8803 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8804 but check for a vector mode before rather than after the call.
8805 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8806 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8807 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8808 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8809 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8810 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8811 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8812 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8813 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8814 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8815 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8816 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8817 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8818 (aarch64_sqdmull2_n<mode>): Update accordingly.
8820 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8821 Alan Hayward <alan.hayward@arm.com>
8822 David Sherwood <david.sherwood@arm.com>
8824 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8825 the number of units too.
8826 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8827 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8828 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8829 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8830 (vec_store_lanesxi<mode>): Update accordingly.
8832 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8833 Alan Hayward <alan.hayward@arm.com>
8834 David Sherwood <david.sherwood@arm.com>
8836 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8837 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8838 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8839 of units rather than the mode.
8840 * config/aarch64/iterators.md (nunits): New mode attribute.
8841 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8842 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8843 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8844 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8845 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8846 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8847 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8848 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8849 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8850 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8851 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8852 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8853 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8854 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8855 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8856 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8857 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8858 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8859 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8860 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8861 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8862 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8863 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8864 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8865 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8866 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8867 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8868 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8869 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8870 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8871 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8872 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8873 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8874 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8875 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8876 (aarch64_simd_vec_setv2di): Likewise.
8878 2017-11-06 Carl Love <cel@us.ibm.com>
8880 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8882 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8883 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8885 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8887 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8888 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8889 Add power 8 macro expansions.
8890 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8891 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8892 power 8 instructions. (VSX_XXBR): Add iterator.
8894 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8896 * config/arm/arm.md (predicable_short_it): Change default to "no",
8897 improve documentation, remove uses that are identical to the default.
8898 (enabled_for_depr_it): Rename to enabled_for_short_it.
8899 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8900 Remove default uses.
8901 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8902 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8903 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8904 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8906 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8909 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8910 float128 helper macros, which are no longer used after deleting
8911 the old 'q' built-in functions, and moving the round to odd
8912 built-in functions to being special built-in functions.
8913 (BU_FLOAT128_2): Likewise.
8914 (BU_FLOAT128_1_HW): Likewise.
8915 (BU_FLOAT128_2_HW): Likewise.
8916 (BU_FLOAT128_3_HW): Likewise.
8917 (FABSQ): Delete old 'q' built-in functions.
8918 (COPYSIGNQ): Likewise.
8919 (SQRTF128_ODD): Move round to odd built-in functions to be
8920 special built-in functions, so that we can handle
8921 -mabi=ieeelongdouble.
8922 (TRUNCF128_ODD): Likewise.
8923 (ADDF128_ODD): Likewise.
8924 (SUBF128_ODD): Likewise.
8925 (MULF128_ODD): Likewise.
8926 (DIVF128_ODD): Likewise.
8927 (FMAF128_ODD): Likewise.
8928 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8929 built-in names to 'f128'.
8930 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8931 old 'q' built-in functions, as the machine independent code for
8932 'f128' built-in functions handles this.
8933 (rs6000_expand_builtin): Add expansion for float128 round to odd
8934 functions, keying off on -mabi=ieeelongdouble of whether to use
8935 the KFmode or TFmode variant.
8936 (rs6000_init_builtins): Initialize the _Float128 round to odd
8938 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8939 _Float128 'q' built-in functions are now mapped into the new
8940 'f128' built-in functions.
8942 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8944 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8945 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8947 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8950 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8951 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8952 * toplev.c: Include "ipa-fnsummary.h".
8953 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8955 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8957 PR tree-optimization/82838
8958 * gimple-ssa-store-merging.c
8959 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8960 on a separate gimple_seq which is then appended to seq.
8962 2017-11-06 Jeff Law <law@redhat.com>
8965 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8966 (get_probe_interval): New functions.
8967 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8968 (ix86_adjust_stack_and_probe): Likewise.
8969 (output_adjust_stack_and_probe): Likewise.
8970 (ix86_emit_probe_stack_range): Likewise.
8971 (ix86_expand_prologue): Likewise.
8973 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8975 PR tree-optimization/82816
8976 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8977 if the modes of the two types are the same.
8978 (convert_plusminus_to_widen): Likewise.
8980 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8982 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8984 (usadv16qi): New define_expand.
8985 (usadv8hi): New define_expand.
8987 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8990 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8992 (inline_transform): Likewise.
8993 * predict.c (counts_to_freqs): Remove useless conditional.
8994 * profile-count.h (profile_count::apply_scale): Move sanity check.
8995 * tree-inline.c (copy_bb): Always scale.
8996 (copy_cfg_body): Likewise.
8998 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
9001 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
9003 (*cmp_ite0): Add enabled_for_depr_it attribute.
9004 (*cmp_ite1): Likewise.
9006 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
9008 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
9011 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9013 * tree-vrp.c (vrp_int_const_binop): Return true on success and
9014 return the value by pointer.
9015 (extract_range_from_multiplicative_op_1): Update accordingly.
9016 Return as soon as an operation fails.
9018 2017-11-05 Tom de Vries <tom@codesourcery.com>
9021 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
9022 (DEF_SANITIZER_BUILTIN): ... here.
9023 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
9024 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
9026 2017-11-05 Tom de Vries <tom@codesourcery.com>
9029 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
9031 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
9032 ASM_OUTPUT_BEFORE_CASE_LABEL call.
9033 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
9035 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9036 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9037 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9039 2017-11-05 Tom de Vries <tom@codesourcery.com>
9042 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
9045 2017-11-04 Michael Clark <michaeljclark@mac.com>
9047 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
9048 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
9050 (*addsi3_extended): Likewise.
9051 (*addsi3_extended2): Likewise.
9052 (<optab>si3): Likewise.
9053 (<optab>di3): Likewise.
9054 (<optab><mode>3): Likewise.
9055 (<*optabe>si3_internal): Likewise.
9056 (zero_extendqi<SUPERQI:mode>2): Likewise.
9057 (*add<mode>hi3): Likewise.
9058 (*xor<mode>hi3): Likewise.
9059 (<optab>di3): Likewise.
9060 (*<optab>si3_extend): Likewise.
9061 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9062 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9063 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9065 2017-11-04 Andrew Waterman <andrew@sifive.com>
9067 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9068 TARGET_STRICT_ALIGN based upon -mtune argument.
9070 2017-11-04 Andrew Waterman <andrew@sifive.com>
9072 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9074 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9076 * config/i386/i386.c (choose_basereg): Use optional scratch
9077 register and add assertion.
9078 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9079 needed, and don't allocate stack.
9080 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9081 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9082 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9084 2017-11-03 Jeff Law <law@redhat.com>
9086 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9087 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9088 to probe at the start of a noreturn function.
9090 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9092 PR tree-optimization/78821
9093 * gimple-ssa-store-merging.c: Update the file comment.
9094 (MAX_STORE_ALIAS_CHECKS): Define.
9095 (struct store_operand_info): New type.
9096 (store_operand_info::store_operand_info): New constructor.
9097 (struct store_immediate_info): Add rhs_code and ops data members.
9098 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9099 and op1r arguments to the ctor, initialize corresponding data members.
9100 (struct merged_store_group): Add load_align_base and load_align
9102 (merged_store_group::merged_store_group): Initialize them.
9103 (merged_store_group::do_merge): Update them.
9104 (merged_store_group::apply_stores): Pick the constant for
9105 encode_tree_to_bitpos from one of the two operands, or skip
9106 encode_tree_to_bitpos if neither operand is a constant.
9107 (class pass_store_merging): Add process_store method decl. Remove
9108 bool argument from terminate_all_aliasing_chains method decl.
9109 (pass_store_merging::terminate_all_aliasing_chains): Remove
9110 var_offset_p argument and corresponding handling.
9111 (stmts_may_clobber_ref_p): New function.
9112 (compatible_load_p): New function.
9113 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9114 if there is overlap and rhs_code is not INTEGER_CST. For
9115 non-overlapping stores terminate group if rhs is not mergeable.
9116 (get_alias_type_for_stmts): Change first argument from
9117 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9118 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9119 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9121 (get_location_for_stmts): Change first argument from
9122 auto_vec<gimple *> & to vec<gimple *> &.
9123 (struct split_store): Remove orig_stmts data member, add orig_stores.
9124 (split_store::split_store): Create orig_stores rather than orig_stmts.
9125 (find_constituent_stmts): Renamed to ...
9126 (find_constituent_stores): ... this. Change second argument from
9127 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9128 to info structures rather than the statements.
9129 (split_group): Rename ALLOW_UNALIGNED argument to
9130 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9131 it. Adjust find_constituent_stores caller.
9132 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9133 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9134 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9135 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9136 (mem_valid_for_store_merging): New function.
9137 (handled_load): New function.
9138 (pass_store_merging::process_store): New method.
9139 (pass_store_merging::execute): Use process_store method. Adjust
9140 terminate_all_aliasing_chains caller.
9142 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9144 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9145 Return true for more constants, symbols and label references.
9146 (aarch64_valid_floating_const): Remove unused function.
9148 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9151 * config/aarch64/aarch64.c (aarch64_layout_frame):
9152 Undo forcing of LR at bottom of frame.
9154 2017-11-03 Jeff Law <law@redhat.com>
9157 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9158 for int_registers_saved.
9160 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9161 extracted from tree-ssa-dom.c.
9162 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9163 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9164 (record_equivalences_from_incoming_edge): Add additional argument
9165 to single_pred_edge_ignoring_loop_edges call.
9166 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9167 (uncprop_dom_walker::before_dom_children): Add additional argument
9168 to single_pred_edge_ignoring_loop_edges call.
9169 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9170 single_pred_edge_ignoring_loop_edges rather than open coding.
9171 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9173 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9175 * match.pd (-(-A)): Rewrite.
9177 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9179 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9180 (rs6000_emit_int_cmove): New declaration.
9181 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9182 (rs6000_emit_sISEL): Delete.
9183 (rs6000_emit_int_cmove): Make non-static.
9184 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9185 instead of rs6000_emit_sISEL.
9187 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9189 * asan.c (create_cond_insert_point): Maintain profile.
9190 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9192 * basic-block.h (struct basic_block_def): Remove frequency.
9193 (EDGE_FREQUENCY): Use to_frequency
9194 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9196 (find_traces): Update to use to_frequency.
9197 (find_traces_1_round): Likewise; use only IPA counts.
9198 (bb_to_key): Likewise.
9199 (connect_traces): Use IPA counts only.
9200 (copy_bb_p): Update to use to_frequency.
9201 (fix_up_crossing_landing_pad): Likewise.
9202 (sanitize_hot_paths): Likewise.
9203 * bt-load.c (basic_block_freq): Likewise.
9204 * cfg.c (init_flow): Set count_max to uninitialized.
9205 (check_bb_profile): Remove frequencies; check counts.
9206 (dump_bb_info): Do not dump frequencies.
9207 (update_bb_profile_for_threading): Update counts only.
9208 (scale_bbs_frequencies_int): Likewise.
9209 (MAX_SAFE_MULTIPLIER): Remove.
9210 (scale_bbs_frequencies_gcov_type): Update counts only.
9211 (scale_bbs_frequencies_profile_count): Update counts only.
9212 (scale_bbs_frequencies): Update counts only.
9213 * cfg.h (struct control_flow_graph): Add count-max.
9214 (update_bb_profile_for_threading): Update prototype.
9215 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9216 (find_many_sub_basic_blocks): Likewise.
9217 * cfgcleanup.c (try_forward_edges): Likewise.
9218 (try_crossjump_to_edge): Likewise.
9219 * cfgexpand.c (expand_gimple_cond): Likewise.
9220 (expand_gimple_tailcall): Likewise.
9221 (construct_init_block): Likewise.
9222 (construct_exit_block): Likewise.
9223 * cfghooks.c (verify_flow_info): Check consistency of counts.
9224 (dump_bb_for_graph): Do not dump frequencies.
9225 (split_block_1): Do not update frequencies.
9226 (split_edge): Do not update frequencies.
9227 (make_forwarder_block): Do not update frequencies.
9228 (duplicate_block): Do not update frequencies.
9229 (account_profile_record): Do not update frequencies.
9230 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9231 for global heuristics.
9232 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9233 (expected_loop_iterations_unbounded): Use counts only.
9234 * cfgloopmanip.c (scale_loop_profile): Simplify.
9235 (create_empty_loop_on_edge): Simplify
9237 (duplicate_loop_to_header_edge): Simplify
9238 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9239 (update_br_prob_note): Take care of removing note when profile
9241 (relink_block_chain): Do not dump frequency.
9242 (rtl_account_profile_record): Use to_frequency.
9243 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9244 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9245 (cgraph_update_edges_for_call_stmt_node): Likewise.
9246 (cgraph_edge::verify_count_and_frequency): Update.
9247 (cgraph_node::verify_node): Temporarily disable frequency verification.
9248 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9249 to_cgraph_frequency.
9250 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9251 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9253 (cgraph_node::expand_thunk): Update profile.
9254 * except.c (dw2_build_landing_pads): Do not update frequency.
9255 * final.c (compute_alignments): Use to_frequency.
9256 (dump_basic_block_info): Do not dump frequency.
9257 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9258 (dump_gimple_bb_header): Do not dump frequency.
9259 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9261 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9262 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9263 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9264 (init_before_recovery): Do not update frequency.
9265 (sched_create_recovery_edges): Do not update frequency.
9266 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9267 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9268 (ipa_cp_c_finalize): Set max_count to uninitialized.
9269 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9270 (param_change_prob): Use counts.
9271 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9273 * ipa-split.c (consider_split): Use to_frequency.
9274 (split_function): Use to_frequency.
9275 * ira-build.c (loop_compare_func): Likewise.
9276 (mark_loops_for_removal): Likewise.
9277 (mark_all_loops_for_removal): Likewise.
9278 * loop-doloop.c (doloop_modify): Do not update frequency.
9279 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9281 * lto-streamer-in.c (input_function): Update count_max.
9282 * omp-expand.c (expand_omp_taskreg): Update count_max.
9283 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9284 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9285 (maybe_hot_count_p): Use ipa counts only.
9286 (maybe_hot_bb_p): Simplify.
9287 (maybe_hot_edge_p): Simplify.
9288 (probably_never_executed): Do not take frequency argument.
9289 (probably_never_executed_bb_p): Do not pass frequency.
9290 (probably_never_executed_edge_p): Likewise.
9291 (combine_predictions_for_bb): Check that profile is nonzero.
9292 (propagate_freq): Do not set frequency.
9293 (drop_profile): Simplify.
9294 (counts_to_freqs): Simplify.
9295 (expensive_function_p): Use to_frequency.
9296 (propagate_unlikely_bbs_forward): Simplify.
9297 (determine_unlikely_bbs): Simplify.
9298 (estimate_bb_frequencies): Add hack to silence graphite issues.
9299 (compute_function_frequency): Use ipa counts.
9300 (pass_profile::execute): Update.
9301 (rebuild_frequencies): Use counts only.
9302 (force_edge_cold): Use counts only.
9303 * profile-count.c (profile_count::dump): Dump new count types.
9304 (profile_count::differs_from_p): Check compatiblity.
9305 (profile_count::to_frequency): New function.
9306 (profile_count::to_cgraph_frequency): New function.
9307 * profile-count.h (struct function): Declare.
9308 (enum profile_quality): Add profile_guessed_local and
9309 profile_guessed_global0.
9310 (class profile_proability): Decrease number of bits to 29;
9311 update from_reg_br_prob_note and to_reg_br_prob_note.
9312 (class profile_count: Update comment; decrease number of bits
9313 to 61. Check compatibility.
9314 (profile_count::compatible_p): New private member function.
9315 (profile_count::ipa_p): New member function.
9316 (profile_count::operator<): Handle global zero correctly.
9317 (profile_count::operator>): Handle global zero correctly.
9318 (profile_count::operator<=): Handle global zero correctly.
9319 (profile_count::operator>=): Handle global zero correctly.
9320 (profile_count::nonzero_p): New member function.
9321 (profile_count::force_nonzero): New member function.
9322 (profile_count::max): New member function.
9323 (profile_count::apply_scale): Handle IPA scalling.
9324 (profile_count::guessed_local): New member function.
9325 (profile_count::global0): New member function.
9326 (profile_count::ipa): New member function.
9327 (profile_count::to_frequency): Declare.
9328 (profile_count::to_cgraph_frequency): Declare.
9329 * profile.c (OVERLAP_BASE): Delete.
9330 (compute_frequency_overlap): Delete.
9331 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9332 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9333 * sched-ebb.c (rank): Use counts only.
9334 * shrink-wrap.c (handle_simple_exit): Use counts only.
9335 (try_shrink_wrapping): Use counts only.
9336 (place_prologue_for_one_component): Use counts only.
9337 * tracer.c (find_best_predecessor): Use to_frequency.
9338 (find_trace): Use to_frequency.
9339 (tail_duplicate): Use to_frequency.
9340 * trans-mem.c (expand_transaction): Do not update frequency.
9341 * tree-call-cdce.c: Do not update frequency.
9342 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9343 (gimple_merge_blocks): Likewise.
9344 (gimple_split_edge): Likewise.
9345 (gimple_duplicate_sese_region): Likewise.
9346 (gimple_duplicate_sese_tail): Likewise.
9347 (move_sese_region_to_fn): Likewise.
9348 (gimple_account_profile_record): Likewise.
9349 (insert_cond_bb): Likewise.
9350 * tree-complex.c (expand_complex_div_wide): Likewise.
9351 * tree-eh.c (lower_resx): Update profile.
9352 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9354 (initialize_cfun): Do not initialize frequencies
9355 (freqs_to_counts): Delete.
9356 (copy_cfg_body): Ignore count parameter.
9357 (copy_body): Update.
9358 (expand_call_inline): Update count_max.
9359 (optimize_inline_calls): Update count_max.
9360 (tree_function_versioning): Update count_max.
9361 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9362 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9364 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9365 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9366 (try_peel_loop): Likewise.
9367 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9369 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9370 (tree_transform_and_unroll_loop): Do not use frequencies
9371 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9372 Use reliable prediction only.
9373 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9374 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9375 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9376 probability scaling.
9377 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9378 not update frequency
9379 (any_remaining_duplicated_blocks): Likewise.
9380 (update_profile): Likewise.
9381 (estimated_freqs_path): Delete.
9382 (freqs_to_counts_path): Delete.
9383 (clear_counts_path): Delete.
9384 (ssa_fix_duplicate_block_edges): Likewise.
9385 (duplicate_thread_path): Likewise.
9386 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9387 * tree-tailcall.c (decrease_profile): Do not update frequency.
9388 (eliminate_tail_call): Likewise.
9389 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9390 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9391 (optimize_mask_stores): Likewise.
9392 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9393 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9394 (ubsan_expand_ptr_ifn): Update profile.
9395 * value-prof.c (gimple_ic): Simplify.
9396 * value-prof.h (gimple_ic): Update prototype.
9397 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9398 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9400 (want_inline_self_recursive_call_p): Likewise.
9401 (resolve_noninline_speculation): Only cummulate defined counts.
9402 (inline_small_functions): Use nonzero_p.
9403 (ipa_inline): Do not access freed node.
9405 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9407 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9408 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9410 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9412 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9413 non-legitimate address.
9415 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9417 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9418 (*lt0_<mode>di, *lt0_<mode>si): New.
9420 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9422 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9423 TARGET_PAIRED_FLOAT.
9425 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9426 Jim Wilson <jim.wilson@linaro.org>
9428 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9429 * config/aarch64/aarch64-tune.md: Regenerated.
9430 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9431 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9433 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9435 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9436 (arc_expand_prologue): Restore blink for millicode.
9437 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9439 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9442 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9443 gen_vec_duplicate after forcing the scalar into a register.
9445 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9447 * combine (try_combine): Print the insns input to try_combine to the
9450 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9453 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9454 Remove second argument from aarch64_process_target_attr call.
9455 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9457 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9459 (aarch64_handle_attr_arch): Remove second argument.
9460 (aarch64_handle_attr_cpu): Ditto.
9461 (aarch64_handle_attr_tune): Ditto.
9462 (aarch64_handle_attr_isa_flags): Ditto.
9463 (aarch64_process_one_target_attr): Ditto.
9464 (aarch64_process_target_attr): Ditto.
9465 (aarch64_option_valid_attribute_p): Remove second argument.
9466 on aarch64_process_target_attr call.
9468 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9470 * diagnostic.c: Include "selftest-diagnostic.h".
9471 (selftest::assert_location_text): New function.
9472 (selftest::test_diagnostic_get_location_text): New function.
9473 (selftest::diagnostic_c_tests): Call it.
9475 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9477 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9478 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9479 (class selftest::test_diagnostic_context): Move to...
9480 * selftest-diagnostic.c: New file.
9481 * selftest-diagnostic.h: New file.
9483 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9485 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9486 offset range for FT32B.
9487 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9488 * config/ft32/ft32.md: Add TARGET_NOPM.
9489 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9490 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9492 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9494 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9496 2017-11-02 Jeff Law <law@redhat.com>
9498 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9499 virtual keyword on FINAL OVERRIDE members.
9501 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9502 virtuals together. Add virtual destructor.
9503 (substitute_and_fold_engine): Similarly.
9505 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9507 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9509 2017-11-02 Richard Biener <rguenther@suse.de>
9511 PR tree-optimization/82795
9512 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9514 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9516 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9518 (gcc_GAS_CHECK_FEATURE): Remove.
9519 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9520 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9521 * configure: Regenerate.
9523 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9525 * config/arc/arc.c (hwloop_optimize): Account for empty
9528 2017-11-02 Richard Biener <rguenther@suse.de>
9531 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9532 Truncate ARRAY_REF index and element size.
9534 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9536 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9538 2017-11-01 Jeff Law <law@redhat.com>
9540 * tree-ssa-ccp.c (ccp_folder): New class derived from
9541 substitute_and_fold_engine.
9542 (ccp_folder::get_value): New member function.
9543 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9544 (ccp_fold_stmt): Remove prototype.
9545 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9546 * tree-ssa-copy.c (copy_folder): New class derived from
9547 substitute_and_fold_engine.
9548 (copy_folder::get_value): Renamed from get_value.
9549 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9550 * tree-vrp.c (vrp_folder): New class derived from
9551 substitute_and_fold_engine.
9552 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9553 (vrp_folder::get_value): New member function.
9554 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9555 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9556 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9557 provide a class interface to folder/substitute routines.
9558 (ssa_prop_fold_stmt_fn): Remove typedef.
9559 (ssa_prop_get_value_fn): Likewise.
9560 (subsitute_and_fold): Remove prototype.
9561 (replace_uses_in): Likewise.
9562 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9563 Renamed from replace_uses_in. Call the virtual member function
9564 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9565 (substitute_and_fold_dom_walker): Remove initialization of
9566 data member entries for calbacks. Add substitute_and_fold_engine
9567 member and initialize it.
9568 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9569 member functions for get_value, replace_phi_args_in c
9570 replace_uses_in, and fold_stmt calls.
9571 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9572 substitute_and_fold. Remove assert. Update ctor call.
9574 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9575 (ssa_prop_visit_phi_fn): Likewise.
9576 (class ssa_propagation_engine): New class to provide an interface
9578 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9580 (ssa_prop_visit_phi): Likewise.
9581 (ssa_propagation_engine::simulate_stmt): Moved into class.
9582 Call visit_phi/visit_stmt from the class rather than via
9583 file scoped static variables.
9584 (ssa_propagation_engine::simulate_block): Moved into class.
9585 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9586 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9587 set file scoped statics for the visit_stmt/visit_phi callbacks.
9588 * tree-complex.c (complex_propagate): New class derived from
9589 ssa_propagation_engine.
9590 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9591 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9592 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9594 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9595 ssa_propagation_engine.
9596 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9597 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9598 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9599 * tree-ssa-copy.c (copy_prop): New class derived from
9600 ssa_propagation_engine.
9601 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9602 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9603 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9604 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9605 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9606 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9607 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9609 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9611 PR rtl-optimization/82778
9612 PR rtl-optimization/82597
9613 * compare-elim.c (struct comparison): Add in_a_setter field.
9614 (find_comparison_dom_walker::before_dom_children): Remove killed
9615 bitmap and df_simulate_find_defs call, instead walk the defs.
9616 Compute last_setter and initialize in_a_setter. Merge definitions
9617 with first initialization for a few variables.
9618 (try_validate_parallel): Use insn_invalid_p instead of
9619 recog_memoized. Return insn rather than just the pattern.
9620 (try_merge_compare): Fix up comment. Don't uselessly test if
9621 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9623 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9626 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9627 Alan Hayward <alan.hayward@arm.com>
9628 David Sherwood <david.sherwood@arm.com>
9630 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9631 aarch64_hard_regno_nregs to get the number of registers
9634 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9635 Alan Hayward <alan.hayward@arm.com>
9636 David Sherwood <david.sherwood@arm.com>
9638 * config/aarch64/constraints.md (Upl): Rename to...
9640 * config/aarch64/aarch64.md
9641 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9644 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9645 Alan Hayward <alan.hayward@arm.com>
9646 David Sherwood <david.sherwood@arm.com>
9648 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9649 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9652 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9653 Alan Hayward <alan.hayward@arm.com>
9654 David Sherwood <david.sherwood@arm.com>
9656 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9657 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9658 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9660 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9661 of a single element.
9662 * config/aarch64/iterators.md: Add a comment above the permute
9663 unspecs to say that they are generated directly by
9664 aarch64_expand_vec_perm_const.
9665 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9667 2017-11-01 Nathan Sidwell <nathan@acm.org>
9669 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9671 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9673 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9674 and medany code models, and describe what they do.
9676 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9678 Revert accidental duplicate:
9680 * combine.c (can_change_dest_mode): Reject changes in
9681 REGMODE_NATURAL_SIZE.
9683 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9685 PR rtl-optimization/64682
9686 PR rtl-optimization/69567
9687 PR rtl-optimization/69737
9688 PR rtl-optimization/82683
9689 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9690 register mentioned in the note, drop the note, unless it came from I3,
9691 in which case it should go to I3 again.
9693 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9695 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9696 and return false if not.
9697 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9699 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9701 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9702 range is known to be empty.
9704 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9705 Alan Hayward <alan.hayward@arm.com>
9706 David Sherwood <david.sherwood@arm.com>
9708 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9709 and CONST_VECTOR_NUNITS instead of computing the number of units from
9710 the byte sizes of the vector and element.
9711 (simplify_binary_operation_1): Likewise.
9712 (simplify_const_binary_operation): Likewise.
9713 (simplify_ternary_operation): Likewise.
9715 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9716 Alan Hayward <alan.hayward@arm.com>
9717 David Sherwood <david.sherwood@arm.com>
9719 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9720 (int_mem_offset): ...this new function.
9721 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9722 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9723 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9726 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9727 Alan Hayward <alan.hayward@arm.com>
9728 David Sherwood <david.sherwood@arm.com>
9730 * lower-subreg.c (interesting_mode_p): New function.
9731 (compute_costs, find_decomposable_subregs, decompose_register)
9732 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9733 (resolve_clobber, dump_choices): Use it.
9735 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9736 Alan Hayward <alan.hayward@arm.com>
9737 David Sherwood <david.sherwood@arm.com>
9739 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9741 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9742 Alan Hayward <alan.hayward@arm.com>
9743 David Sherwood <david.sherwood@arm.com>
9745 * alias.c (find_base_value, find_base_term): Only process integer
9746 truncations. Check the precision rather than the size.
9748 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9749 Alan Hayward <alan.hayward@arm.com>
9750 David Sherwood <david.sherwood@arm.com>
9752 * machmode.h (is_narrower_int_mode): New function
9753 * optabs.c (expand_float, expand_fix): Use it.
9754 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9756 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9757 Alan Hayward <alan.hayward@arm.com>
9758 David Sherwood <david.sherwood@arm.com>
9760 * rtl.h (narrower_subreg_mode): New function.
9761 * ira-color.c (update_costs_from_allocno): Use it.
9763 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9764 Alan Hayward <alan.hayward@arm.com>
9765 David Sherwood <david.sherwood@arm.com>
9767 * optabs-query.h (convert_optab_p): New function, split out from...
9768 (convert_optab_handler): ...here.
9769 (widening_optab_handler): Delete.
9770 (find_widening_optab_handler): Remove permit_non_widening parameter.
9771 (find_widening_optab_handler_and_mode): Likewise. Provide an
9772 override that operates on mode class wrappers.
9773 * optabs-query.c (widening_optab_handler): Delete.
9774 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9775 parameter. Assert that the two modes are the same class and that
9776 the "from" mode is narrower than the "to" mode. Use
9777 convert_optab_handler instead of widening_optab_handler.
9778 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9779 instead of widening_optab_handler.
9780 * expr.c (expand_expr_real_2): Update calls to
9781 find_widening_optab_handler.
9782 * optabs.c (expand_widen_pattern_expr): Likewise.
9783 (expand_binop_directly): Take the insn_code as a parameter.
9784 (expand_binop): Only call find_widening_optab_handler for
9785 conversion optabs; use optab_handler otherwise. Update calls
9786 to find_widening_optab_handler and expand_binop_directly.
9787 Use convert_optab_handler instead of widening_optab_handler.
9788 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9789 find_widening_optab_handler and use scalar_mode rather than
9791 (convert_plusminus_to_widen): 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 * machmode.h (fixed_size_mode): New class.
9798 * rtl.h (get_pool_mode): Return fixed_size_mode.
9799 * gengtype.c (main): Add fixed_size_mode.
9800 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9801 (get_raw_arg_mode): Likewise.
9802 * doc/tm.texi: Regenerate.
9803 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9804 * targhooks.c (default_get_reg_raw_mode): Likewise.
9805 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9806 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9807 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9808 (msp430_get_raw_result_mode): Likewise.
9809 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9810 * dbxout.c (dbxout_parms): Require fixed-size modes.
9811 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9812 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9813 * omp-low.c (lower_oacc_reductions): Likewise.
9814 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9815 (simplify_subreg): Update accordingly.
9816 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9817 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9818 that aren't fixed-size.
9819 (get_pool_mode): Return a fixed_size_mode.
9820 (output_constant_pool_2): Take a fixed_size_mode.
9822 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9823 Alan Hayward <alan.hayward@arm.com>
9824 David Sherwood <david.sherwood@arm.com>
9826 * doc/rtl.texi (vec_series): Document.
9827 (const): Say that the operand can be a vec_series.
9828 * rtl.def (VEC_SERIES): New rtx code.
9829 * rtl.h (const_vec_series_p_1): Declare.
9830 (const_vec_series_p): New function.
9831 * emit-rtl.h (gen_const_vec_series): Declare.
9832 (gen_vec_series): Likewise.
9833 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9834 (gen_vec_series): Likewise.
9835 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9836 * simplify-rtx.c (simplify_unary_operation): Handle negations
9838 (simplify_binary_operation_series): New function.
9839 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9840 (test_vector_ops_series): New function.
9841 (test_vector_ops): Call it.
9842 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9843 gen_const_vec_series.
9844 (altivec_lvsr): Likewise.
9845 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9847 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9848 Alan Hayward <alan.hayward@arm.com>
9849 David Sherwood <david.sherwood@arm.com>
9851 * doc/rtl.texi (const): Update description of address constants.
9852 Say that vector constants are allowed too.
9853 * common.md (E, F): Use CONSTANT_P instead of checking for
9855 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9856 checking for CONST_VECTOR.
9857 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9859 * expr.c (expand_expr_real_2): Check for vector modes instead
9860 of checking for CONST_VECTOR.
9861 * rtl.h (const_vec_p): New function.
9862 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9863 (unwrap_const_vec_duplicate): Handle them here too.
9865 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9866 David Malcolm <dmalcolm@redhat.com>
9867 Alan Hayward <alan.hayward@arm.com>
9868 David Sherwood <david.sherwood@arm.com>
9870 * rtl.h (vec_duplicate_p): New function.
9871 * selftest-rtl.c (assert_rtx_eq_at): New function.
9872 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9873 (assert_rtx_eq_at): Declare.
9874 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9875 * selftest-run-tests.c (selftest::run_tests): Call it.
9876 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9877 (simplify_unary_operation_1): Recursively handle vector duplicates.
9878 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9880 (simplify_subreg): Handle subregs of vector duplicates.
9881 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9882 (selftest::simplify_rtx_c_tests): New functions.
9884 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9885 Alan Hayward <alan.hayward@arm.com>
9886 David Sherwood <david.sherwood@arm.com>
9888 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9889 (gen_vec_duplicate): Likewise.
9890 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9892 (gen_const_vector): ...here.
9893 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9894 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9895 whose elements are all equal.
9896 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9897 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9898 (simplify_relational_operation): Likewise.
9899 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9901 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9902 (aarch64_expand_vector_init): Likewise.
9903 * config/arm/arm.c (neon_vdup_constant): Likewise.
9904 (neon_expand_vector_init): Likewise.
9905 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9906 (arm_block_set_unaligned_vect): Likewise.
9907 (arm_block_set_aligned_vect): Likewise.
9908 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9909 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9910 (expand_vec_perm_even_odd_pack): Likewise.
9911 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9912 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9913 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9914 gen_const_vec_duplicate.
9915 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9916 * config/mips/mips.c (mips_gen_const_int_vector): Use
9917 gen_const_vec_duplicate.
9918 (mips_expand_vector_init): Use CONST0_RTX.
9919 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9920 (define_split): Use gen_const_vec_duplicate.
9921 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9922 (define_split): Use gen_const_vec_duplicate.
9923 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9924 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9925 * config/spu/spu.c (spu_const): Likewise.
9927 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9928 Alan Hayward <alan.hayward@arm.com>
9929 David Sherwood <david.sherwood@arm.com>
9931 * combine.c (can_change_dest_mode): Reject changes in
9932 REGMODE_NATURAL_SIZE.
9934 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9936 * configure.ac (--enable-libssp): New.
9937 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9938 trying to determine target-specific default. Adjust indentation.
9939 * configure: Regenerated.
9940 * doc/install.texi (Configuration): Expand --disable-libssp
9943 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9945 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9948 2017-10-31 Martin Jambor <mjambor@suse.cz>
9951 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9953 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9955 * auto-profile.c (autofdo_source_profile::read): Use
9956 UNKNOWN_LOCATION rather than 0.
9957 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9958 (warning_at): ...this overload.
9959 (warning_at_rich_loc_n): Rename to...
9960 (warning_n): ...this overload.
9961 (error_at_rich_loc): Rename to...
9962 (error_at): ...this overload.
9963 (pedwarn_at_rich_loc): Rename to...
9964 (pedwarn): ...this overload.
9965 (permerror_at_rich_loc): Rename to...
9966 (permerror): ...this overload.
9967 (inform_at_rich_loc): Rename to...
9968 (inform): ...this overload.
9969 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9970 (diagnostic_n_impl_richloc): Rename to...
9971 (diagnostic_n_impl): ...this rich_location *-based decl.
9972 (inform_at_rich_loc): Rename to...
9973 (inform): ...this, and add an assertion.
9974 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9975 (warning_at_rich_loc): Rename to...
9976 (warning_at): ...this, and add an assertion.
9977 (warning_at_rich_loc_n): Rename to...
9978 (warning_n): ...this, and add an assertion.
9979 (warning_n): Update location_t-based implementation for removal of
9980 location_t-based diagnostic_n_impl.
9981 (pedwarn_at_rich_loc): Rename to...
9982 (pedwarn): ...this, and add an assertion.
9983 (permerror_at_rich_loc): Rename to...
9984 (permerror): ...this, and add an assertion.
9985 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9986 (error_at_rich_loc): Rename to...
9987 (error_at): ...this, and add an assertion.
9988 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9989 (driver::do_spec_on_infiles): Likewise.
9990 * substring-locations.c (format_warning_va): Update for renaming
9991 of inform_at_rich_loc.
9993 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9995 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9996 _Float<N>X built-in functions so that the variant without the
9997 "__builtin_" prefix is only enabled for the GNU C and Objective C
9998 languages when they are in non-strict ANSI/ISO mode.
9999 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
10000 * target.def (floatn_builtin_p): Add a target hook to control
10001 whether _Float<N> and _Float<N>X built-in functions without the
10002 "__builtin_" prefix are enabled, and return true for C and
10003 Objective C in the default hook. Include langhooks.h in
10005 * targhooks.h (default_floatn_builtin_p): Likewise.
10006 * targhooks.c (default_floatn_builtin_p): Likewise.
10007 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
10008 floatn_builtin_p target hook.
10009 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
10011 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
10012 Eric Botcazou <ebotcazou@adacore.com>
10014 PR rtl-optimization/81803
10015 * lra-constraints.c (curr_insn_transform): Also reload the whole
10016 register for a strict subreg no wider than a word if this is for
10017 a WORD_REGISTER_OPERATIONS target.
10019 2017-10-31 Jason Merrill <jason@redhat.com>
10021 * gdbinit.in: Skip over inlines from timevar.h.
10023 2017-10-31 Martin Liska <mliska@suse.cz>
10025 * doc/gcov.texi: Document new option.
10026 * gcov.c (print_usage): Likewise print it.
10027 (process_args): Support the argument.
10028 (format_count): New function.
10029 (format_gcov): Use the function.
10031 2017-10-31 Martin Liska <mliska@suse.cz>
10033 * gcov.c (struct name_map): do not use typedef.
10034 Define operator== and operator<.
10035 (name_search): Remove.
10036 (name_sort): Remove.
10037 (main): Do not allocate names.
10038 (process_file): Add vertical space.
10039 (generate_results): Use std::find.
10040 (release_structures): Do not release memory.
10041 (find_source): Use std::find.
10043 2017-10-31 Martin Liska <mliska@suse.cz>
10045 * gcov.c (struct line_info): Remove it's typedef.
10046 (line_info::line_info): Add proper ctor.
10047 (line_info::has_block): Do not use a typedef.
10048 (struct source_info): Do not use typedef.
10049 (circuit): Likewise.
10050 (get_cycles_count): Likewise.
10051 (output_intermediate_file): Iterate via vector iterator.
10052 (add_line_counts): Use std::vector methods.
10053 (accumulate_line_counts): Likewise.
10054 (output_lines): Likewise.
10056 2017-10-31 Martin Liska <mliska@suse.cz>
10058 * gcov.c (struct source_info): Remove typedef.
10059 (source_info::source_info): Add proper ctor.
10060 (accumulate_line_counts): Use struct, not it's typedef.
10061 (output_gcov_file): Likewise.
10062 (output_lines): Likewise.
10063 (main): Do not allocate an array.
10064 (output_intermediate_file): Use size of vector container.
10065 (process_file): Resize the vector.
10066 (generate_results): Do not preallocate, use newly added vector
10068 (release_structures): Do not release sources.
10069 (find_source): Use vector methods.
10070 (add_line_counts): Do not use typedef.
10072 2017-10-31 Martin Liska <mliska@suse.cz>
10074 * doc/gcov.texi: Document that.
10075 * gcov.c (add_line_counts): Mark lines with a non-executed
10077 (output_line_beginning): Handle such lines.
10078 (output_lines): Pass new argument.
10079 (output_intermediate_file): Print it in intermediate format.
10081 2017-10-31 Martin Liska <mliska@suse.cz>
10083 * color-macros.h: New file.
10084 * diagnostic-color.c: Factor out color related to macros to
10086 * doc/gcov.texi: Document -k option.
10087 * gcov.c (INCLUDE_STRING): Include string.h.
10088 (print_usage): Add -k option.
10089 (process_args): Parse it.
10090 (pad_count_string): New function.
10091 (output_line_beginning): Likewise.
10092 (DEFAULT_LINE_START): New macro.
10093 (output_lines): Support color output.
10095 2017-10-31 Martin Liska <mliska@suse.cz>
10097 PR gcov-profile/82633
10098 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10099 their interaction with GCOV infrastructure.
10100 * configure.ac: Add -fkeep-{inline,static}-functions to
10102 * configure: Regenerate.
10104 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10107 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10109 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10112 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10113 into a register if it does not fit into an immediate offset field.
10115 2017-10-31 Olivier Hainque <hainque@adacore.com>
10117 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10119 2017-10-31 Julia Koval <julia.koval@intel.com>
10121 * config.gcc: Add gfniintrin.h.
10122 * config/i386/gfniintrin.h: New.
10123 * config/i386/i386-builtin-types.def
10124 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10125 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10126 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10127 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10128 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10129 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10130 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10131 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10132 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10133 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10134 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10135 * config/i386/immintrin.h: Include gfniintrin.h.
10136 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10138 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10140 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10142 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10144 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10145 (arm_ashldi3_1bit): Remove pattern.
10146 (ashrdi3): Remove shift by 1 expansion.
10147 (arm_ashrdi3_1bit): Remove pattern.
10148 (lshrdi3): Remove shift by 1 expansion.
10149 (arm_lshrdi3_1bit): Remove pattern.
10150 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10151 cost of ashldi3 by 1.
10152 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10153 (<shift>di3_neon): Likewise.
10155 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10157 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10158 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10159 and (*aarch64_simd_mov<VQ:mode>).
10160 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10161 pattern alternative.
10162 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10163 attributes to match pattern alternative.
10165 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10167 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10168 * config/rs6000/emmintrin.h: New file.
10169 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10171 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10173 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10174 * (movdi_vfp_cortexa8): Remove pattern.
10176 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10178 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10180 (Specific, alpha*-dec-osf5.1): Remove.
10181 (Specific, mips-sgi-irix5): Remove.
10182 (Specific, mips-sgi-irix6): Remove.
10184 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10186 PR middle-end/22141
10187 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10188 arguments to clear_bit_region_be.
10190 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10192 * gimplify.c: Include memmodel.h.
10194 2017-10-30 Martin Jambor <mjambor@suse.cz>
10196 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10197 condition whether loop should be executed at all.
10199 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10201 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10202 gimple folding of vec_madd() intrinsics.
10203 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10204 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10205 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10207 2017-10-30 Richard Biener <rguenther@suse.de>
10209 PR tree-optimization/82762
10211 2017-10-23 Richard Biener <rguenther@suse.de>
10213 PR tree-optimization/82129
10215 2017-08-01 Richard Biener <rguenther@suse.de>
10217 PR tree-optimization/81181
10218 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10219 (compute_antic): ... end of iteration here.
10221 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10223 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10224 -std=iso9899:2017 and -std=gnu17.
10225 * doc/standards.texi (C Language): Document C17 support.
10226 * doc/cpp.texi (Overview): Mention -std=c17.
10227 (Standard Predefined Macros): Document C11 and C17 values of
10228 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10229 * doc/extend.texi (Inline): Do not list individual options for
10230 standards newer than C99.
10231 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10233 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10236 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10238 * asan.c (asan_finish_file): Align asan globals array by shadow
10241 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10243 PR middle-end/22141
10244 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10245 (struct store_immediate_info): Add bitregion_start and bitregion_end
10247 (store_immediate_info::store_immediate_info): Add brs and bre
10248 arguments and initialize bitregion_{start,end} from those.
10249 (struct merged_store_group): Add bitregion_start, bitregion_end,
10250 align_base and mask fields. Drop unnecessary struct keyword from
10251 struct store_immediate_info. Add do_merge method.
10252 (clear_bit_region_be): Use memset instead of loop storing zeros.
10253 (merged_store_group::do_merge): New method.
10254 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10255 stores as long as the surrounding bitregions have no gaps.
10256 (merged_store_group::merge_overlapping): Use do_merge.
10257 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10258 is byte aligned, rather than requiring that start and width are
10259 byte aligned. Drop unnecessary struct keyword from
10260 struct store_immediate_info. Allocate and populate also mask array.
10261 Make start of the arrays relative to bitregion_start rather than
10262 start and size them according to bitregion_{end,start} difference.
10263 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10264 struct store_immediate_info.
10265 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10266 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10267 struct keyword from struct store_immediate_info.
10268 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10269 between stores as long as the surrounding bitregions have no gaps.
10271 (struct split_store): Add orig non-static data member.
10272 (split_store::split_store): Initialize orig to false.
10273 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10274 if there is exactly a single original stmt. Change stmts argument
10275 to pointer from reference, if NULL, don't push anything to it. Add
10276 first argument, use it to optimize skipping over orig stmts that
10277 are known to be before bitpos already. Simplify.
10278 (split_group): Return unsigned int count how many stores are or
10279 would be needed rather than a bool. Add allow_unaligned argument.
10280 Change split_stores argument from reference to pointer, if NULL,
10281 only do a dry run computing how many stores would be produced.
10282 Rewritten algorithm to use both alignment and misalign if
10283 !allow_unaligned and handle bitfield stores with gaps.
10284 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10285 from bitregion_start instead of start. Compute allow_unaligned
10286 here, if true, do 2 split_group dry runs to compute which one
10287 produces fewer stores and prefer aligned if equal. Punt if
10288 new count is bigger or equal than original before emitting any
10289 statements, rather than during that. Remove no longer needed
10290 new_ssa_names tracking. Replace num_stmts with
10291 split_stores.length (). Use 32-bit stack allocated entries
10292 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10293 if possible. Handle bitfields with gaps.
10294 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10295 Compute bitregion_{start,end} for the stores and construct
10296 store_immediate_info with that. Formatting fixes.
10298 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10301 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10302 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10304 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10306 * gimplify.c: Include tm_p.h.
10308 * common.opt (gcoff): Re-add as ignored option.
10309 (gcoff1, gcoff2, gcoff3): Likewise.
10311 * Makefile.in (OBJS): Delete sdbout.o.
10312 (GTFILES): Delete $(srcdir)/sdbout.c.
10313 * debug.h: Delete sdb_debug_hooks.
10314 * final.c: Delete sdbout.h include.
10315 (final_scan_insn): Delete SDB_DEBUG check.
10316 (rest_of_clean_state): Likewise.
10317 * output.h: Delete sdb_begin_function_line.
10318 * sdbout.c: Delete.
10319 * sdbout.h: Delete.
10320 * toplev.c: Delete sdbout.h include.
10321 (process_options): Delete SDB_DEBUG check.
10322 * tree-core.h (tree_type_common): Delete pointer field of
10324 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10325 TYPE_SYMTAB_POINTER.
10326 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10327 (TYPE_SYMTAB_IS_POINTER): Delete.
10328 (TYPE_SYMTAB_IS_DIE): Renumber.
10329 * xcoffout.c: Refer to former sdbout.c file.
10330 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10332 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10333 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10334 (Debugging Options): Delete -gcoff.
10335 (-gstabs): Delete SDB reference.
10337 (-gcoff@var{level}): Delete.
10338 * doc/passes.texi (Debugging information output): Delete SDB and
10339 sdbout.c references.
10340 * doc/tm.texi: Regenerate.
10341 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10342 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10344 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10345 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10346 (SDB_DEBUGGING_INFO): Delete.
10347 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10348 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10349 * target.def (output_source_filename): Delete COFF reference.
10351 * common.opt (gcoff): Delete.
10352 (gxcoff+): Update Negative chain.
10353 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10355 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10356 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10357 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10358 and SDB references.
10359 (expand_function_start): Change sdb reference to past tense.
10360 (expand_function_end): Change sdb reference to past tense.
10361 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10362 * opts.c (debug_type_names): Delete coff entry.
10363 (common_handle_option): Delete OPT_gcoff case.
10364 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10366 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10367 * config/cris/cris.h: Delete SDB reference in comment.
10368 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10369 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10370 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10371 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10373 (ix86_expand_prologue): Likewise.
10374 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10375 * config/ia64/ia64.h: Likewise.
10376 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10377 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10379 * config/mmix/mmix.h: Likewise.
10380 * config/nds32/nds32.c: Likewise.
10381 * config/stormy/storym16.h: Likewise.
10382 * config/visium/visium.h: Likewise.
10383 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10385 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10387 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10388 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10389 FRAME_POINTER_REGNUM point at high end of local var area.
10391 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10393 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10394 Move comment around. Do not reset best_edge for a copiable
10395 destination if the copy would cause a partition change.
10396 (better_edge_p): Remove redundant check.
10398 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10400 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10402 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10404 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10405 for math functions that have _Float<N> and _Float<N>X variants.
10406 (mathfn_built_in_2): Add support for math functions that have
10407 _Float<N> and _Float<N>X variants.
10408 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10409 (expand_builtin_mathfn_ternary): Add support for fma with
10410 _Float<N> and _Float<N>X variants.
10411 (expand_builtin): Likewise.
10412 (fold_builtin_3): Likewise.
10413 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10414 create math function _Float<N> and _Float<N>X variants as external
10416 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10417 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10418 the __builtin_ prefix and if not strict ansi, without the prefix.
10419 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10420 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10421 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10422 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10423 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10424 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10425 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10426 function signatures for fma _Float<N> and _Float<N>X variants.
10427 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10428 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10429 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10430 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10431 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10432 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10433 * gencfn-macros.c (print_case_cfn): Add support for math functions
10434 that have _Float<N> and _Float<N>X variants.
10435 (print_define_operator_list): Likewise.
10436 (fltfn_suffixes): Likewise.
10438 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10439 for math functions that have _Float<N> and _Float<N>X variants.
10440 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10441 and _Float<N>X variants.
10442 (COPYSIGN): Likewise.
10445 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10446 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10448 (integer_valued_read_call_p): Likewise.
10449 * fold-const-call.c (fold_const_call_ss): Likewise.
10450 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10451 _Float<N> and _Float<N>X variants.
10452 (fold_const_call_ssss): Add support for fma _Float<N> and
10453 _Float<N>X variants.
10454 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10455 support for copysign and fma _Float<N> and _Float<N>X variants.
10456 (backprop::process_builtin_call_use): Likewise.
10457 * tree-call-cdce.c (can_test_argument_range); Add support for
10458 sqrt _Float<N> and _Float<N>X variants.
10459 (edom_only_function): Likewise.
10460 (get_no_error_domain): Likewise.
10461 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10462 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10463 copysign _Float<N> and _Float<N>X variants.
10464 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10465 handled by machine independent code.
10466 (FMAF128): Likewise.
10467 * doc/cpp.texi (Common Predefined Macros): Document defining
10468 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10469 fma _Float<N> and _Float<N>X variants.
10471 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10474 * config/i386/i386-modes.def (CCFPU): Remove definition.
10475 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10476 (ix86_cc_modes_compatible): Ditto.
10477 (ix86_expand_carry_flag_compare): Ditto.
10478 (ix86_expand_int_movcc): Ditto.
10479 (ix86_expand_int_addcc): Ditto.
10480 (ix86_reverse_condition): Ditto.
10481 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10482 Return true/false for unordered/ordered fp comparisons.
10483 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10484 (ix86_prepare_fp_compare_args): Update for rename.
10485 (ix86_expand_fp_compare): Update for rename. Generate unordered
10486 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10487 (ix86_expand_sse_compare_and_jump): Ditto.
10488 * config/i386/predicates.md (fcmov_comparison_operator):
10489 Remove CCFPU mode handling.
10490 (ix86_comparison_operator): Ditto.
10491 (ix86_carry_flag_operator): Ditto.
10492 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10493 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10494 (*cmpu<mode>_cc_i387): Ditto.
10495 (FPCMP): Remove mode iterator.
10496 (unord): Remove mode attribute.
10497 (unord_subst): New define_subst transformation
10498 (unord): New define_subst attribute.
10499 (unordered): Ditto.
10500 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10501 (*cmpi<unord>xf_i387): Ditto.
10502 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10503 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10504 using unord_subst transformation.
10505 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10506 (round_saeonly): Also handle CCFP mode.
10507 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10508 Remove UNSPEC_SAHF unspec handling.
10510 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10512 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10514 2017-10-27 Jeff Law <law@redhat.com>
10516 * gimple-ssa-sprintf.c: Include domwalk.h.
10517 (class sprintf_dom_walker): New class, derived from dom_walker.
10518 (sprintf_dom_walker::before_dom_children): New function.
10519 (struct call_info): Moved into sprintf_dom_walker class
10520 (compute_formath_length, handle_gimple_call): Likewise.
10521 (sprintf_length::execute): Call the dominator walker rather
10522 than walking the statements.
10524 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10525 gimple statement locations.
10526 (check_array_bounds): Corresponding changes. Get the statement's
10527 location directly from wi->stmt.
10529 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10532 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10534 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10536 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10537 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10539 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10542 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10543 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10544 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10545 maybe_get_pool_constant.
10546 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10549 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10551 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10552 2.26 caveat. Update gas and gld versions.
10553 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10556 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10558 * cgraph.h (set_malloc_flag): Declare.
10559 * cgraph.c (set_malloc_flag_1): New function.
10560 (set_malloc_flag): Likewise.
10561 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10562 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10564 (read_ipa_call_summary): Add support for reading is_return_callee.
10565 (write_ipa_call_summary): Stream is_return_callee.
10566 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10567 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10568 ipa-prop.h, ipa-fnsummary.h.
10569 (pure_const_names): Change to static.
10570 (malloc_state_e): Define.
10571 (malloc_state_names): Define.
10572 (funct_state_d): Add field malloc_state.
10573 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10574 (check_retval_uses): New function.
10575 (malloc_candidate_p): Likewise.
10576 (analyze_function): Add support for malloc attribute.
10577 (pure_const_write_summary): Stream malloc_state.
10578 (pure_const_read_summary): Add support for reading malloc_state.
10579 (dump_malloc_lattice): New function.
10580 (propagate_malloc): New function.
10581 (warn_function_malloc): New function.
10582 (ipa_pure_const::execute): Call propagate_malloc and
10583 ipa_free_fn_summary.
10584 (pass_local_pure_const::execute): Add support for malloc attribute.
10585 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10586 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10588 2017-10-27 Martin Liska <mliska@suse.cz>
10590 PR gcov-profile/82457
10591 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10592 for fork-like functions to be properly instrumented.
10594 2017-10-27 Richard Biener <rguenther@suse.de>
10596 PR middle-end/81659
10597 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10598 info when we redirected EH.
10600 2017-10-26 Michael Collison <michael.collison@arm.com>
10602 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10604 (<optab>_trunchf<GPI:mode>2: New pattern.
10605 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10606 * config/aarch64/iterators.md (wv): New mode attribute.
10607 (vf, VF): New mode attributes.
10608 (vgp, VGP): New mode attributes.
10609 (s): Update attribute with SImode and DImode prefixes.
10611 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10613 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10614 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10615 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10616 (nios2_option_overide): Initialize it. Don't allow R0-relative
10617 addressing with PIC.
10618 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10619 (nios2_symbolic_constant_p): Likewise.
10620 (nios2_legitimate_address_p): Likewise.
10621 (nios2_r0rel_section_name_p): New.
10622 (nios2_symbol_ref_in_r0rel_data_p): New.
10623 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10624 (r0rel_constant_p): New.
10625 (nios2_print_operand_address): Handle r0rel_constant_p.
10626 (nios2_cdx_narrow_form_p): Likewise.
10627 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10628 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10629 (Nios II Options): Document -mr0rel-sec.
10631 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10633 * config/nios2/nios2.c: Include xregex.h.
10634 (nios2_gprel_sec_regex): New.
10635 (nios2_option_overide): Initialize it. Don't allow GP-relative
10636 addressing with PIC.
10637 (nios2_small_section_name_p): Check for regex match.
10638 * config/nios2/nios2.opt (mgprel-sec=): New option.
10639 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10640 (Nios II Options): Document -mgprel-sec.
10642 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10644 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10646 2017-10-26 Tom de Vries <tom@codesourcery.com>
10648 PR tree-optimization/82707
10649 * gimple.c (gimple_copy): Fix unsharing of
10650 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10652 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10654 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10655 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10656 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10657 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10658 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10659 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10660 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10661 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10662 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10663 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10664 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10665 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10666 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10667 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10668 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10669 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10670 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10672 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10674 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10676 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10677 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10678 warning to rs6000.c. Remove the Undocumented flag, since it has
10680 (-mabi=ibmlongdouble): Likewise.
10681 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10682 already set, set the default format for long double.
10683 (rs6000_debug_reg_global): Print whether long double is IBM or
10685 (rs6000_option_override_internal): Rework setting long double
10686 format. Only warn if the user is changing the long double default
10687 and they did not use -Wno-psabi.
10688 * doc/invoke.texi (PowerPC options): Update the documentation for
10689 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10691 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10692 Alan Hayward <alan.hayward@arm.com>
10693 David Sherwood <david.sherwood@arm.com>
10695 * rtl.h (wider_subreg_mode): New function.
10696 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10697 rather than an unsigned int *.
10698 * ira-color.c (regno_max_ref_width): Replace with...
10699 (regno_max_ref_mode): ...this new variable.
10700 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10701 Use wider_subreg_mode.
10702 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10703 rather than an unsigned int *.
10704 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10705 (process_alt_operands): Likewise.
10706 (invariant_p): Likewise.
10707 * lra-spills.c (assign_mem_slot): Likewise.
10708 (add_pseudo_to_slot): Likewise.
10709 * lra.c (collect_non_operand_hard_regs): Likewise.
10710 (add_regs_to_insn_regno_info): Likewise.
10711 * reload1.c (regno_max_ref_width): Replace with...
10712 (regno_max_ref_mode): ...this new variable.
10713 (reload): Update accordingly. Update call to
10714 ira_sort_regnos_for_alter_reg.
10715 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10716 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10717 (scan_paradoxical_subregs): Likewise.
10719 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10721 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10722 (aarch64_frame): Add emit_frame_chain boolean.
10723 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10724 Move eh_return case to aarch64_layout_frame.
10725 (aarch64_layout_frame): Initialize emit_frame_chain.
10726 (aarch64_expand_prologue): Use emit_frame_chain.
10728 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10730 * config/aarch64/aarch64.c (aarch64_layout_frame):
10731 Ensure LR is always stored at the bottom of the callee-saves.
10732 Remove rarely used frame layout which saves callee-saves at top of
10733 frame, so the store of LR can be used as a valid probe in all cases.
10735 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10737 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10738 Improve unaligned TImode/TFmode base/offset split.
10740 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10741 Alan Hayward <alan.hayward@arm.com>
10742 David Sherwood <david.sherwood@arm.com>
10744 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10745 * combine.c (find_single_use_1): Likewise.
10746 (expand_field_assignment): Likewise.
10747 (move_deaths): Likewise.
10748 * lra-constraints.c (simplify_operand_subreg): Likewise.
10749 (curr_insn_transform): Likewise.
10750 * lra.c (collect_non_operand_hard_regs): Likewise.
10751 (add_regs_to_insn_regno_info): Likewise.
10752 * rtlanal.c (reg_referenced_p): Likewise.
10753 (covers_regno_no_parallel_p): Likewise.
10755 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10757 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10758 Don't print any bits outside the precision of the value.
10759 * wide-int.cc (test_printing): Add some new tests.
10761 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10763 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10764 supports -xbrace_comment option.
10765 * configure: Regenerate.
10766 * config.in: Regenerate.
10767 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10768 (ASM_CPU_SPEC): Use it.
10770 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10772 * target.def (static_rtx_alignment): New hook.
10773 * targhooks.h (default_static_rtx_alignment): Declare.
10774 * targhooks.c (default_static_rtx_alignment): New function.
10775 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10776 * doc/tm.texi: Regenerate.
10777 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10778 instead of targetm.constant_alignment. Remove call to
10779 set_mem_attributes.
10780 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10781 (cris_preferred_mininum_alignment): New function, split out from...
10782 (cris_constant_alignment): ...here.
10783 (cris_static_rtx_alignment): New function.
10784 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10786 (ix86_constant_alignment): ...here.
10787 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10788 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10789 (mmix_static_rtx_alignment): New function.
10790 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10791 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10793 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10796 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10797 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10799 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10801 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10802 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10803 mark as a sign-extending load.
10804 (local_pic_load_u): Define.
10806 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10808 PR middle-end/82062
10809 * fold-const.c (operand_equal_for_comparison_p): Also return true
10810 if ARG0 is a simple variant of ARG1 with narrower precision.
10811 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10813 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10815 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10817 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10818 scatter_static, scatter_per_elt.
10819 * x86-tune-costs.h: Add new cost entries.
10821 2017-10-25 Richard Biener <rguenther@suse.de>
10823 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10824 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10825 class pass_fre): Move to ...
10826 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10827 class pass_fre): ... here and adjust for statistics.
10829 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10832 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10833 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10834 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10835 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10838 2017-10-25 Richard Biener <rguenther@suse.de>
10840 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10841 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10842 eliminate_push_avail, eliminate_insert): Move inside...
10843 (class eliminate_dom_walker): ... this class in preparation
10845 (fini_eliminate): Remove by merging with ...
10846 (eliminate): ... this function. Adjust for class changes.
10847 (pass_pre::execute): Remove fini_eliminate call.
10848 (pass_fre::execute): Likewise.
10850 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10853 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10854 (VPERMI2, VPERMI2I): New mode iterators.
10855 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10856 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10858 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10859 mode iterator. Remove 3 old define_insn patterns.
10860 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10861 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10862 VPERMI2 mode iterator, remove the other two expanders.
10863 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10864 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10865 instructions, remove the other two patterns.
10866 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10867 mode iterator, remove the other two patterns.
10868 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10869 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10870 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10871 and adjust argument order accordingly.
10872 (ix86_expand_vec_perm): Adjust caller.
10873 (expand_vec_perm_1): Likewise.
10874 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10875 (expand_vec_perm_vpermt2_vpshub2): ... this.
10876 (ix86_expand_vec_perm_const_1): Adjust caller.
10877 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10880 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10881 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10882 (vec_shl_<mode>): Remove unused expander.
10883 (avx512bw_<shift_insn><mode>3): New define_insn.
10884 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10885 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10887 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10890 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10893 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10895 PR rtl-optimization/82396
10896 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10897 (autopref_multipass_init): Simplify initialization.
10898 (autopref_rank_data): Simplify sort order.
10899 * gcc/sched-int.h (autopref_multipass_data_): Remove
10900 multi_mem_insn_p, min_offset and max_offset.
10902 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10904 PR middle-end/60580
10905 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10906 Check special value of flag_omit_frame_pointer.
10907 (aarch64_can_eliminate): Likewise.
10908 (aarch64_override_options_after_change_1): Simplify handling of
10909 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10911 2017-10-24 Richard Biener <rguenther@suse.de>
10913 PR tree-optimization/82697
10914 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10915 zero for conditional load and unconditional store.
10917 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10919 * doc/install.texi: Document bootstrap-cet.
10921 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10924 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10925 ENDBR instruction at function entrance if function is only
10928 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10931 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10932 patterns to better describe from which operation the CF is computed.
10933 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10934 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10935 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10936 is 0, use _0 suffixed expanders instead of emitting a comparison
10939 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10941 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10942 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10944 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10946 PR middle-end/82569
10947 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10948 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10949 * loop-iv.c (iv_get_reaching_def): Likewise.
10950 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10951 variable is promoted and the partition contains undefined values.
10953 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10955 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10957 (nios2_address_cost): Define.
10958 (nios2_legitimize_address): Recognize (exp + constant) directly.
10959 (TARGET_ADDRESS_COST): Define.
10961 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10963 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10964 (nios2_symbolic_memory_operand_p): Declare.
10965 (nios2_split_large_constant): Declare.
10966 (nios2_split_symbolic_memory_operand): Declare.
10967 * config/nios2/nios2.c: Adjust includes.
10968 (nios2_symbolic_constant_allowed): New.
10969 (nios2_symbolic_constant_p): New.
10970 (nios2_plus_symbolic_constant_p): New.
10971 (nios2_valid_addr_expr_p): Recognize addresses involving
10972 symbolic constants.
10973 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10974 (nios2_symbolic_memory_operand_p): New.
10975 (nios2_large_constant_p): New.
10976 (nios2_split_large_constant): New.
10977 (nios2_split_plus_large_constant): New.
10978 (nios2_split_symbolic_memory_operand): New.
10979 (nios2_legitimize_address): Code refactoring. Handle addresses
10980 involving symbolic constants.
10981 (nios2_emit_move_sequence): Likewise.
10982 (nios2_print_operand): Improve error output.
10983 (nios2_print_operand_address): Handle LO_SUM.
10984 (nios2_cdx_narrow_form_p): Likewise.
10985 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10986 operands involving symbolic constants.
10987 (movhi_internal, movsi_internal): Likewise.
10988 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10989 (extendhisi2, extendqi<mode>2): Likewise.
10991 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10993 * tree-pass.h (PROP_rtl_split_insns): Define.
10994 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10996 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10998 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
11000 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11003 * target.def (const_not_ok_for_debug_p): Default to
11004 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
11005 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
11006 * targhooks.c (default_const_not_ok_for_debug_p): New function.
11007 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
11008 which targetm.const_not_ok_for_debug_p returned true.
11009 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
11011 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
11013 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
11014 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
11015 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
11017 (ix86_const_not_ok_for_debug_p): New function.
11018 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
11019 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
11021 2017-10-23 David Malcolm <dmalcolm@redhat.com>
11024 * system.h: Conditionally include "unique-ptr.h" if
11025 INCLUDE_UNIQUE_PTR is defined.
11026 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
11027 of defining INCLUDE_UNIQUE_PTR before including "system.h".
11029 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
11031 * config/rl78/rl78.md: New define_expand "subdi3".
11033 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
11036 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
11037 DF_REF_INSN if DF_REF_INSN_INFO is false.
11039 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
11041 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
11042 xmm_move instead of sse_move.
11043 (sse_store_index): New function.
11044 (ix86_register_move_cost): Be more sensible about mismatch stall;
11045 model AVX moves correctly; make difference between sse->integer and
11047 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
11048 moves; make difference between SSE and AVX.
11049 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
11050 and zmm_move. Increase size of sse load and store tables;
11051 add unaligned load and store tables; add ssemmx_to_integer.
11052 * x86-tune-costs.h: Update all entries according to real
11053 move latencies from Agner Fog's manual and chip documentation.
11055 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11058 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
11059 * config/i386/constraints.md (Wf): New constraint.
11060 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11061 (cmp<dwi>_doubleword): Removed.
11062 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11063 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11064 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11065 expand with cmp<dwi>_doubleword. For LTU and GEU use
11066 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11068 * common.opt (gcolumn-info): Enable by default.
11069 * doc/invoke.texi (gcolumn-info): Document new default.
11071 2017-10-23 Richard Biener <rguenther@suse.de>
11073 PR tree-optimization/82672
11074 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11075 Fold the stmt if we propagated into it.
11077 2017-10-23 Richard Biener <rguenther@suse.de>
11079 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11080 (bitmap_remove_expr_from_set): ... this. All callers call this
11081 for non-constant values.
11082 (bitmap_set_subtract): Rename to...
11083 (bitmap_set_subtract_expressions): ... this. Adjust and
11085 (bitmap_set_contains_value): Remove superfluous check.
11086 (bitmap_set_replace_value): Inline into single caller ...
11087 (bitmap_value_replace_in_set): ... here and simplify.
11088 (dependent_clean): Merge into ...
11089 (clean): ... this using an overload. Adjust.
11090 (prune_clobbered_mems): Adjust.
11091 (compute_antic_aux): Likewise.
11092 (compute_partial_antic_aux): Likewise.
11094 2017-10-23 Richard Biener <rguenther@suse.de>
11096 PR tree-optimization/82129
11098 2017-08-01 Richard Biener <rguenther@suse.de>
11100 PR tree-optimization/81181
11101 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11102 (compute_antic): ... end of iteration here.
11104 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11106 * target.def (starting_frame_offset): New hook.
11107 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11108 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11109 * doc/tm.texi.in: Regenerate.
11110 * hooks.h (hook_hwi_void_0): Declare.
11111 * hooks.c (hook_hwi_void_0): New function.
11112 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11113 STARTING_FRAME_OFFSET.
11114 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11115 * reload1.c (reload): Likewise.
11116 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11117 instead of STARTING_FRAME_OFFSET.
11118 * function.c (try_fit_stack_local): Likewise.
11119 (assign_stack_local_1): Likewise
11120 (instantiate_virtual_regs): Likewise.
11121 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11122 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11123 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11124 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11125 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11126 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11127 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11128 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11129 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11130 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11131 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11132 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11133 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11134 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11135 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11136 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11137 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11138 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11139 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11140 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11141 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11142 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11143 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11144 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11145 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11146 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11147 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11148 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11149 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11150 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11151 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11152 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11153 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11154 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11155 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11156 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11157 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11158 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11159 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11160 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11161 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11162 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11163 return a HOST_WIDE_INT.
11164 (avr_builtin_setjmp_frame_value): Use it instead of
11165 STARTING_FRAME_OFFSET.
11166 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11167 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11168 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11170 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11171 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11172 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11173 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11174 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11175 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11176 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11177 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11178 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11179 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11180 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11181 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11183 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11184 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11185 * config/mips/mips.c (mips_compute_frame_info): Refer to
11186 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11187 (mips_starting_frame_offset): New function.
11188 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11189 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11190 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11191 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11192 and return a HOST_WIDE_INT.
11193 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11194 (mmix_initial_elimination_offset): Refer to
11195 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11196 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11197 * config/pa/pa.c (pa_starting_frame_offset): New function.
11198 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11199 (pa_expand_prologue): Likewise.
11200 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11201 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11202 !FRAME_GROWS_DOWNWARD handling to...
11203 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11204 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11205 !FRAME_GROWS_DOWNWARD handling to...
11206 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11207 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11208 !FRAME_GROWS_DOWNWARD handling to...
11209 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11210 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11212 (rs6000_starting_frame_offset): New function.
11213 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11214 !FRAME_GROWS_DOWNWARD handling to...
11215 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11216 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11217 !FRAME_GROWS_DOWNWARD handling to...
11218 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11219 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11220 !FRAME_GROWS_DOWNWARD handling to...
11221 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11222 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11223 (rs6000_starting_frame_offset): New function.
11224 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11225 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11226 * config/vax/vax.c (vax_starting_frame_offset): New function.
11227 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11228 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11229 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11230 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11231 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11232 * system.h (STARTING_FRAME_OFFSET): Poison.
11234 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11236 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11237 SCALAR_TYPE_MODE instead of TYPE_MODE.
11239 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11240 Alan Hayward <alan.hayward@arm.com>
11241 David Sherwood <david.sherwood@arm.com>
11243 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11245 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11246 Alan Hayward <alan.hayward@arm.com>
11247 David Sherwood <david.sherwood@arm.com>
11249 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11251 2017-10-23 Richard Biener <rguenther@suse.de>
11253 PR tree-optimization/82129
11254 * tree-ssa-pre.c (bitmap_set_and): Remove.
11255 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11256 canonicalizing expressions in the set to those with lowest
11257 ID rather than taking that from the first edge.
11259 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11261 * combine.c (rtx_equal_for_field_assignment_p): Use
11262 byte_lowpart_offset.
11264 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11265 Alan Hayward <alan.hayward@arm.com>
11266 David Sherwood <david.sherwood@arm.com>
11268 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11269 to a SUBREG_PROMOTED_VAR.
11271 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11272 Alan Hayward <alan.hayward@arm.com>
11273 David Sherwood <david.sherwood@arm.com>
11275 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11276 (expand_debug_source_expr): Likewise.
11277 * combine.c (combine_simplify_rtx): Likewise.
11278 * cse.c (fold_rtx): Likewise.
11279 * optabs.c (expand_float): Likewise.
11280 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11281 (simplify_binary_operation_1): Likewise.
11283 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11284 Alan Hayward <alan.hayward@arm.com>
11285 David Sherwood <david.sherwood@arm.com>
11287 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11288 (record_promoted_value): Likewise.
11289 * expr.c (expand_expr_real_2): Likewise.
11290 * ree.c (update_reg_equal_equiv_notes): Likewise.
11291 (combine_set_extension): Likewise.
11292 * rtlanal.c (low_bitmask_len): Likewise.
11293 * simplify-rtx.c (neg_const_int): Likewise.
11294 (simplify_binary_operation_1): Likewise.
11296 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11297 Alan Hayward <alan.hayward@arm.com>
11298 David Sherwood <david.sherwood@arm.com>
11300 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11301 * regcprop.c (maybe_mode_change): Likewise.
11302 * reload1.c (alter_reg): Likewise.
11304 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11306 * inchash.h (inchash::hash::add_wide_int): New function.
11307 * lto-streamer-out.c (hash_tree): Use it.
11309 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11311 * inchash.h (inchash::hash::add_wide_int): Rename to...
11312 (inchash::hash::add_hwi): ...this.
11313 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11314 (polymorphic_call_target_hasher::hash): Likewise.
11315 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11316 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11317 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11318 * lto-streamer-out.c (hash_tree): Likewise.
11319 * optc-save-gen.awk: Likewise.
11320 * tree.c (add_expr): Likewise.
11322 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11325 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11326 for ordered inequality comparisons even with TARGET_IEEE_FP.
11328 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11331 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11332 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11333 Expand with cmp<dwi>_doubleword.
11335 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11337 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11338 List CET intrinsics.
11339 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11340 specific to -fcf-protection option.
11342 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11344 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11345 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11346 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11347 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11348 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11349 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11350 (extra_objs): Add cet.o for Linux/x86 targets.
11351 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11352 * config/i386/cet.c: New file.
11353 * config/i386/cetintrin.h: Likewise.
11354 * config/i386/t-cet: Likewise.
11355 * config/i386/cpuid.h (bit_SHSTK): New.
11356 (bit_IBT): Likewise.
11357 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11358 pass IBT and SHSTK bits.
11359 * config/i386/i386-builtin-types.def
11360 (VOID_FTYPE_UNSIGNED_PVOID): New.
11361 (VOID_FTYPE_UINT64_PVOID): Likewise.
11362 * config/i386/i386-builtin.def: Add CET intrinsics.
11363 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11364 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11365 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11366 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11368 * config/i386/i386.c (rest_of_insert_endbranch): New.
11369 (pass_data_insert_endbranch): Likewise.
11370 (pass_insert_endbranch): Likewise.
11371 (make_pass_insert_endbranch): Likewise.
11372 (ix86_notrack_prefixed_insn_p): Likewise.
11373 (ix86_target_string): Add -mibt, -mshstk flags.
11374 (ix86_option_override_internal): Add flag_cf_protection
11376 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11377 (ix86_print_operand): Add 'notrack' prefix output.
11378 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11379 (ix86_expand_builtin): Expand CET intrinsics.
11380 (x86_output_mi_thunk): Add 'endbranch' instruction.
11381 * config/i386/i386.h (TARGET_IBT): New.
11382 (TARGET_IBT_P): Likewise.
11383 (TARGET_SHSTK): Likewise.
11384 (TARGET_SHSTK_P): Likewise.
11385 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11386 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11387 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11388 (builtin_setjmp_setup): New pattern.
11389 (builtin_longjmp): Likewise.
11390 (rdssp<mode>): Likewise.
11391 (incssp<mode>): Likewise.
11392 (saveprevssp): Likewise.
11393 (rstorssp): Likewise.
11394 (wrss<mode>): Likewise.
11395 (wruss<mode>): Likewise.
11396 (setssbsy): Likewise.
11397 (clrssbsy): Likewise.
11398 (nop_endbr): Likewise.
11399 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11401 * config/i386/immintrin.h: Include <cetintrin.h>.
11402 * config/i386/linux-common.h
11403 (file_end_indicate_exec_stack_and_cet): New prototype.
11404 (TARGET_ASM_FILE_END): New.
11406 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11408 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11409 latencies instead of having separate table; make difference between
11410 integer and float costs.
11411 * i386.h (processor_costs): Remove scalar_stmt_cost,
11412 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11413 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11415 * x86-tune-costs.h: Remove entries which has been removed in
11416 procesor_costs from all tables; make cond_taken_branch_cost
11417 and cond_not_taken_branch_cost COST_N_INSNS based.
11419 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11421 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11423 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11425 * config/i386/i386.md (isa): Remove fma_avx512f.
11426 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11427 <avx512>_fmadd_<mode>_mask3<round_name>,
11428 <avx512>_fmsub_<mode>_mask<round_name>,
11429 <avx512>_fmsub_<mode>_mask3<round_name>,
11430 <avx512>_fnmadd_<mode>_mask<round_name>,
11431 <avx512>_fnmadd_<mode>_mask3<round_name>,
11432 <avx512>_fnmsub_<mode>_mask<round_name>,
11433 <avx512>_fnmsub_<mode>_mask3<round_name>,
11434 <avx512>_fmaddsub_<mode>_mask<round_name>,
11435 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11436 <avx512>_fmsubadd_<mode>_mask<round_name>,
11437 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11438 (*vec_widen_umult_even_v16si<mask_name>,
11439 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11440 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11442 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11444 * extend.texi: Add 'nocf_check' documentation.
11445 * gimple.texi: Add second parameter to
11446 gimple_build_call_from_tree.
11447 * invoke.texi: Add -fcf-protection documentation.
11448 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11450 2017-10-20 Richard Biener <rguenther@suse.de>
11452 PR tree-optimization/82473
11453 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11454 the largest input type.
11456 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11458 * c-attribs.c (handle_nocf_check_attribute): New function.
11459 (c_common_attribute_table): Add 'nocf_check' handling.
11460 * gimple-parser.c: Add second argument NULL to
11461 gimple_build_call_from_tree.
11462 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11463 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11465 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11466 * common.opt: Add fcf-protection flag.
11467 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11468 * flag-types.h: Add enum cf_protection_level.
11469 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11470 Add 'nocf_check' attribute propagation to gimple call.
11471 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11472 (gimple_build_call_from_tree): Update prototype.
11473 (gimple_call_nocf_check_p): New function.
11474 (gimple_call_set_nocf_check): Likewise.
11475 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11476 * ipa-icf.c: Add nocf_check attribute in statement hash.
11477 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11478 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11479 * toplev.c (process_options): Add flag_cf_protection handling.
11481 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11483 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11485 2017-10-20 Richard Biener <rguenther@suse.de>
11487 PR tree-optimization/82603
11488 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11489 remove false predicated stores.
11491 2017-10-20 Richard Biener <rguenther@suse.de>
11493 * graphite-isl-ast-to-gimple.c
11494 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11495 Remove return value and simplify, dump copied stmt after lhs
11497 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11498 Reduce dump verbosity.
11499 (gsi_insert_earliest): Likewise.
11500 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11501 * graphite.c (print_global_statistics): Adjust dumping.
11502 (print_graphite_scop_statistics): Likewise.
11503 (print_graphite_statistics): Do not dump loops here.
11504 (graphite_transform_loops): But here.
11506 2017-10-20 Nicolas Roche <roche@adacore.com>
11508 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11509 * configure: Regenerate.
11511 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11514 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11515 functions when optimizing replace GIMPLE_RETURN stmts with
11516 calls to __builtin_unreachable ().
11519 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11520 for -fsanitize=thread link of executables.
11521 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11522 link of executables.
11525 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11526 New mode iterators.
11527 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11528 define_insns for logical vector shifts to use VI248_AVX512BW
11529 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11530 condition, useless isa and prefix attributes. Change the first
11531 2 of these define_insns to ...
11532 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11533 define_insn for avx512vl.
11534 (<shift_insn><mode>3): ... and this, new define_insn without
11535 masking for non-avx512vl.
11538 * config/i386/sse.md (*andnot<mode>3,
11539 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11540 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11541 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11542 not applied use empty suffix even for TARGET_AVX512VL.
11543 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11544 is applied, supply evex,evex or evex,evex,evex instead of just
11547 2017-10-20 Julia Koval <julia.koval@intel.com>
11549 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11550 (OPTION_MASK_ISA_GFNI_UNSET): New.
11551 (ix86_handle_option): Handle OPT_mgfni.
11552 * config/i386/cpuid.h (bit_GFNI): New.
11553 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11554 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11555 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11556 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11557 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11558 * config/i386/i386.opt: Add mgfni.
11560 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11562 * config/msp430/msp430.c (msp430_option_override): Disable
11563 -fdelete-null-pointer-checks.
11564 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11566 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11568 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11569 of x87 and SSE instructions.
11571 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11573 * asan.c (create_cond_insert_point): Do not update edge count.
11574 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11575 (afdo_propagate_circuit): Likewise.
11576 (afdo_calculate_branch_prob): Likewise.
11577 (afdo_annotate_cfg): Likewise.
11578 * basic-block.h (struct edge_def): Remove count.
11579 (edge_def::count): New accessor.
11580 * bb-reorder.c (rotate_loop): Update.
11581 (find_traces_1_round): Update.
11582 (connect_traces): Update.
11583 (sanitize_hot_paths): Update.
11584 * cfg.c (unchecked_make_edge): Update.
11585 (make_single_succ_edge): Update.
11586 (check_bb_profile): Update.
11587 (dump_edge_info): Update.
11588 (update_bb_profile_for_threading): Update.
11589 (scale_bbs_frequencies_int): Update.
11590 (scale_bbs_frequencies_gcov_type): Update.
11591 (scale_bbs_frequencies_profile_count): Update.
11592 (scale_bbs_frequencies): Update.
11593 * cfganal.c (connect_infinite_loops_to_exit): Update.
11594 * cfgbuild.c (compute_outgoing_frequencies): Update.
11595 (find_many_sub_basic_blocks): Update.
11596 * cfgcleanup.c (try_forward_edges): Update.
11597 (try_crossjump_to_edge): Update
11598 * cfgexpand.c (expand_gimple_cond): Update
11599 (expand_gimple_tailcall): Update
11600 (construct_exit_block): Update
11601 * cfghooks.c (verify_flow_info): Update
11602 (redirect_edge_succ_nodup): Update
11603 (split_edge): Update
11604 (make_forwarder_block): Update
11605 (duplicate_block): Update
11606 (account_profile_record): Update
11607 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11608 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11609 * cfgloopmanip.c (scale_loop_profile): Update.
11611 (lv_adjust_loop_entry_edge): Update.
11612 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11613 (force_nonfallthru_and_redirect): Update.
11614 (purge_dead_edges): Update.
11615 (rtl_flow_call_edges_add): Update.
11616 * cgraphunit.c (init_lowered_empty_function): Update.
11617 (cgraph_node::expand_thunk): Update.
11618 * gimple-pretty-print.c (dump_probability): Update.
11619 (dump_edge_probability): Update.
11620 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11621 * haifa-sched.c (sched_create_recovery_edges): Update.
11622 * hsa-gen.c (convert_switch_statements): Update.
11623 * ifcvt.c (dead_or_predicable): Update.
11624 * ipa-inline-transform.c (inline_transform): Update.
11625 * ipa-split.c (split_function): Update.
11626 * ipa-utils.c (ipa_merge_profiles): Update.
11627 * loop-doloop.c (add_test): Update.
11628 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11629 * lto-streamer-in.c (input_cfg): Update.
11630 (input_function): Update.
11631 * lto-streamer-out.c (output_cfg): Update.
11632 * modulo-sched.c (sms_schedule): Update.
11633 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11634 * predict.c (maybe_hot_edge_p): Update.
11635 (unlikely_executed_edge_p): Update.
11636 (probably_never_executed_edge_p): Update.
11637 (dump_prediction): Update.
11638 (drop_profile): Update.
11639 (propagate_unlikely_bbs_forward): Update.
11640 (determine_unlikely_bbs): Update.
11641 (force_edge_cold): Update.
11642 * profile.c (compute_branch_probabilities): Update.
11643 * reg-stack.c (better_edge): Update.
11644 * shrink-wrap.c (handle_simple_exit): Update.
11645 * tracer.c (better_p): Update.
11646 * trans-mem.c (expand_transaction): Update.
11647 (split_bb_make_tm_edge): Update.
11648 * tree-call-cdce.c: Update.
11649 * tree-cfg.c (gimple_find_sub_bbs): Update.
11650 (gimple_split_edge): Update.
11651 (gimple_duplicate_sese_region): Update.
11652 (gimple_duplicate_sese_tail): Update.
11653 (gimple_flow_call_edges_add): Update.
11654 (insert_cond_bb): Update.
11655 (execute_fixup_cfg): Update.
11656 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11657 * tree-complex.c (expand_complex_div_wide): Update.
11658 * tree-eh.c (lower_resx): Update.
11659 (unsplit_eh): Update.
11660 (cleanup_empty_eh_move_lp): Update.
11661 * tree-inline.c (copy_edges_for_bb): Update.
11662 (freqs_to_counts): Update.
11663 (copy_cfg_body): Update.
11664 * tree-ssa-dce.c (remove_dead_stmt): Update.
11665 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11666 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11667 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11668 (unloop_loops): Update.
11669 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11670 * tree-ssa-loop-split.c (connect_loops): Update.
11671 (split_loop): Update.
11672 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11673 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11674 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11675 * tree-ssa-reassoc.c (branch_fixup): Update.
11676 * tree-ssa-tail-merge.c (replace_block_by): Update.
11677 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11678 (compute_path_counts): Update.
11679 (update_profile): Update.
11680 (recompute_probabilities): Update.
11681 (update_joiner_offpath_counts): Update.
11682 (estimated_freqs_path): Update.
11683 (freqs_to_counts_path): Update.
11684 (clear_counts_path): Update.
11685 (ssa_fix_duplicate_block_edges): Update.
11686 (duplicate_thread_path): Update.
11687 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11688 (case_bit_test_cmp): Update.
11689 (collect_switch_conv_info): Update.
11690 (gen_inbound_check): Update.
11691 (do_jump_if_equal): Update.
11692 (emit_cmp_and_jump_insns): Update.
11693 * tree-tailcall.c (decrease_profile): Update.
11694 (eliminate_tail_call): Update.
11695 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11696 (vect_do_peeling): Update.
11697 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11698 * ubsan.c (ubsan_expand_null_ifn): Update.
11699 (ubsan_expand_ptr_ifn): Update.
11700 * value-prof.c (gimple_divmod_fixed_value): Update.
11701 (gimple_mod_pow2): Update.
11702 (gimple_mod_subtract): Update.
11703 (gimple_ic): Update.
11704 (gimple_stringop_fixed_value): Update.
11706 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11709 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11711 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11713 PR rtl-optimization/82395
11714 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11715 based on non_spilled_static_chain_regno_p.
11717 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11719 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11720 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11721 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11723 2017-10-19 Martin Sebor <msebor@redhat.com>
11725 PR tree-optimization/82596
11726 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11728 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11730 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11731 (get_mem_refs_of_builtin_call): Likewise.
11732 * builtins.c (expand_builtin_apply): Adjust call to
11733 allocate_dynamic_stack_space.
11734 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11735 the third argument to allocate_dynamic_stack_space, otherwise -1.
11736 (expand_builtin): Deal with all alloca variants.
11737 (is_inexpensive_builtin): Likewise.
11738 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11739 * calls.c (special_function_p): Deal with all alloca variants.
11740 (initialize_argument_information): Adjust call to
11741 allocate_dynamic_stack_space.
11742 (expand_call): Likewise.
11743 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11744 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11745 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11746 use it for the stack usage computation.
11747 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11748 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11749 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11750 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11751 (in_loop_p): Remove first argument and useless check.
11752 (pass_walloca::execute): Remove useless test and adjust call to above.
11753 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11754 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11755 (gimplify_call_expr): Deal with all alloca variants.
11756 * hsa-gen.c (gen_hsa_alloca): Likewise.
11757 (gen_hsa_insns_for_call): Likewise.
11758 * ipa-pure-const.c (special_builtin_state): Likewise.
11759 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11760 * tree-object-size.c (alloc_object_size): Likewise.
11761 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11762 (call_may_clobber_ref_p_1): Likewise.
11763 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11764 (ccp_fold_stmt): Likewise.
11765 (optimize_stack_restore): Likewise.
11766 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11767 (mark_all_reaching_defs_necessary_1): Likewise.
11768 (propagate_necessity): Likewise.
11769 (eliminate_unnecessary_stmts): Likewise.
11770 * tree.c (build_common_builtin_nodes): Build
11771 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11772 (build_alloca_call_expr): New function.
11773 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11774 (CASE_BUILT_IN_ALLOCA): Likewise.
11775 (build_alloca_call_expr): Declare.
11776 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11778 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11781 * dwarf2out.c (new_die_raw): New static inline function.
11782 (new_die): Use it to create the DIE.
11783 (add_AT_external_die_ref): Likewise.
11784 (clone_die): Likewise.
11785 (clone_as_declaration): Likewise.
11786 (dwarf2out_vms_debug_main_pointer): Likewise.
11787 (base_type_die): Likewise. Remove early return for corner cases.
11788 Do not call add_pubtype on the DIE here.
11789 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11790 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11791 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11792 native order exists for base types, attach the DIE manually and call
11793 add_pubtype on it. Do not equate a reverse order DIE to the type.
11795 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11797 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11798 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11800 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11802 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11804 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11805 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11806 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11807 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11808 * opts.c (sanitizer_opts): Add builtin.
11809 * ubsan.c (instrument_builtin): New function.
11810 (pass_ubsan::execute): Call it.
11811 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11812 * doc/invoke.texi: Document -fsanitize=builtin.
11814 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11815 builtins, store max (log2 (align), 0) into uchar field instead of
11816 align into uptr field.
11817 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11818 store uchar 0 field instead of uptr 0 field.
11819 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11820 instead of passing one address of struct with 2 locations pass
11821 two addresses of structs with 1 location each.
11822 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11823 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11824 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11825 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11826 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11827 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11828 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11829 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11831 2017-10-19 Martin Liska <mliska@suse.cz>
11834 * file-find.c (remove_prefix): Remove.
11835 * file-find.h (remove_prefix): Likewise.
11836 * gcc-ar.c: Remove smartness of lookup.
11838 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11840 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11841 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11842 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11844 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11847 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11848 (setcc + and to xor + setcc): New peephole2.
11850 2017-10-19 Tom de Vries <tom@codesourcery.com>
11852 * doc/sourcebuild.texi (Test Directives, Variants of
11853 dg-require-support): Add dg-require-stack-size.
11855 2017-10-19 Martin Liska <mliska@suse.cz>
11858 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11859 that have a large alignment.
11860 (gimplify_target_expr): Likewise.
11862 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11864 PR rtl-optimization/82602
11865 * ira.c (rtx_moveable_p): Return false for volatile asm.
11867 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11870 * config/i386/i386-modes.def (CCGZ): New CC mode.
11871 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11872 * config/i386/predicates.md (ix86_comparison_operator):
11874 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11875 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11876 with double-word subtraction.
11877 (put_condition_code): Handle CCGZmode.
11879 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11881 * wide-int.cc (debug (const wide_int &)): New.
11882 (debug (const wide_int *)): New.
11883 (debug (const widest_int &)): New.
11884 (debug (const widest_int *)): New.
11886 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11888 PR middle-end/82556
11889 * lra-constraints.c (curr_insn_transform): Use non-input operand
11890 instead of output one for matched reload.
11892 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11894 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11895 (tree-ssa-loop-ivopts.h): New header file.
11896 (struct builtin_info): New fields.
11897 (classify_builtin_1): Compute and record base and offset parts for
11898 memset builtin partition by calling strip_offset.
11899 (offset_cmp, fuse_memset_builtins): New functions.
11900 (finalize_partitions): Fuse adjacent memset partitions by calling
11902 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11903 Expose the interface.
11904 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11906 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11908 PR tree-optimization/82574
11909 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11910 that data reference must be executed exactly once per iteration
11911 against the outermost loop in nest.
11912 (classify_partition): Update call to above function.
11914 2017-10-18 Richard Biener <rguenther@suse.de>
11916 PR tree-optimization/82591
11917 * graphite.c (graphite_transform_loops): Move code gen message
11919 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11920 Here. Handle scop_to_isl_ast failing.
11921 (scop_to_isl_ast): Limit the number of ISL operations.
11923 2017-10-18 Richard Biener <rguenther@suse.de>
11925 * graphite-isl-ast-to-gimple.c
11926 (translate_isl_ast_to_gimple::set_rename): Simplify.
11927 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11928 (graphite_copy_stmts_from_block): ... here.
11929 (copy_bb_and_scalar_dependences): Simplify.
11930 (add_parameters_to_ivs_params): Canonicalize.
11931 (generate_entry_out_of_ssa_copies): Simplify.
11932 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11933 by passing in ISL dimension.
11934 (parameter_index_in_region_1): Rename to ...
11935 (parameter_index_in_region): ... this.
11936 (extract_affine): Adjust assert, pass down parameter index.
11937 (add_param_constraints): Use range-info when available.
11938 (build_scop_context): Adjust.
11939 * sese.c (new_sese_info): Adjust.
11940 (free_sese_info): Likewise.
11941 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11942 Remove unused typedefs.
11943 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11945 2017-10-18 Martin Liska <mliska@suse.cz>
11947 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11949 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11951 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11952 (s390_sched_init): Do not reset s390_sched_state if we entered the
11953 current basic block via a fallthru edge and all others are unlikely.
11955 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11957 * config/s390/s390.c (NUM_SIDES): New variable.
11958 (LONGRUNNING_THRESHOLD): New variable.
11959 (LATENCY_FACTOR): New variable.
11960 (s390_sched_score): Decrease score for long-running instructions on
11962 (s390_sched_variable_issue): Perform bookkeeping for long-running
11965 2017-10-18 Richard Biener <rguenther@suse.de>
11967 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11968 Simplify with removal of the parameter rename map.
11969 (set_rename): Likewise.
11970 (should_copy_to_new_region): Likewise.
11971 (graphite_copy_stmts_from_block): Likewise.
11972 (copy_bb_and_scalar_dependences): Remove initialization of
11973 unused copied_bb_map.
11974 (copy_def): Remove.
11975 (copy_internal_parameters): Likewise.
11976 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11977 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11978 Use INTEGRAL_TYPE_P.
11979 (parameter_index_in_region_1): Rename to ...
11980 (assign_parameter_index_in_region): ... this. Assert we have
11981 a parameter we handle.
11982 (scan_tree_for_params): Adjust.
11983 * sese.h (parameter_rename_map_t): Remove.
11984 (struct sese_info_t): Remove unused parameter_rename_map and
11985 copied_bb_map members.
11986 * sese.c (new_sese_info): Adjust.
11987 (free_sese_info): Likewise.
11989 2017-10-18 Martin Liska <mliska@suse.cz>
11992 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11993 on an abnormal edge.
11995 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11997 * doc/invoke.texi (ffunction-sections and fdata-sections):
12000 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12002 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
12003 the use statement can throw internally.
12005 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12007 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
12008 any RTX present on the RHS of a SET.
12009 * compare-elim.c (try_eliminate_compare): Restore comment.
12011 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12013 * langhooks.h (struct lang_hooks): Document that tree_size langhook
12014 may be also called on tcc_type nodes.
12015 * langhooks.c (lhd_tree_size): Likewise.
12017 2017-10-17 David Malcolm <dmalcolm@redhat.com>
12019 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
12020 format_warning_at_substring.
12021 (maybe_warn): Convert source_range * param to a location_t. Pass
12022 UNKNOWN_LOCATION rather than NULL to fmtwarn.
12023 (format_directive): Remove code to extract source_ranges and
12024 source_range * in favor of just a location_t.
12025 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
12027 * substring-locations.c (format_warning_va): Convert
12028 source_range * param to a location_t.
12029 (format_warning_at_substring): Likewise.
12030 * substring-locations.h (format_warning_va): Likewise.
12031 (format_warning_at_substring): Likewise.
12033 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
12035 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
12037 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
12038 and scatter/gather ops.
12040 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
12041 vec_gather_load and vec_scatter_store.
12042 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
12043 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
12044 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
12045 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
12046 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
12047 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
12049 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
12051 * reg-stack.c (compare_for_stack_reg): Add bool argument.
12052 Detect FTST instruction and handle its register pops. Only pop
12053 second operand if can_pop_second_op is true.
12054 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
12055 set can_pop_second_op to false in the compare_for_stack_reg call.
12057 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
12058 output_fp_compare for stack register operands.
12059 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
12060 instructions here. Do not emit stack register pops here. Assert
12061 that FCOMPP pops next to top stack register. Rewrite function.
12063 2017-10-17 Nathan Sidwell <nathan@acm.org>
12065 PR middle-end/82577
12066 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12067 use DECL_ASSEMBLER_NAME_RAW.
12069 PR middle-end/82546
12070 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12073 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12074 Wilco Dijkstra <wilco.dijkstra@arm.com>
12076 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12077 converstion to Pmode from the buf_addr.
12079 2017-10-17 Richard Biener <rguenther@suse.de>
12081 * graphite-dependences.c (scop_get_reads_and_writes): Change
12082 output parameters to references.
12084 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12086 PR 71026/tree-optimization
12087 * fold-const.c (distribute_real_division): Removed.
12088 (fold_binary_loc): Remove calls to distribute_real_divison.
12090 2017-10-17 Richard Biener <rguenther@suse.de>
12092 * graphite-scop-detection.c
12093 (scop_detection::stmt_has_simple_data_refs_p): Always use
12094 the full nest as region.
12095 (try_generate_gimple_bb): Likewise.
12096 * sese.c (scalar_evolution_in_region): Simplify now that
12097 SCEV can handle instantiation in regions.
12098 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12099 in the non-loop part of a function if requested.
12101 2017-10-17 Richard Biener <rguenther@suse.de>
12103 PR tree-optimization/82563
12104 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12106 (graphite_regenerate_ast_isl): Call it.
12107 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12109 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12111 PR tree-optimization/82549
12112 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12113 Formatting fixes. Instead of calling make_bit_field_ref with negative
12116 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12118 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12119 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12120 _mm_maskz_reduce_ss): New.
12121 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12122 __builtin_ia32_reducess_mask): Ditto..
12123 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12124 * config/i386/sse.md (reduces<mode>): Renamed to ...
12125 (reduces<mode><mask_scalar_name>): ... this.
12126 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12128 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12129 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12131 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12133 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12134 * selftest-run-tests.c (selftest::run_tests): Call
12135 selftest::unique_ptr_tests_cc_tests.
12136 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12137 * unique-ptr-tests.cc: New file.
12139 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12142 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12144 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12145 (make_hard_regno_born, make_hard_regno_dead): Update
12146 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12148 2017-10-16 Jeff Law <law@redhat.com>
12150 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12152 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12154 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12156 2017-10-16 Olivier Hainque <hainque@adacore.com>
12158 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12159 with_cpu if we were configured for an e500v2 target cpu name.
12161 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12163 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12164 * doc/invoke.texi: Document +nodsp as a valid extension for
12167 2017-10-16 Martin Liska <mliska@suse.cz>
12169 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12170 (test_set_range): Likewise.
12171 (test_range_functions): Rename to ...
12172 (test_bit_in_range): ... this.
12173 (sbitmap_c_tests): Add new test.
12175 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12177 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12179 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12180 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12182 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12184 * config/aarch64/aarch64-builtins.c
12185 (aarch64_types_quadopu_lane_qualifiers): New.
12186 (TYPES_QUADOPU_LANE): New.
12187 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12188 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12189 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12190 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12191 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12192 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12193 (Vdottype, DOTPROD): New.
12194 (sur): Add SDOT and UDOT.
12196 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12198 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12199 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12200 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12201 Add TARGET_DOTPROD.
12202 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12203 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12204 Enable TARGET_DOTPROD.
12205 (cortex-a75.cortex-a55): Likewise.
12206 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12208 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12210 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12211 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12213 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12215 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12216 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12217 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12218 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12219 * config/arm/types.md (neon_dot, neon_dot_q): New.
12220 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12222 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12224 * config/arm/arm.h (TARGET_DOTPROD): New.
12225 * config/arm/arm.c (arm_arch_dotprod): New.
12226 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12227 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12228 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12229 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12230 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12231 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12232 * doc/invoke.texi (armv8.2-a): Document dotprod
12234 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12236 * i386.c (ix86_vec_cost): New function.
12237 (ix86_rtx_costs): Handle vector operations better.
12238 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12239 * x86-tune-costs.h: Add new costs to all tables.
12241 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12243 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12245 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12246 divsd, sqrtss and sqrtsd
12247 * x86-tune-costs.h: Add new entries to all costs.
12248 (znver1_cost): Fix to match real instruction latencies.
12250 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12251 Michael Collison <michael.collison@arm.com>
12253 * compare-elim.c: Include emit-rtl.h.
12254 (can_merge_compare_into_arith): New function.
12255 (try_validate_parallel): Likewise.
12256 (try_merge_compare): Likewise.
12257 (try_eliminate_compare): Call the above when no previous clobber
12259 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12262 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12264 PR middle-end/62263
12265 PR middle-end/82498
12266 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12267 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12269 PR middle-end/62263
12270 PR middle-end/82498
12271 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12272 to be any operand_equal_p operands. For & (B - 1) require
12273 B to be power of 2. Recognize
12274 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12276 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12279 * optabs.c (expand_memory_blockage): Fix call of
12280 targetm.have_memory_blockage.
12282 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12285 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12286 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12287 objects to extra_objs instead of overwriting it.
12289 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12291 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12292 Use any_fp_register_operand as operand[3] predicate. Simplify
12293 equality test for operands[2] and operands[4] memory location.
12294 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12295 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12296 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12297 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12298 any_fp_register_operand as operand[1] predicate. Simplify
12299 equality test for operands[0] and operands[3] memory location.
12300 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12301 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12302 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12304 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12306 * target-insns.def: Add memory_blockage.
12307 * optabs.c (expand_memory_blockage): New function.
12308 (expand_asm_memory_barrier): Rename ...
12309 (expand_asm_memory_blockage): ... to this.
12310 (expand_mem_thread_fence): Call expand_memory_blockage
12311 instead of expand_asm_memory_barrier.
12312 (expand_mem_singnal_fence): Ditto.
12313 (expand_atomic_load): Ditto.
12314 (expand_atomic_store): Ditto.
12315 * doc/md.texi (Standard Pattern Names For Generation):
12316 Document memory_blockage instruction pattern.
12318 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12320 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12321 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12322 * config/rl78/rl78.md: New define_expand "adddi3".
12324 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12326 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12329 2017-10-13 Jeff Law <law@redhat.com>
12331 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12333 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12336 * internal-fn.c (expand_mul_overflow): If both operands have
12337 the same highpart of -1 or 0 and the topmost bit of lowpart
12338 is different, overflow is if res <= 0 rather than res < 0.
12340 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12342 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12343 TARGET_P9_VECTOR code for unaligned_load case.
12345 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12347 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12349 2017-10-13 Nathan Sidwell <nathan@acm.org>
12351 * tree-core.h (tree_contains_struct): Make bool.
12352 * tree.c (tree_contains_struct): Likewise.
12353 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12354 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12355 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12356 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12357 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12359 2017-10-13 Richard Biener <rguenther@suse.de>
12361 * graphite-isl-ast-to-gimple.c
12362 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12363 parameters and dominance check.
12364 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12365 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12366 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12367 Do not update SSA form here or do intermediate IL verification.
12368 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12369 (graphite_initialize): Remove check on the number of loops in
12370 the function and inline into graphite_transform_loops.
12371 (graphite_finalize): Inline into graphite_transform_loops.
12372 (graphite_transform_loops): Perform SSA update and IL verification
12374 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12376 2017-10-13 Richard Biener <rguenther@suse.de>
12378 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12379 graphite_expression_type_precision): Avoid global constructor
12381 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12382 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12383 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12384 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12385 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12386 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12388 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12391 * config/i386/i386.h (ix86_red_zone_size): New.
12392 * config/i386/i386.md (push peephole2s): Replace
12393 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12395 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12396 Alan Hayward <alan.hayward@arm.com>
12397 David Sherwood <david.sherwood@arm.com>
12399 * combine.c (can_change_dest_mode): Reject changes in
12400 REGMODE_NATURAL_SIZE.
12402 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12403 Alan Hayward <alan.hayward@arm.com>
12404 David Sherwood <david.sherwood@arm.com>
12406 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12407 (expand_debug_source_expr): Likewise.
12408 * combine.c (combine_simplify_rtx): Likewise.
12409 * cse.c (fold_rtx): Likewise.
12410 * fwprop.c (canonicalize_address): Likewise.
12411 * targhooks.c (default_shift_truncation_mask): Likewise.
12413 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12414 Alan Hayward <alan.hayward@arm.com>
12415 David Sherwood <david.sherwood@arm.com>
12417 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12418 (widened_mode): Likewise.
12419 (expand_unop): Likewise.
12420 * ree.c (transform_ifelse): Likewise.
12421 (merge_def_and_ext): Likewise.
12422 (combine_reaching_defs): Likewise.
12423 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12425 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12426 Alan Hayward <alan.hayward@arm.com>
12427 David Sherwood <david.sherwood@arm.com>
12429 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12430 * combine.c (gen_lowpart_for_combine): Likewise.
12431 * dwarf2out.c (rtl_for_decl_location): Likewise.
12432 * final.c (alter_subreg): Likewise.
12433 * rtlhooks.c (gen_lowpart_general): Likewise.
12434 (gen_lowpart_if_possible): Likewise.
12436 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12437 Alan Hayward <alan.hayward@arm.com>
12438 David Sherwood <david.sherwood@arm.com>
12440 * calls.c (expand_call): Use subreg_lowpart_offset.
12441 * cse.c (cse_insn): Likewise.
12442 * regcprop.c (copy_value): Likewise.
12443 (copyprop_hardreg_forward_1): Likewise.
12445 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12448 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12449 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12450 =Q constraints to +Q and into insn condition add check
12451 that operands[0] and operands[1] are equal.
12452 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12453 =Q constraints to +Q and into insn condition add check
12454 that operands[0] is equal to either operands[1] or operands[2].
12457 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12458 instead of handling MINUS_EXPR twice (once for each argument),
12459 canonicalize operand order and handle just once, use rtype where
12460 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12463 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12464 any values of __C while still being pattern recognizable as a simple
12465 rotate instruction.
12467 2017-10-13 Richard Biener <rguenther@suse.de>
12469 PR tree-optimization/82451
12471 2017-10-02 Richard Biener <rguenther@suse.de>
12473 PR tree-optimization/82355
12474 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12475 a mapping for the enclosing loop but avoid generating one for
12476 the loop tree root.
12477 (copy_bb_and_scalar_dependences): Remove premature codegen
12478 error on PHIs in blocks duplicated into multiple places.
12479 * graphite-scop-detection.c
12480 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12481 in the region use it as loop and nest to analyze the DR in.
12482 (try_generate_gimple_bb): Likewise.
12483 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12484 (add_loop_constraints): For blocks in a loop not in the region
12485 create a dimension with a single iteration.
12486 * sese.h (gbb_loop_at_index): Remove assert.
12488 * cfgloop.c (loop_preheader_edge): For the loop tree root
12489 return the single successor of the entry block.
12490 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12491 Reset the SCEV hashtable and niters.
12492 * graphite-scop-detection.c
12493 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12494 assert that we only have POLYNOMIAL_CHREC that vary in loops
12495 contained in the region.
12496 (scop_detection::graphite_can_represent_expr): Adjust.
12497 (scop_detection::stmt_has_simple_data_refs_p): For loops
12498 not in the region set loop to NULL. The nest is now the
12499 entry edge to the region.
12500 (try_generate_gimple_bb): Likewise.
12501 * sese.c (scalar_evolution_in_region): Adjust for
12502 instantiate_scev change.
12503 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12504 Make nest parameter the edge into the region.
12505 (create_data_ref): Likewise.
12506 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12507 entry edge into a region and adjust instantiate_scev calls.
12508 (create_data_ref): Likewise.
12509 (graphite_find_data_references_in_stmt): Likewise.
12510 (find_data_references_in_stmt): Pass the loop preheader edge
12511 from the nest argument.
12512 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12513 parameter the edge into the region.
12514 (instantiate_parameters): Use the loop preheader edge as entry.
12515 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12517 (get_instantiated_value_entry): Make instantiate_below parameter
12518 the edge into the region.
12519 (instantiate_scev_name): Likewise. Adjust dominance checks,
12520 when we cannot use loop-based instantiation instantiate by
12521 walking use-def chains.
12522 (instantiate_scev_poly): Adjust.
12523 (instantiate_scev_binary): Likewise.
12524 (instantiate_scev_convert): Likewise.
12525 (instantiate_scev_not): Likewise.
12526 (instantiate_array_ref): Remove.
12527 (instantiate_scev_3): Likewise.
12528 (instantiate_scev_2): Likewise.
12529 (instantiate_scev_1): Likewise.
12530 (instantiate_scev_r): Do not blindly handle N-operand trees.
12531 Do not instantiate array-refs. Handle all constants and invariants.
12532 (instantiate_scev): Make instantiate_below parameter
12533 the edge into the region.
12534 (resolve_mixers): Use the loop preheader edge for the region
12535 parameter to instantiate_scev_r.
12536 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12538 2017-10-13 Richard Biener <rguenther@suse.de>
12540 PR tree-optimization/82525
12541 * graphite-isl-ast-to-gimple.c
12542 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12544 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12545 Fail code generation when we cannot represent the isl integer.
12546 (binary_op_to_tree): Elide modulo operations that are no-ops
12547 in the type we code generate. Remove now superfluous code
12550 2017-10-13 Richard Biener <rguenther@suse.de>
12552 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12553 (scop_detection::harmful_loop_in_region): Remove premature
12554 IV type restriction.
12555 (scop_detection::graphite_can_represent_scev): We can handle
12556 pointer IVs just fine.
12558 2017-10-13 Alan Modra <amodra@gmail.com>
12560 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12561 "Clobbers and Scratch Registers". Add paragraph on
12562 alternative to clobbers for scratch registers and OpenBLAS
12565 2017-10-13 Alan Modra <amodra@gmail.com>
12567 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12568 example of a memory input for a string of known length. Move
12569 commentary out of table. Add a number of new examples
12570 covering array memory inputs.
12572 2017-10-12 Martin Liska <mliska@suse.cz>
12574 PR tree-optimization/82493
12575 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12576 (test_range_functions): New function.
12577 (sbitmap_c_tests): Likewise.
12578 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12579 * selftest.h (sbitmap_c_tests): New function.
12581 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12583 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12585 * config/rs6000/amo.h: Fix spacing issue.
12587 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12590 * config/i386/i386.md (*ashl<mode>3_mask_1,
12591 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12592 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12595 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12597 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12598 (profile_probability): Set max_probability
12599 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12601 * profile-count.c (profile_probability::differs_from_p): Do not
12602 rely on max_probaiblity == 10000
12604 2017-10-12 Jeff Law <law@redhat.com>
12606 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12609 2017-10-12 Martin Sebor <msebor@redhat.com>
12613 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12614 (handle_alias_pairs): Call it.
12615 * common.opt (-Wattribute-alias): New option.
12616 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12617 * doc/invoke.texi (-Wattribute-alias): Document.
12619 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12622 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12624 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12626 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12627 (make_hard_regno_born, make_hard_regno_dead): Update
12628 bb_killed_pseudos and bb_gen_pseudos.
12630 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12632 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12634 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12636 * config/alpha/alpha.c (alpha_split_conditional_move):
12637 Use std::swap instead of manually swapping.
12638 (alpha_stdarg_optimize_hook): Ditto.
12639 (alpha_canonicalize_comparison): Ditto.
12641 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12643 * tree-loop-distribution.c (struct builtin_info): New struct.
12644 (struct partition): Refactor fields into struct builtin_info.
12645 (partition_free): Free struct builtin_info.
12646 (build_size_arg_loc, build_addr_arg_loc): Delete.
12647 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12648 information from struct builtin_info.
12649 (find_single_drs): New function refactored from classify_partition.
12650 Also moved builtin validity checks to this function.
12651 (compute_access_range, alloc_builtin): New functions.
12652 (classify_builtin_st, classify_builtin_ldst): New functions.
12653 (classify_partition): Refactor code into functions find_single_drs,
12654 classify_builtin_st and classify_builtin_ldst.
12655 (distribute_loop): Don't do runtime alias check when distributing
12657 (find_seed_stmts_for_distribution): New function.
12658 (pass_loop_distribution::execute): Refactor code finding seed
12659 stmts into above function. Support distribution for the innermost
12660 two-level loop nest. Adjust dump information.
12662 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12664 * tree-loop-distribution.c: Adjust the general comment.
12665 (NUM_PARTITION_THRESHOLD): New macro.
12666 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12667 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12668 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12669 in call to build_partition_graph.
12670 (finalize_partitions): New parameter. Make loop distribution more
12671 conservative by fusing more partitions.
12672 (distribute_loop): Don't do runtime alias check in case of loop nest
12674 (find_seed_stmts_for_distribution): New function.
12675 (prepare_perfect_loop_nest): New function.
12676 (pass_loop_distribution::execute): Refactor code finding seed stmts
12677 and loop nest into above functions. Support loop nest distribution.
12678 Adjust dump information accordingly.
12680 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12682 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12683 and set PTYPE_SEQUENTIAL for merged partition.
12685 2017-10-12 Richard Biener <rguenther@suse.de>
12687 PR tree-optimization/69728
12689 2017-09-19 Richard Biener <rguenther@suse.de>
12691 PR tree-optimization/69728
12692 * graphite-sese-to-poly.c (schedule_error): New global.
12693 (add_loop_schedule): Handle empty domain by failing the
12695 (build_original_schedule): Handle schedule_error.
12697 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12698 domain by returning an unchanged schedule.
12700 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12702 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12703 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12705 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12707 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12710 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12713 * expr.c (store_field): Don't optimize away bitsize == 0 store
12714 from CALL_EXPR with addressable return type.
12716 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12718 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12719 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12720 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12721 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12722 TARGET_ISEL instead of TARGET_ISEL<sel>.
12724 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12726 * config/rs6000/rs6000.c
12727 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12729 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12731 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12733 (reg_or_cint_operand): Fix comment.
12734 (reg_or_zero_operand): New predicate.
12735 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12736 * config/rs6000/rs6000.c (output_isel): Delete.
12737 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12738 instead of reg_or_cint_operand. Output instruction directly (not via
12740 (isel_unsigned_<mode>): Ditto.
12741 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12742 gpc_reg_operand. Add an instruction alternative for this. Output
12743 instruction directly.
12744 (*isel_reversed_unsigned_<mode>): Ditto.
12746 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12748 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12749 (TARGET_CANONICALIZE_COMPARISON): Define.
12751 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12754 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12755 Check whether the dest is REG before adding REG_EQUIV note.
12757 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12760 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12762 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12763 (make_hard_regno_born, make_hard_regno_dead): Update
12764 bb_killed_pseudos and bb_gen_pseudos.
12766 2017-10-11 Nathan Sidwell <nathan@acm.org>
12768 * incpath.h (enum incpath_kind): Name enum, prefix values.
12769 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12770 * incpath.c (heads, tails): Use INC_MAX.
12771 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12772 (merge_include_chains, split_quote_chain,
12773 register_include_chains): Update incpath_kind names.
12774 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12775 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12777 (add_framework_path, darwin_register_objc_includes): Likewise.
12778 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12780 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12782 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12783 Do not use float_operator operator predicate.
12784 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12785 * config/i386/predicates.md (float_operator): Remove predicate.
12787 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12789 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12790 (*jccxf_i387): Ditto.
12791 (*jcc<mode>_i387): Ditto.
12792 (*jccu<mode>_i387): Ditto.
12793 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12794 (*jcc_*_i387 splitters): Remove.
12795 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12796 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12797 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12800 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12802 * profile-count.h (slow_safe_scale_64bit): New function.
12803 (safe_scale_64bit): New inline.
12804 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12805 * profile-count.c: Include wide-int.h
12806 (slow_safe_scale_64bit): New.
12808 2017-10-11 Nathan Sidwell <nathan@acm.org>
12810 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12811 HAS_DECL_ASSEMBLER_NAME_P.
12812 * gimple-expr.c (gimple_decl_printable_name: Check
12813 HAS_DECL_ASSEMBLER_NAME_P too.
12814 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12815 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12816 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12817 * passes.c (rest_of_decl_compilation): Check
12818 HAS_DECL_ASSEMBLER_NAME_P too.
12819 * recog.c (verify_changes): Likewise.
12820 * tree-pretty-print.c (dump_decl_name): Likewise.
12821 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12823 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12824 (SET_DECL_ASSEMBLER_NAME): Use it.
12825 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12826 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12827 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12829 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12831 * config.gcc (i386, x86_64): Add extra objects.
12832 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12833 (ix86_min_insn_size): Declare.
12834 (ix86_issue_rate): Declare.
12835 (ix86_adjust_cost): Declare.
12836 (ia32_multipass_dfa_lookahead): Declare.
12837 (ix86_macro_fusion_p): Declare.
12838 (ix86_macro_fusion_pair_p): Declare.
12839 (ix86_bd_has_dispatch): Declare.
12840 (ix86_bd_do_dispatch): Declare.
12841 (ix86_core2i7_init_hooks): Declare.
12842 (ix86_atom_sched_reorder): Declare.
12843 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12844 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12845 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12846 (rip_relative_addr_p): Rename to ...
12847 (ix86_rip_relative_addr_p): ... this one; export.
12848 (memory_address_length): Update.
12849 (ix86_issue_rate): Move to x86-tune-sched.c.
12850 (ix86_flags_dependent): Move to x86-tune-sched.c.
12851 (ix86_agi_dependent): Move to x86-tune-sched.c.
12852 (exact_dependency_1): Move to x86-tune-sched.c.
12853 (exact_store_load_dependency): Move to x86-tune-sched.c.
12854 (ix86_adjust_cost): Move to x86-tune-sched.c.
12855 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12856 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12857 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12858 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12859 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12860 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12861 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12862 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12863 (min_insn_size): Rename to ...
12864 (ix86_min_insn_size): ... this one; export.
12865 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12866 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12867 (core2i7_first_cycle_multipass_backtrack): Move to
12868 x86-tune-sched-core.c.
12869 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12870 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12871 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12872 (ix86_avoid_jump_mispredicts): Update.
12873 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12874 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12875 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12876 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12877 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12878 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12879 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12880 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12881 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12882 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12883 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12884 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12885 (BIG): Move to ix86-tune-sched-bd.c.
12886 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12887 (enum insn_path): Move to ix86-tune-sched-bd.c.
12888 (get_mem_group): Move to ix86-tune-sched-bd.c.
12889 (is_cmp): Move to ix86-tune-sched-bd.c.
12890 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12891 (is_branch): Move to ix86-tune-sched-bd.c.
12892 (is_prefetch): Move to ix86-tune-sched-bd.c.
12893 (init_window): Move to ix86-tune-sched-bd.c.
12894 (allocate_window): Move to ix86-tune-sched-bd.c.
12895 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12896 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12897 (process_end_window): Move to ix86-tune-sched-bd.c.
12898 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12899 (find_constant): Move to ix86-tune-sched-bd.c.
12900 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12901 (has_immediate): Move to ix86-tune-sched-bd.c.
12902 (get_insn_path): Move to ix86-tune-sched-bd.c.
12903 (get_insn_group): Move to ix86-tune-sched-bd.c.
12904 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12905 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12906 (add_insn_window): Move to ix86-tune-sched-bd.c.
12907 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12908 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12909 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12910 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12911 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12912 (do_dispatch): Move to ix86-tune-sched-bd.c.
12913 (has_dispatch): Move to ix86-tune-sched-bd.c.
12914 * config/i386/t-i386: Add new object files.
12915 * config/i386/x86-tune-costs.h: New file.
12916 * config/i386/x86-tune-sched-atom.c: New file.
12917 * config/i386/x86-tune-sched-bd.c: New file.
12918 * config/i386/x86-tune-sched-core.c: New file.
12919 * config/i386/x86-tune-sched.c: New file.
12921 2017-10-11 Liu Hao <lh_mouse@126.com>
12923 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12924 the generic version below instead.
12925 (should_colorize): Recognize Windows consoles as terminals
12927 * pretty-print.c [__MINGW32__] (write_all): New function.
12928 [__MINGW32__] (find_esc_head): Likewise.
12929 [__MINGW32__] (find_esc_terminator): Likewise.
12930 [__MINGW32__] (eat_esc_sequence): Likewise.
12931 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12933 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12936 2017-10-11 Richard Biener <rguenther@suse.de>
12938 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12939 Properly call analyze_scalar_evolution with the loop of the stmt.
12941 2017-10-11 Richard Biener <rguenther@suse.de>
12943 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12944 * tree-core.h (tree_base): Add chrec_var union member.
12945 * tree.h (CHREC_VAR): Remove.
12946 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12947 * tree-chrec.h (build_polynomial_chrec): Adjust.
12948 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12949 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12951 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12953 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12954 * match.pd: ... here.
12955 ((T) X == (T) Y): Relax condition.
12957 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12959 PR tree-optimization/82472
12960 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12962 (break_alias_scc_partitions): Update postorder number.
12964 2017-10-11 Martin Liska <mliska@suse.cz>
12967 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12969 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12971 2017-10-11 Martin Liska <mliska@suse.cz>
12973 * print-rtl.c (print_insn): Move declaration of idbuf
12974 to same scope as name.
12976 2017-10-11 Martin Liska <mliska@suse.cz>
12981 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12982 volatile arguments.
12984 2017-10-11 Martin Liska <mliska@suse.cz>
12987 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12988 volatile arguments.
12990 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12992 * config.gcc (default_gnu_indirect_function): Default to yes for
12993 arm*-*-linux* with glibc.
12995 2017-10-11 Richard Biener <rguenther@suse.de>
12997 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12998 default-defs and types we do not want to analyze.
12999 (interpret_loop_phi): Replace unreachable code with an assert.
13000 (compute_scalar_evolution_in_loop): Remove and inline ...
13001 (analyze_scalar_evolution_1): ... here, replacing condition with
13002 what makes the intent clearer. Remove handling of cases
13003 get_scalar_evolution now handles.
13005 2017-10-10 Jim Wilson <wilson@tuliptree.org>
13007 PR rtl-optimization/81434
13008 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
13009 comment for main loop. In sched_group_found if, also add checks for
13010 pass and min_cost_group.
13012 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
13014 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
13015 (rs6000_insn_cost): New function.
13016 * config/rs6000/rs6000.md (cost): New attribute.
13018 2017-10-10 Jakub Jelinek <jakub@redhat.com>
13019 H.J. Lu <hongjiu.lu@intel.com>
13023 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
13024 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
13025 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
13026 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
13027 to OPTION_MASK_ISA_AVX512VL - builtins that have both
13028 OPTION_MASK_ISA_MMX and some other bit set require both
13029 mmx and the ISAs without the mmx bit.
13030 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
13031 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
13032 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
13033 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
13034 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
13035 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
13036 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
13037 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
13038 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
13039 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
13040 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
13041 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
13042 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
13043 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
13044 Add OPTION_MASK_ISA_MMX.
13046 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
13048 * config.gcc (armv7*-*-freebsd*): New target.
13049 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
13051 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
13053 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
13054 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
13057 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
13059 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
13060 a template parameter.
13061 (WIDE_INT_REF_FOR): Update accordingly.
13062 * tree.h (wi::int_traits <const_tree>): Delete.
13063 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13064 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13065 (wi::tree_to_wide_ref): New typedef.
13066 (wi::to_wide): New function.
13067 * calls.c (get_size_range): Use wi::to_wide when operating on
13068 trees as wide_ints.
13069 * cgraph.c (cgraph_node::create_thunk): Likewise.
13070 * config/i386/i386.c (ix86_data_alignment): Likewise.
13071 (ix86_local_alignment): Likewise.
13072 * dbxout.c (stabstr_O): Likewise.
13073 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13074 * expr.c (const_vector_from_tree): Likewise.
13075 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13076 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13077 (fold_negate_expr_1, int_const_binop_1, const_binop)
13078 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13079 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13080 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13081 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13082 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13083 (fold_not_const, round_up_loc): Likewise.
13084 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13085 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13086 (alloca_call_type): Likewise.
13087 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13088 * godump.c (go_output_typedef): Likewise.
13089 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13090 * internal-fn.c (get_min_precision): Likewise.
13091 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13092 * ipa-polymorphic-call.c
13093 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13094 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13095 (ipa_modify_call_arguments): Likewise.
13096 * match.pd: Likewise.
13097 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13098 * print-tree.c (print_node_brief, print_node): Likewise.
13099 * stmt.c (expand_case): Likewise.
13100 * stor-layout.c (layout_type): Likewise.
13101 * tree-affine.c (tree_to_aff_combination): Likewise.
13102 * tree-cfg.c (group_case_labels_stmt): Likewise.
13103 * tree-data-ref.c (dr_analyze_indices): Likewise.
13104 (prune_runtime_alias_test_list): Likewise.
13105 * tree-dump.c (dequeue_and_dump): Likewise.
13106 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13107 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13108 * tree-pretty-print.c (dump_generic_node): Likewise.
13109 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13110 (simple_iv_with_niters): Likewise.
13111 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13112 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13113 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13114 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13115 (refine_value_range_using_guard, number_of_iterations_ne_max)
13116 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13117 (get_cst_init_from_scev, record_nonwrapping_iv)
13118 (scev_var_range_cant_overflow): Likewise.
13119 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13120 * tree-ssa-pre.c (compute_avail): Likewise.
13121 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13122 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13123 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13124 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13125 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13126 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13127 (dump_case_nodes, try_switch_expansion): Likewise.
13128 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13129 (vect_do_peeling): Likewise.
13130 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13131 * tree-vect-stmts.c (vectorizable_load): Likewise.
13132 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13133 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13134 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13135 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13136 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13137 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13138 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13139 (evrp_dom_walker::before_dom_children): Likewise.
13140 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13141 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13142 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13143 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13144 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13145 (get_range_pos_neg): Likewise.
13146 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13147 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13148 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13149 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13150 * config/avr/avr.c (avr_fold_builtin): Likewise.
13151 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13152 * config/msp430/msp430.c (msp430_attr): Likewise.
13153 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13154 * config/powerpcspe/powerpcspe-c.c
13155 (altivec_resolve_overloaded_builtin): Likewise.
13156 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13157 (rs6000_expand_ternop_builtin): Likewise.
13158 * config/rs6000/rs6000-c.c
13159 (altivec_resolve_overloaded_builtin): Likewise.
13160 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13161 (rs6000_expand_ternop_builtin): Likewise.
13162 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13164 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13166 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13167 when copying loop nest with only one inner loop.
13169 2017-10-10 Richard Biener <rguenther@suse.de>
13171 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13172 blocks if SCEV is active.
13173 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13175 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13176 (scev_initialize): Assert we are not yet initialized.
13178 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13180 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13181 inner loop's exit stmt by making it always exit the loop, otherwise
13182 we would generate an infinite empty loop.
13184 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13186 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13187 renaming variables in new preheader if it's deleted.
13189 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13191 * tree-loop-distribution.c (struct partition): Remove unused field
13192 loops of the structure.
13193 (partition_alloc, partition_free): Ditto.
13194 (build_rdg_partition_for_vertex): Ditto.
13196 2017-10-09 Jeff Law <law@redhat.com>
13198 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13199 return type to match prototype and documentation.
13201 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13203 * config/rs6000/rs6000.c (processor_costs): Move to ...
13204 * config/rs6000/rs6000.h: ... here.
13205 (rs6000_cost): Declare.
13207 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13209 * except.c (setjmp_fn): New global variable.
13210 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13211 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13212 if DONT_USE_BUILTIN_SETJMP is defined.
13214 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13216 * target.def (insn_cost): New hook.
13217 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13218 * doc/tm.texi: Regenerate.
13219 * rtlanal.c (insn_cost): Use the new hook.
13221 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13223 * combine.c (combine_validate_cost): Compute the new insn_cost,
13224 not just pattern_cost.
13225 (try_combine): Adjust comment.
13227 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13229 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13231 * combine.c (uid_insn_cost): Adjust comment.
13232 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13234 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13235 * dse.c (find_shift_sequence): Ditto.
13236 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13237 (bb_valid_for_noce_process_p): Use pattern_cost.
13238 * rtl.h (insn_rtx_cost): Delete.
13239 (pattern_cost): New prototype.
13240 (insn_cost): New prototype.
13241 * rtlanal.c (insn_rtx_cost): Rename to...
13242 (pattern_cost): ... this.
13245 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13247 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13248 (*jcc<mode>_0_r_i387): Ditto.
13249 (*jccxf_r_i387): Ditto.
13250 (*jcc<mode>_r_i387): Ditto.
13251 (*jccu<mode>_r_i387): Ditto.
13252 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13253 (*jcc): Rename from *jcc_1.
13255 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13257 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13258 deferred rescans after the lvx/stvx recombination pre-pass.
13260 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13262 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13263 memory operation instruction support.
13264 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13265 (rs6000-ibm-aix[789]*): Likewise.
13266 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13267 Document new functions.
13269 2017-10-09 Richard Biener <rguenther@suse.de>
13271 PR tree-optimization/82397
13272 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13273 equality only for semantically equal trees.
13275 2017-10-09 Richard Biener <rguenther@suse.de>
13277 PR tree-optimization/82449
13278 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13279 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13280 allow constant addresses.
13281 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13284 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13286 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13289 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13292 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13295 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13298 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13300 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13303 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13304 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13306 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13308 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13309 (WI_BINARY_PREDICATE_RESULT): Likewise.
13310 (wi::binary_traits::operator_result): New type.
13311 (wi::binary_traits::predicate_result): Likewise.
13312 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13313 (generic_wide_int::operator==, generic_wide_int::operator!=)
13314 (generic_wide_int::operator&, generic_wide_int::and_not)
13315 (generic_wide_int::operator|, generic_wide_int::or_not)
13316 (generic_wide_int::operator^, generic_wide_int::operator+
13317 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13319 (operator~, unary operator-, operator==, operator!=, operator&)
13320 (operator|, operator^, operator+, binary operator-, operator*): New
13322 * expr.c (get_inner_reference): Use wi::bit_and_not.
13323 * fold-const.c (fold_binary_loc): Likewise.
13324 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13325 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13326 (bit_value_binop): Likewise.
13327 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13328 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13329 (extract_range_from_binary_expr_1): Likewise.
13330 (masked_increment): Likewise.
13331 (simplify_bit_ops_using_ranges): Likewise.
13333 2017-10-09 Martin Jambor <mjambor@suse.cz>
13336 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13337 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13338 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13340 (hsa_fixup_mov_insn_type): New function.
13341 (hsa_op_with_type::get_in_type): Use it.
13342 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13343 immediates in an assert.
13344 (hsa_op_with_type::extend_int_to_32bit): New method.
13345 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13346 types. Convert to dest type if necessary.
13347 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13348 (reg_for_gimple_ssa): Pass false as min32int to
13349 hsa_type_for_scalar_tree_type.
13350 (gen_hsa_addr): Fixup type when creating addresable temporary.
13351 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13352 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13353 necessary. Call hsa_fixup_mov_insn_type.
13354 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13355 extend operands and convert to dest type if necessary.
13356 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13357 to dest type if necessary.
13358 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13359 if conversion nt necessary and size matches.
13360 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13361 to dest type if necessary.
13362 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13363 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13365 (gen_hsa_clrsb): Likewise.
13366 (gen_hsa_ffs): Likewise.
13367 (gen_hsa_divmod): Extend operands and convert to dest type if
13369 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13371 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13373 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13374 Remove empty default arguments. Use a brace block as output
13376 (conditional return): Ditto.
13378 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13379 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13381 (group_ending_nop): Ditto.
13382 (doloop_end): Ditto.
13383 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13384 (splitters for those): Ditto.
13386 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13388 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13389 a conditional jump (and the compare for it) so that pc_rtx is the
13391 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13392 for the deleted and renamed ctr<mode>_internal[234] patterns.
13393 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13394 Delete second conditional return pattern.
13395 (ctr<mode>_internal2): Delete this second bdnz pattern.
13396 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13397 (ctr<mode>_internal4): Delete this second bdz pattern.
13399 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13401 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13402 (always_initialized_rtx_for_ssa_name_p): New predicate.
13403 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13404 (finish_out_of_ssa): Free new field of SA.
13405 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13406 * tree-ssa-coalesce.c: Include tree-ssa.h.
13407 (get_parm_default_def_partitions): Remove extern keyword.
13408 (get_undefined_value_partitions): New function.
13409 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13410 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13411 uninitialized bits.
13412 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13414 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13416 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13418 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13420 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13421 for targets that preffer 128bit.
13423 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13425 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13427 2017-10-08 Olivier Hainque <hainque@adacore.com>
13429 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13430 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13432 (thumb_set_return_address): Likewise.
13434 2017-10-08 Olivier Hainque <hainque@adacore.com>
13436 * common/config/arm/arm-common.c (arm_except_unwind_info):
13437 Handle DWARF2_UNWIND_INFO.
13439 2017-10-07 Michael Collison <michael.collison@arm.com>
13441 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13444 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13446 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13447 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13448 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13449 defined, force the creation of a new block for a dispatch label.
13451 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13453 * invoke.texi (Wsuggest-attribute=cold): Document.
13454 * common.opt (Wsuggest-attribute=cold): New
13455 * ipa-pure-const.c (warn_function_cold): New function.
13456 * predict.c (compute_function_frequency): Use it.
13457 * predict.h (warn_function_cold): Declare.
13459 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13461 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13464 2017-10-06 Martin Liska <mliska@suse.cz>
13466 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13467 keyword for member functions.
13468 (struct sanopt_tree_couple): New struct.
13469 (struct sanopt_tree_couple_hash): New function.
13470 (struct sanopt_ctx): Add new hash_map.
13471 (has_dominating_ubsan_ptr_check): New function.
13472 (record_ubsan_ptr_check_stmt): Likewise.
13473 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13474 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13475 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13477 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13480 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13481 aarch64_simd_valid_immediate on CONST_VECTORs.
13482 (aarch64_reg_or_bic_imm): Likewise.
13484 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13486 PR rtl-optimization/82396
13487 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13489 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13491 * graphite-dependences.c (scop_get_reads): Move code to...
13492 (scop_get_must_writes): Move code to...
13493 (scop_get_may_writes): Move code to...
13494 (scop_get_reads_and_writes): ... here.
13495 (scop_get_dependences): Call scop_get_reads_and_writes.
13497 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13499 PR tree-optimization/82434
13500 * fold-const.h (can_native_encode_type_p,
13501 can_native_encode_string_p): Remove.
13502 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13503 don't encode anything, just return what would be otherwise returned.
13504 (native_encode_fixed, native_encode_complex, native_encode_vector):
13506 (native_encode_string): Likewise. Inline by hand
13507 can_native_encode_string_p.
13508 (can_native_encode_type_p): Remove.
13509 (can_native_encode_string_p): Remove.
13510 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13511 STRING_CSTs using can_native_encode_string_p, test all
13512 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13513 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13514 argument from native_encode_expr.
13515 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13516 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13519 2017-10-06 Richard Biener <rguenther@suse.de>
13521 PR tree-optimization/82397
13522 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13523 operand_equal_p but rely on data_ref_compare_tree for detecting
13525 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13526 to match up with dr_group_sort_cmp.
13528 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13531 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13533 * config/s390/s390-builtin-types.def: Regenerate.
13535 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13538 * config/s390/s390-builtin-types.def: Regenerate.
13539 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13540 Change flag from B_VXE to B_VX.
13541 (s390_vec_min_dbl): Remove B_VXE flag.
13543 2017-10-06 Richard Biener <rguenther@suse.de>
13545 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13546 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13547 translate_isl_ast_to_gimple::is_valid_rename,
13548 translate_isl_ast_to_gimple::get_rename,
13549 translate_isl_ast_to_gimple::get_def_bb_for_const,
13550 translate_isl_ast_to_gimple::get_new_name,
13551 translate_isl_ast_to_gimple::collect_all_ssa_names,
13552 translate_isl_ast_to_gimple::copy_loop_phi_args,
13553 translate_isl_ast_to_gimple::collect_all_ssa_names,
13554 translate_isl_ast_to_gimple::copy_loop_phi_args,
13555 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13556 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13557 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13558 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13559 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13560 translate_isl_ast_to_gimple::copy_cond_phi_args,
13561 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13562 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13563 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13564 translate_isl_ast_to_gimple::rename_uses,
13565 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13566 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13567 (set_rename_for_each_def): Likewise.
13568 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13569 here. Handle rewriting SCEV analyzable uses here.
13570 (copy_bb_and_scalar_dependences): Generate code for PHI
13572 (graphite_regenerate_ast_isl): Adjust.
13573 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13574 (add_write, add_read): New functions.
13575 (build_cross_bb_scalars_def): Use it and simplify.
13576 (build_cross_bb_scalars_use): Likewise.
13577 (graphite_find_cross_bb_scalar_vars): Inline into...
13578 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13579 simulating out-of-SSA. Compute liveout and add dependencies.
13580 (build_scops): Force an empty entry block.
13581 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13583 (sese_build_liveouts): Declare.
13584 (sese_trivially_empty_bb_p): Likewise.
13585 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13586 compute liveout and debug_liveout.
13587 (sese_bad_liveouts_use): Remove.
13588 (sese_reset_debug_liveouts_bb): Likewise.
13589 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13590 (sese_build_liveouts): Build liveout and debug_liveout and store
13592 (new_sese_info): Adjust.
13593 (free_sese_info): Likewise.
13594 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13595 do not build liveout here.
13596 (move_sese_in_condition): Adjust region entry.
13597 (scev_analyzable_p): Match up with chrec_apply requirements.
13598 (sese_trivially_empty_bb_p): New.
13599 * tree-into-ssa.c (get_reaching_def): Properly support generating
13600 default-defs for incremental rewrite of anonymous names.
13602 2017-10-06 Richard Biener <rguenther@suse.de>
13604 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13605 precision do not perform modulo reduction.
13607 2017-10-06 Richard Biener <rguenther@suse.de>
13609 PR tree-optimization/82436
13610 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13611 conservatively choose the vectorization factor when checking
13612 whether we can perform the required load permutation.
13613 (vect_transform_slp_perm_load): Assert when we may not fail.
13615 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13617 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13618 message for incompatible -msdata=* and -mcall-* options.
13620 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13622 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13623 rate for post-reload scheduling.
13625 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13627 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13629 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13631 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13632 to improve monte carlo in scimark.
13634 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13636 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13637 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13638 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13639 pentium4_cost, nocona_cost): Set reassociation width to 1.
13640 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13641 width to 2 for fp operations and 1 otherwise.
13642 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13644 (atom_cost): Set reassociation width to 2.
13645 (slm_cost, generic_cost): Set fp reassociation width
13646 to 2 and 1 otherwise.
13647 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13648 (core_cost): Set fp reassociation width to 4 and vector to 2.
13649 (ix86_reassociation_width): Rewrite using cost table; special case
13650 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13651 and TARGET_AVX128_OPTIMAL.
13652 * config/i386/i386.h (processor_costs): Add
13653 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13654 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13655 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13656 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13657 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13658 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13660 2017-10-05 Nathan Sidwell <nathan@acm.org>
13662 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13664 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13666 * config/arm/arm.c (arm_test_fpu_data): New.
13667 (arm_run_selftests): Call arm_test_fpu_data.
13669 2017-10-04 Nathan Sidwell <nathan@acm.org>
13671 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13673 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13675 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13677 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13678 check type for aarch64_simd_valid_immediate.
13679 (aarch64_output_simd_mov_immediate): Update prototype.
13680 (aarch64_simd_valid_immediate): Update prototype.
13681 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13682 support for ORR-immediate.
13683 (and<mode>3): modified pattern to add support for BIC-immediate.
13684 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13685 now checks for valid immediate for BIC and ORR based on new enum
13687 (aarch64_output_simd_mov_immediate): Function now used to output
13688 BIC/ORR imm as well based on new enum argument.
13689 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13691 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13692 (aarch64_reg_or_bic_imm): Likewise.
13694 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13696 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13697 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13699 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13703 PR rtl-optimization/82396
13704 * haifa-sched.c (autopref_multipass_init): Simplify
13706 (autopref_rank_data): Simplify sort order.
13707 * sched-int.h (autopref_multipass_data_): Remove
13708 multi_mem_insn_p, min_offset and max_offset.
13710 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13712 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13714 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13716 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13717 vect_doubleint_cvt.
13719 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13721 * doc/sourcebuild.texi: Document vect_long_mult.
13723 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13725 PR tree-optimization/82413
13726 * fold-const.c (build_range_check): Use widest_int when comparing
13727 the maximum ETYPE value with HIGH.
13729 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13731 PR rtl-optimization/82396
13732 * haifa-sched.c (autopref_multipass_init): Simplify
13734 (autopref_rank_data): Simplify sort order.
13735 * sched-int.h (autopref_multipass_data_): Remove
13736 multi_mem_insn_p, min_offset and max_offset.
13738 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13740 PR tree-optimization/82381
13741 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13742 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13745 PR tree-optimization/82374
13746 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13747 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13748 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13749 current_function_decl to the new decl.
13751 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13753 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13754 helper macro for IEEE float128 hardware built-in functions.
13755 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13757 (TRUNCF128_ODD): Likewise.
13758 (ADDF128_ODD): Likewise.
13759 (SUBF128_ODD): Likewise.
13760 (MULF128_ODD): Likewise.
13761 (DIVF128_ODD): Likewise.
13762 (FMAF128_ODD): Likewise.
13763 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13764 UNSPEC_TRUNC_ROUND_TO_ODD.
13765 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13766 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13767 floating point round to odd instructions.
13768 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13769 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13770 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13771 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13772 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13773 (trunc<mode>sf2_hw): Change the truncate with round to odd
13774 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13775 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13776 to odd hardware instructions.
13777 (sub<mode>3_odd): Likewise.
13778 (mul<mode>3_odd): Likewise.
13779 (div<mode>3_odd): Likewise.
13780 (sqrt<mode>2_odd): Likewise.
13781 (fma<mode>4_odd): Likewise.
13782 (fms<mode>4_odd): Likewise.
13783 (nfma<mode>4_odd): Likewise.
13784 (nfms<mode>4_odd): Likewise.
13785 (trunc<mode>df2_odd): Change the truncate with round to odd
13786 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13788 * doc/extend.texi (PowerPC built-in functions): Update documentation
13789 for existing IEEE float128-bit built-in functions. Add built-in
13790 functions that generate the IEEE 128-bit floating point round to
13793 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13795 PR rtl-optimization/77729
13796 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13797 to (X&(C1&~C2))|C2 transformations.
13799 2017-10-03 Martin Jambor <mjambor@suse.cz>
13801 PR tree-optimization/82363
13802 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13803 mismatch, mark lacc written regardless of racc.
13805 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13807 PR tree-optimization/82381
13808 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13809 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13810 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13811 fallthrough into reassoc_stmt_dominates_stmt_p.
13814 * combine.c (combine_instructions): Don't combine in unreachable
13817 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13820 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13821 function to not use the have_cpu variable. Do not set cpu_index,
13822 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13823 or the default cpu.
13824 (rs6000_valid_attribute_p): Remove duplicate initializations of
13825 old_optimize and func_optimize.
13826 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13827 (rs6000_activate_target_options): Make global.
13828 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13831 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13833 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13834 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13835 Set *poffset to 0 and *psize and *pmax_size to -1 if
13836 *poffset + *psize overflows in HOST_WIDE_INT.
13838 PR tree-optimization/82387
13839 PR tree-optimization/82388
13840 PR tree-optimization/82389
13841 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13842 instead of live_bytes non-NULL.
13844 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13847 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13850 2017-10-02 Richard Biener <rguenther@suse.de>
13852 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13853 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13854 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13856 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13858 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13859 requested precision matches the type's.
13860 * calls.c (alloc_max_size): Calculate the new candidate size as
13861 a widest_int and use wi::to_widest when comparing it with the
13862 current candidate size.
13863 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13864 zero rather than integer_zero_node.
13865 * match.pd: Check for a no-op conversion before using wi::add
13866 rather than after. Use tree_to_uhwi when summing small shift
13867 counts into an unsigned int.
13869 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13870 Alan Hayward <alan.hayward@arm.com>
13871 David Sherwood <david.sherwood@arm.com>
13874 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13875 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13876 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13877 POINTER_AND_FP_REGS.
13879 2017-10-02 Richard Biener <rguenther@suse.de>
13881 PR tree-optimization/82355
13882 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13883 a mapping for the enclosing loop but avoid generating one for
13884 the loop tree root.
13885 (copy_bb_and_scalar_dependences): Remove premature codegen
13886 error on PHIs in blocks duplicated into multiple places.
13887 * graphite-scop-detection.c
13888 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13889 in the region use it as loop and nest to analyze the DR in.
13890 (try_generate_gimple_bb): Likewise.
13891 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13892 (add_loop_constraints): For blocks in a loop not in the region
13893 create a dimension with a single iteration.
13894 * sese.h (gbb_loop_at_index): Remove assert.
13896 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13898 * omp-expand.c (adjust_context_scope): New function.
13899 (expand_parallel_call): Call adjust_context_scope.
13901 2017-10-01 Jeff Law <law@redhat.com>
13903 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13904 dom_opt_dom_walker class with direct access to private members.
13905 Add comments. Call test_for_singularity.
13906 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13907 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13908 m_dummy_cond anymore.
13909 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13911 (pass_dominator:execute): Build the dummy_cond here and pass it
13912 to the dom_opt_dom_walker ctor.
13913 (test_for_singularity): New function.
13915 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13916 Maya Rashish <coypu@sdf.org>
13918 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13919 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13920 (alpha*-*-netbsd*) Use nbsd_tm_file.
13921 (arm*-*-netbsdelf*) Likewise.
13922 (i[34567]86-*-netbsdelf*) Likewise.
13923 (x86_64-*-netbsd*) Likewise.
13924 (mips*-*-netbsd*) Likewise.
13925 (powerpc-*-netbsd*) Likewise.
13926 (sh*-*-netbsd*) Likewise.
13927 (sparc-*-netbsdelf*) Likewise.
13928 (sparc64-*-netbsd*) Likewise.
13929 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13931 (vax-*-netbsdelf*) Likewise.
13932 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13933 (UINT_FAST8_TYPE) Likewise.
13934 (INT_FAST16_TYPE) Check CHAR_FAST16.
13935 (UINT_FAST16_TYPE) Likewise.
13937 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13940 * config/i386/i386.md
13941 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13942 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13943 *divmodsi4_zext_2): New define_insn_and_split.
13944 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13945 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13946 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13947 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13948 New define_insn_and_split.
13949 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13950 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13951 operands[1] having DImode when mode is SImode.
13953 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13954 always SImode for DIV and MOD in REG_EQUAL notes.
13956 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13958 PR middle-end/82319
13959 * match.pd: Fix handling of NaNs in pattern.
13961 2017-09-29 Jeff Law <law@redhat.com>
13963 * sbitmap.c (bitmap_bit_in_range_p): New function.
13964 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13965 * tree-ssa-dse.c (live_bytes_read): New function.
13966 (dse_classify_store): Ignore reads of dead bytes.
13968 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13969 typos and whitespace errors.
13970 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13971 * config/s390/s390.c (s390_emit_prologue): Likewise.
13973 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13976 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13977 with a symbol for LRA.
13979 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13981 PR rtl-optimization/82338
13982 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13984 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13986 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13987 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13988 (qsort_chk): Declare.
13989 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13990 (qsort_chk): New function.
13992 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13994 PR tree-optimization/82337
13995 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13996 phi definition if the PHI result appears in an abnormal PHI.
13997 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13998 basis appears in an abnormal PHI.
14000 2017-09-29 Richard Biener <rguenther@suse.de>
14002 * graphite-isl-ast-to-gimple.c
14003 (translate_isl_ast_to_gimple::set_codegen_error): New function.
14004 (binary_op_to_tree): Use it.
14005 (get_rename_from_scev): Likewise.
14006 (copy_loop_phi_nodes): Likewise.
14007 (copy_bb_and_scalar_dependences): Likewise.
14008 (translate_pending_phi_nodes): Likewise.
14010 2017-09-29 Jakub Jelinek <jakub@redhat.com>
14013 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
14014 for movabsq $(i32 << shift), r64.
14016 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
14018 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
14019 index when encoding %esp as %rsp to avoid 0x67 prefix.
14021 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
14023 * config/i386/i386.md (*movsf_internal, *movdf_internal):
14024 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
14026 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
14028 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
14029 Extensions with more than 16 double VFP registers.
14030 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
14031 to_clear_mask and all code related to it. Replace the remaining
14032 entry by a sbitmap and adapt code accordingly.
14034 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
14036 * brig-builtins.def: Change pure attributes to const.
14038 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14040 * config.gcc (default_gnu_indirect_function): Default to yes for
14041 sparc*-*-linux* with glibc.
14043 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14045 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
14046 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
14047 when creating .init_array and .fini_array sections with priority
14050 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
14053 * config/aarch64/aarch64.c
14054 (aarch64_builtin_support_vector_misalignment): Always return false
14055 when misalignment is unknown.
14057 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14059 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
14060 this function to return false if the definition used by the swap
14061 instruction is artificial, or if the memory address from which the
14062 constant value is loaded is not represented by a base address held
14063 in a register or if the base address register is a frame or stack
14064 pointer. Additionally, return false if the base address of the
14065 loaded constant is a SYMBOL_REF but is not considered to be a
14067 (replace_swapped_load_constant): New function.
14068 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14069 loaded constant vector with a load of a swapped constant vector.
14071 2017-09-27 Carl Love <cel@us.ibm.com>
14073 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14074 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14075 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14078 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14080 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14081 first, always call autopref_rank_data otherwise.
14083 2017-09-27 Richard Biener <rguenther@suse.de>
14085 * graphite-scop-detection.c (find_scop_parameters): Move
14086 loop bound handling ...
14087 (gather_bbs::before_dom_children): ... here, avoiding the need
14088 to build scop_info->loop_nest.
14089 (record_loop_in_sese): Remove.
14090 * sese.h (sese_info_t::loop_nest): Remove.
14091 * sese.c (new_sese_info): Do not allocate loop_nest.
14092 (free_sese_info): Do not free loop_nest.
14094 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14097 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14098 lhs from calls if the lhs has addressable type.
14100 2017-09-27 Richard Biener <rguenther@suse.de>
14102 * graphite.h (scop::max_alias_set): New member.
14103 * graphite-scop-detection.c: Remove references to non-existing
14104 --param in comments.
14105 (build_alias_sets): Record the maximum alias set used for drs.
14106 (build_scops): Support zero as unlimited for
14107 --param graphite-max-arrays-per-scop.
14108 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14109 and inline into ...
14110 (build_poly_sr_1): ... here. Compute alias set based on the
14111 maximum alias set used for drs rather than
14112 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14114 2017-09-27 Richard Biener <rguenther@suse.de>
14116 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14117 --param loop-block-tile-size=0 to disable tiling.
14119 2017-09-27 Richard Biener <rguenther@suse.de>
14121 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14122 (graphite-max-nb-scop-params): Document special value zero.
14123 * domwalk.h (dom_walker::STOP): New symbolical constant.
14124 (dom_walker::dom_walker): Add optional parameter for bb to
14126 (dom_walker::~dom_walker): Declare.
14127 (dom_walker::before_dom_children): Document STOP return value.
14128 (dom_walker::m_user_bb_to_rpo): New member.
14129 (dom_walker::m_bb_to_rpo): Likewise.
14130 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14131 mapping here if not provided by the user.
14132 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14133 provided by the user.
14134 (dom_walker::STOP): Define.
14135 (dom_walker::walk): Do not compute bb to RPO mapping here.
14136 Support STOP return value from before_dom_children to stop
14138 * graphite-optimize-isl.c (optimize_isl): If the schedule
14139 is the same still generate code if -fgraphite-identity
14140 or -floop-parallelize-all are given.
14141 * graphite-scop-detection.c: Include cfganal.h.
14142 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14144 (gather_bbs::before_dom_children): Return STOP for BBs
14146 (build_scops): Compute bb to RPO mapping and pass it to
14147 the domwalk. Treat --param graphite-max-nb-scop-params=0
14148 as not limiting the number of params.
14149 * graphite.c (graphite_initialize): Remove limit on the
14150 number of basic-blocks in a function.
14151 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14152 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14153 default value of 10.
14155 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14157 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14158 Adjust code to eliminate needing to do the shift right 32-bits
14159 operation after XSCVDPSPN.
14161 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14163 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14164 ((X / Y) != 0 -> X >= Y): Likewise.
14166 2017-09-26 Carl Love <cel@us.ibm.com>
14168 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14169 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14170 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14171 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14172 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14173 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14174 definitions and overloading.
14175 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14176 statement for P9V_BUILTIN_XST_LEN_R.
14177 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14178 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14179 define_expand and define_insn for the instructions and builtins.
14180 * doc/extend.texi: Update the built-in documentation file for the new
14181 built-in functions.
14182 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14183 define_insn for the instructions
14185 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14188 * gcc/config/netbsd-protos.h: New file.
14189 * gcc/config/netbsd.c: New file.
14190 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14191 * gcc/config/t-netbsd: New file.
14192 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14193 (tmake_file) Add t-netbsd.
14194 (extra_objs) Add netbsd.o.
14196 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14200 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14202 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14204 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14205 sign extension from a vector register to a GPR by doing a 32-bit
14206 direct move and then an EXTSW.
14207 (extendsi<mode>2 splitter): Likewise.
14208 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14209 right or vector extract after doing XSCVDPSPN. Use
14210 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14212 (movdi_from_sf_zero_ext): Likewise.
14213 (reload_gpr_from_vsxsf): Likewise.
14214 (p8_mfvsrd_4_disf): Delete, no longer used.
14215 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14216 then needing to move the SFmode to a GPR to use the XSCVDPSP
14217 instruction instead of FRSP and XSCVDPSPN.
14218 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14219 it is adjacent to the other XSCVSPDP insns.
14220 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14221 SFmode to be in traditional Altivec registers.
14222 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14223 (vsx_xscvspdpn): Likewise.
14224 (vsx_xscvspdpn_scalar): Likewise.
14226 2017-09-26 Martin Jambor <mjambor@suse.cz>
14228 * tree-sra.c (compare_access_positions): Put integral types first,
14229 stabilize sorting of integral types, remove conditions putting
14230 non-full-precision integers last.
14231 (sort_and_splice_var_accesses): Disable scalarization if a
14232 non-integert would be represented by a non-full-precision integer.
14234 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14236 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14237 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14238 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14239 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14240 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14241 conditionals inside the function instead of around it. Call
14242 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14243 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14245 2017-09-26 Richard Biener <rguenther@suse.de>
14247 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14249 (scop_detection::build_scop_breadth): ... this. Removed.
14250 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14251 (scop_detection::harmful_stmt_in_bb): Likewise.
14252 (scop_detection::graphite_can_represent_stmt): Likewise.
14253 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14254 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14255 (scop_detection::can_represent_loop_1): ... this. Removed.
14256 (scop_detection::harmful_loop_in_region): Simplify after inlining
14257 the above and remove more quadraticness.
14258 (build_scops): Adjust.
14259 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14262 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14265 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14266 REGNO (base) == SP_REG if base is a REG.
14268 PR middle-end/35691
14269 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14270 if it is different SSA_NAME.
14271 (optimize_range_tests_cmp_bitwise): New function.
14272 (optimize_range_tests): Call it.
14274 2017-09-26 Richard Biener <rguenther@suse.de>
14276 PR tree-optimization/82321
14277 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14278 for the def being inside the loop.
14280 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14282 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14284 * config/s390/s390-builtins.def: Fix constraint on op4.
14286 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14288 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14289 independent expanders.
14290 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14291 ("vec_ordered", "vec_unordered"): New expanders.
14293 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14295 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14298 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14300 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14301 vec_unpacks_lo_v16qi.
14302 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14304 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14306 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14307 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14308 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14310 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14312 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14314 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14316 ("*vec_slb<mode>"): New insn pattern.
14317 ("vec_shr_<mode>"): New expander.
14318 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14319 and force the shift count operand to V16QImode.
14320 ("vec_srb<mode>"): Set shift count mode to V16QI.
14322 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14324 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14325 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14326 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14328 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14331 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14333 2017-09-26 Richard Biener <rguenther@suse.de>
14335 PR tree-optimization/82320
14336 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14339 2017-09-25 Jeff Law <law@redhat.com>
14341 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14342 prototype for new argument.
14343 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14344 mostly extracted from rs6000_emit_allocate_stack.
14345 (rs6000_emit_probe_stack_range_stack_clash): New function.
14346 (rs6000_emit_allocate_stack): Call
14347 rs6000_emit_probe_stack_range_stack_clash as needed.
14348 (rs6000_emit_probe_stack_range): Add additional argument
14349 to call to gen_probe_stack_range{si,di}.
14350 (output_probe_stack_range): New.
14351 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14352 (output_probe_stack_range_stack_clash): New.
14353 (rs6000_emit_prologue): Emit notes into dump file as requested.
14354 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14355 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14356 Add additional operand and pass it to output_probe_stack_range.
14358 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14360 PR tree-optimization/82163
14361 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14362 (checking_verify_loop_closed_ssa): New parameter.
14363 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14364 (check_loop_closed_ssa_stmt): Delete.
14365 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14366 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14367 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14370 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14372 * brig-builtins.def: Treat HSAIL barrier builtins as
14373 setjmp/longjump style functions.
14375 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14377 * target.def (constant_alignment): New hook.
14378 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14379 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14380 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14381 * doc/tm.texi: Regenerate.
14382 * targhooks.h (default_constant_alignment): Declare.
14383 (constant_alignment_word_strings): Likewise.
14384 * targhooks.c (default_constant_alignment): New function.
14385 (constant_alignment_word_strings): Likewise.
14386 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14387 instead of CONSTANT_ALIGNMENT.
14388 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14389 (force_const_mem): Likewise.
14390 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14391 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14392 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14393 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14394 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14396 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14397 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14398 constant_alignment_word_strings.
14399 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14400 (CONSTANT_ALIGNMENT): Likewise.
14401 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14402 (arm_constant_alignment): New function.
14403 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14404 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14405 constant_alignment_word_strings.
14406 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14407 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14408 constant_alignment_word_strings.
14409 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14410 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14411 (cris_constant_alignment): New function.
14412 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14413 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14414 (epiphany_constant_alignment): New function.
14415 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14416 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14417 constant_alignment_word_strings.
14418 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14419 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14420 constant_alignment_word_strings.
14421 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14422 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14423 constant_alignment_word_strings.
14424 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14425 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14426 * config/i386/i386.c (ix86_constant_alignment): Make static.
14427 Use the same interface as the target hook.
14428 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14429 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14430 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14431 constant_alignment_word_strings.
14432 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14433 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14434 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14435 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14436 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14437 constant_alignment_word_strings.
14438 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14439 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14440 constant_alignment_word_strings.
14441 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14442 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14443 constant_alignment_word_strings.
14444 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14445 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14447 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14448 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14449 * config/mips/mips.c (mips_constant_alignment): New function.
14450 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14451 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14452 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14453 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14454 (mmix_constant_alignment): Make static. Use the same interface
14455 as the target hook.
14456 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14457 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14458 constant_alignment_word_strings.
14459 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14460 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14461 constant_alignment_word_strings.
14462 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14463 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14464 constant_alignment_word_strings.
14465 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14466 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14467 (rs6000_constant_alignment): New function.
14468 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14469 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14470 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14471 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14472 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14473 (rs6000_constant_alignment): New function.
14474 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14475 * config/s390/s390.c (s390_constant_alignment): New function.
14476 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14477 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14478 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14479 constant_alignment_word_strings.
14480 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14481 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14482 (sparc_constant_alignment): New function.
14483 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14484 * config/spu/spu.c (spu_constant_alignment): New function.
14485 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14486 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14487 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14488 constant_alignment_word_strings.
14489 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14490 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14491 constant_alignment_word_strings.
14492 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14493 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14494 constant_alignment_word_strings.
14495 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14496 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14497 (visium_constant_alignment): New function.
14498 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14499 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14500 (xtensa_constant_alignment): New function.
14501 * system.h (CONSTANT_ALIGNMENT): Poison.
14503 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14505 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14506 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14507 (rs6000_builtin_valid_without_lhs): New helper function.
14508 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14509 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14511 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14513 * target.h (vec_perm_indices): Use unsigned short rather than
14515 (auto_vec_perm_indices): Likewise.
14516 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14517 Use unsigned int rather than unsigned char.
14518 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14520 2017-09-25 Richard Biener <rguenther@suse.de>
14522 * cfgloop.h (sort_sibling_loops): Declare.
14523 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14524 (sort_sibling_loops): New function sorting the sibling loop list
14526 * graphite.c (graphite_transform_loops): Sort sibling loops.
14528 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14530 * target.def (vec_perm_const_ok): Change sel parameter to
14532 * optabs-query.c (can_vec_perm_p): Update accordingly.
14533 * doc/tm.texi: Regenerate.
14534 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14535 auto_vec_perm_indices and remove separate nelt field.
14536 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14537 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14538 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14539 (aarch64_expand_vec_perm_const): Update accordingly.
14540 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14541 to vec_perm_indices.
14542 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14543 auto_vec_perm_indices and remove separate nelt field.
14544 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14545 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14546 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14548 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14549 to vec_perm_indices.
14550 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14551 sel to vec_perm_indices.
14552 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14553 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14554 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14556 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14559 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14562 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14563 on the FUNCTION_DECL function context if it has a DIE that is a
14566 2017-09-25 Richard Biener <rguenther@suse.de>
14568 PR tree-optimization/82285
14569 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14572 2017-09-25 Tom de Vries <tom@codesourcery.com>
14576 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14579 2017-09-25 Richard Biener <rguenther@suse.de>
14581 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14582 ISL errors other than isl_error_quota happen. Dump if the
14583 schedule is the same.
14584 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14585 errors instead of aborting inside ISL.
14587 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14590 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14591 of libgcc_eh for m64.
14592 * config/i386/darwin64.h: Likewise.
14594 2017-09-25 Richard Biener <rguenther@suse.de>
14596 PR middle-end/82144
14597 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14598 attribute for incomplete types nor twice for complete ones.
14600 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14603 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14604 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14607 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14610 * config/i386/i386.opt (mprefer-avx256): Use
14611 ix86_target_flags variable.
14612 * config/i386/i386.c (ix86_target_string): Move
14613 -mprefer-avx256 to flag2_opts.
14615 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14617 PR middle-end/35691
14618 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14619 and x != -1 | y != -1 into (x & y) != -1.
14621 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14623 * config.gcc: Add new case statement to set
14624 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14625 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14626 s390x-*-linux* case statements. Added aarch64 to the list of
14627 supported architectures.
14629 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14631 PR tree-optimization/82289
14632 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14633 STMT_VINFO_RELEVANT_P.
14635 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14636 Alan Hayward <alan.hayward@arm.com>
14637 David Sherwood <david.sherwood@arm.com>
14639 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14640 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14641 (extract_range_from_binary_expr_1): Don't call
14642 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14644 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14645 Alan Hayward <alan.hayward@arm.com>
14646 David Sherwood <david.sherwood@arm.com>
14648 * target.def (preferred_vector_alignment): New hook.
14649 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14651 * doc/tm.texi: Regenerate.
14652 * targhooks.h (default_preferred_vector_alignment): Declare.
14653 * targhooks.c (default_preferred_vector_alignment): New function.
14654 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14656 (DR_TARGET_ALIGNMENT): New macro.
14657 (aligned_access_p): Update commentary.
14658 (vect_known_alignment_in_bytes): New function.
14659 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14661 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14662 Calculate the misalignment based on the target alignment rather than
14664 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14665 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14666 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14667 the target alignment, rather than masking the element misalignment
14668 with the number of elements in a vector. Also use the target
14669 alignment when calculating the maximum number of peels.
14670 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14671 instead of TYPE_ALIGN_UNIT.
14672 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14673 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14674 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14675 (vect_create_data_ref_ptr): Likewise.
14676 (vect_setup_realignment): Realign by ANDing with
14677 -DR_TARGET_MISALIGNMENT.
14678 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14679 the number of peels based on DR_TARGET_ALIGNMENT.
14680 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14681 with the guaranteed alignment boundary when deciding whether
14683 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14684 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14685 (ensure_base_align): Remove stmt_info parameter. Get the
14686 target base alignment from DR_TARGET_ALIGNMENT.
14687 (vectorizable_store): Update call accordingly. Interpret
14688 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14690 (vectorizable_load): Likewise.
14692 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14693 Alan Hayward <alan.hayward@arm.com>
14694 David Sherwood <david.sherwood@arm.com>
14696 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14697 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14698 (vect_enhance_data_refs_alignment): Likewise.
14700 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14702 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14703 error. Only quit immediately if parsing is complete.
14704 (BEGIN): Initialize fatal_err and parse_done.
14705 (begin fpu, end fpu): Check number of arguments.
14706 (begin arch, end arch): Likewise.
14707 (begin cpu, end cpu): Likewise.
14708 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14709 (optalias): Likewise.
14711 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14713 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14714 * config/arm/arm-isa.h: Delete. Move definitions to ...
14715 * arm-cpus.in: ... here. Use new feature and fgroup values.
14716 * config/arm/arm.c (arm_option_override): Use lower case for feature
14718 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14719 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14720 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14722 (print_isa_bits_for): New function.
14723 (gen_isa): New function.
14724 (gen_comm_data): Use print_isa_bits_for.
14725 (define feature): New keyword.
14726 (define fgroup): New keyword.
14727 * config/arm/t-arm (TM_H): Remove.
14728 (GTM_H): Add arm-isa.h.
14729 (arm-isa.h): Add rule to generate file.
14730 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14731 case for feature bit names.
14733 2017-09-22 Richard Biener <rguenther@suse.de>
14735 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14737 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14738 print of no dependency loops ...
14739 * graphite.c (graphite_transform_loops): ... here.
14740 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14742 (same_close_phi_node, remove_duplicate_close_phi,
14743 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14744 (canonicalize_loop_closed_ssa): ... here and simplify.
14745 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14746 (optimize_isl): Use dump_printf_loc to tell when we stopped
14747 optimizing because of an ISL timeout.
14749 2017-09-22 Richard Biener <rguenther@suse.de>
14751 PR tree-optimization/82291
14752 * tree-if-conv.c (predicate_mem_writes): Make sure to
14753 remove writes in blocks predicated with false.
14755 2017-09-22 Richard Biener <rguenther@suse.de>
14757 * sese.c: Include cfganal.h.
14758 (if_region_set_false_region): Remove.
14759 (create_if_region_on_edge): Likewise.
14760 (move_sese_in_condition): Re-implement without destroying
14763 2017-09-22 Richard Biener <rguenther@suse.de>
14765 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14766 Verify both BBs contain loop PHI nodes before dispatching to
14767 copy_loop_phi_args.
14768 (graphite_regenerate_ast_isl): Do not recompute dominators,
14769 do not verify three times. Restructure for clarity.
14770 * graphite-scop-detection.c (same_close_phi_node,
14771 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14772 defined_in_loop_p, canonicalize_loop_closed_ssa,
14773 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14774 checking and SSA rewrite, move to ...
14775 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14776 (graphite_initialize): Do not pass in ctx, do not reset the
14777 SCEV cache, compute only dominators.
14778 (graphite_transform_loops): Allocate ISL ctx after
14779 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14780 Maintain post-dominators only around build_scops.
14781 * sese.c (if_region_set_false_region): Make static. Free
14782 and recompute dominators.
14783 (move_sese_in_condition): Assert we don't get called with
14784 post-dominators computed.
14785 * sese.h (if_region_set_false_region): Remove.
14787 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14789 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14790 mode attribute for TARGET_AVX512VL.
14792 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14794 * config/i386/i386.opt (mprefer-avx256): New option.
14795 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14797 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14798 for TARGET_PREFER_AVX256.
14799 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14801 2017-09-21 Jeff Law <law@redhat.com>
14803 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14804 Fix dump output if the only stack space is for pushed registers.
14806 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14808 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14811 2017-09-21 Martin Sebor <msebor@redhat.com>
14814 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14815 code (in C++) or code that triggers warnings.
14817 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14819 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14821 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14823 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14824 * sched-rgn.c: Ditto.
14825 * sel-sched-ir.c: Ditto.
14827 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14829 * toplev.h (set_random_seed): Adjust return type.
14830 * toplev.c (init_local_tick): Move eager initialization of random_seed
14831 to get_random_seed. Adjust comment.
14832 (init_random_seed): Inline to get_random_seed, delete.
14833 (get_random_seed): Initialize random_seed lazily.
14834 (set_random_seed): Do not return previous value.
14835 (print_switch_value): Do not call get_random_seed.
14837 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14839 * cgraph.c (delete_function_version): New, broken out from...
14840 (cgraph_node::delete_function_version): ...here. Rename to
14841 cgraph_node::delete_function_version_by_decl. Update all uses.
14842 (cgraph_node::remove): Call delete_function_version.
14844 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14847 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14848 VAR_DECLs to which addressable non-volatile parameters are mapped
14849 and for id->retvar after the return value assignment. Clear
14850 id->retval and id->retbnd after inlining.
14852 2017-09-21 Richard Biener <rguenther@suse.de>
14854 PR tree-optimization/82276
14855 PR tree-optimization/82244
14856 * tree-vrp.c (build_assert_expr_for): Set
14857 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14859 (remove_range_assertions): Revert earlier change.
14861 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14864 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14866 2017-09-21 Richard Biener <rguenther@suse.de>
14868 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14869 Restore valid IL after code generation errors.
14870 * graphite.c (graphite_transform_loops): Diagnose code
14871 generation issues as MSG_MISSED_OPTIMIZATION and continue
14872 with processing SCOPs.
14874 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14875 Alan Hayward <alan.hayward@arm.com>
14876 David Sherwood <david.sherwood@arm.com>
14878 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14879 rather than choosing between plus_constant and gen_rtx_<CODE>.
14880 * expr.c (emit_push_insn): Likewise.
14881 (expand_expr_real_2): Likewise.
14883 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14884 Alan Hayward <alan.hayward@arm.com>
14885 David Sherwood <david.sherwood@arm.com>
14887 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14889 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14890 Alan Hayward <alan.hayward@arm.com>
14891 David Sherwood <david.sherwood@arm.com>
14893 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14894 calling tree_to_uhwi.
14896 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14897 Alan Hayward <alan.hayward@arm.com>
14898 David Sherwood <david.sherwood@arm.com>
14900 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14901 INTEGER_CST rather than a negative test for ADDR_EXPR.
14903 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14904 Alan Hayward <alan.hayward@arm.com>
14905 David Sherwood <david.sherwood@arm.com>
14907 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14908 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14910 2017-09-21 Richard Biener <rguenther@suse.de>
14912 PR tree-optimization/71351
14913 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14914 graphite_create_new_loop_guard): Remove, fold remaining parts
14916 (translate_isl_ast_node_for): ... here and simplify.
14918 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14921 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14922 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14923 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14924 alternative always use QI mode, for -Os imov (=R,R) alternative
14925 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14928 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14929 Jeff Law <law@redhat.com>
14931 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14932 (allocate_stack_space): New function, partially extracted from
14933 s390_emit_prologue.
14934 (s390_emit_prologue): Track offset to most recent stack probe.
14935 Code to allocate space moved into allocate_stack_space.
14936 Dump actions when no stack is allocated.
14937 (s390_prologue_plus_offset): New function.
14938 (s390_emit_stack_probe): Likewise.
14940 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14942 * common.opt (Wa, Wl, Wp, g, gz=): Add
14944 (gno-column-info): Remove.
14945 (gcolumn-info): Drop RejectNegative.
14946 (gno-): New prefix.
14947 (gno-record-gcc-switches): Remove.
14948 (grecord-gcc-switches): Drop RejectNegative.
14949 (gno-split-dwarf): Remove.
14950 (gsplit-dwarf): Drop RejectNegative.
14951 (gno-strict-dwarf): Remove.
14952 (gstrict-dwarf): Drop RejectNegative.
14953 * config/darwin.opt (gfull, gused): Add RejectNegative.
14954 * dwarf2out.c (gen_producer_string): Drop
14955 gno-record-gcc-switches handler.
14956 * optc-gen.awk: Add g to prefixes with negative forms.
14957 * opts-common.c (remapping_prefix_p): New.
14958 (find_opt): Check it.
14959 (generate_canonical_option): Test g prefix.
14960 (option_map): Add -gno- mapping.
14961 (add_misspelling_candidates): Check remapping_prefix_p.
14963 2017-09-20 Jeff Law <law@redhat.com>
14965 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14966 thinko in stack clash protection support.
14968 * explow.c (compute_stack_clash_protection_loop_data): Use
14969 CONST_INT_P instead of explicit test. Verify object is a
14970 CONST_INT_P before looking at INTVAL.
14971 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14972 instead of explicit test.
14974 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14977 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14978 address instead of to r1 and r11.
14980 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14982 * config.gcc: Support "knm".
14983 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14984 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14986 * config/i386/i386.c (m_KNM): Define.
14987 (processor_target_table): Add "knm".
14989 (ix86_option_override_internal): Add "knm".
14990 (ix86_issue_rate): Add PROCESSOR_KNM.
14991 (ix86_adjust_cost): Ditto.
14992 (ia32_multipass_dfa_lookahead): Ditto.
14993 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14994 (fold_builtin_cpu): Add M_INTEL_KNM.
14995 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14996 (processor_type): Add PROCESSOR_KNM.
14997 * config/i386/x86-tune.def: Add m_KNM.
14998 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
15000 2017-09-20 Richard Biener <rguenther@suse.de>
15002 PR tree-optimization/80213
15003 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
15004 are allowed in empty BBs as well.
15005 (canonicalize_loop_closed_ssa): Also look for other complex
15007 (scop_detection::get_sese): Include the loop-closed PHI block
15009 (scop_detection::merge_sese): Remove code adding extra blocks.
15010 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
15011 (build_scops): Assert the final returned scop is invalid.
15013 2017-09-20 Richard Biener <rguenther@suse.de>
15015 PR tree-optimization/82264
15016 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
15018 (vn_phi_lookup): Likewise.
15019 (vn_phi_insert): Likewise.
15021 2017-09-20 Jakub Jelinek <jakub@redhat.com>
15023 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
15024 that fits into uhwi or shwi, add DW_AT_const_value regardless
15025 of early_dwarf without going through RTL, using add_AT_unsigned
15028 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
15029 (DEBUG_LTO_ABBREV_SECTION): Likewise.
15030 (DEBUG_LTO_MACINFO_SECTION): Likewise.
15031 (DEBUG_MACRO_SECTION): Likewise.
15032 (DEBUG_LTO_MACRO_SECTION): Likewise.
15033 (DEBUG_STR_DWO_SECTION): Likewise.
15034 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
15035 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
15036 (DEBUG_LTO_DWO_LINE_SECTION): Define.
15037 (DEBUG_LTO_LINE_STR_SECTION): Define.
15038 (init_sections_and_labels): Initialize debug_line_str_section
15039 variable. Initialize debug_loc_section for -gdwarf-5 to
15040 DEBUG_LOCLISTS_SECTION. Formatting fixes.
15042 2017-09-20 Richard Biener <rguenther@suse.de>
15044 * graphite-sese-to-poly.c (extract_affine): Properly handle
15045 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
15047 2017-09-20 Richard Biener <rguenther@suse.de>
15049 PR tree-optimization/81373
15050 * graphite-scop-detection.c (build_cross_bb_scalars_def):
15051 Force SESE live-out defs to be handled even if they are
15054 2017-09-19 Jeff Law <law@redhat.com>
15056 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
15057 nothing for stack adjustments with REG_STACK_CHECK.
15058 * sched-deps.c (parse_add_or_inc): Reject insns with
15059 REG_STACK_CHECK from dependency breaking.
15060 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15061 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15062 * reg-notes.def (STACK_CHECK): New note.
15064 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15065 (ix86_expand_prologue): Dump stack clash info as needed.
15066 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15068 * function.c (dump_stack_clash_frame_info): New function.
15069 * function.h (dump_stack_clash_frame_info): Prototype.
15070 (enum stack_clash_probes): New enum.
15072 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15073 flag_stack_clash_protection.
15074 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15075 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15076 (arm_frame_pointer_required): Likewise.
15077 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15078 (ia64_expand_prologue): Likewise.
15079 * config/mips/mips.c (mips_expand_prologue): Likewise.
15080 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15081 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15082 (sparc_flat_expand_prologue): Likewise.
15083 * config/spu/spu.c (spu_expand_prologue): Likewise.
15085 * explow.c: Include "params.h".
15086 (anti_adjust_stack_and_probe_stack_clash): New function.
15087 (get_stack_check_protect): Likewise.
15088 (compute_stack_clash_protection_loop_data): Likewise.
15089 (emit_stack_clash_protection_loop_start): Likewise.
15090 (emit_stack_clash_protection_loop_end): Likewise.
15091 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15092 Use anti_adjust_stack_and_probe_stack_clash.
15093 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15094 (emit_stack_clash_protection_loop_start): Likewise.
15095 (emit_stack_clash_protection_loop_end): Likewise.
15096 * rtl.h (get_stack_check_protect): Prototype.
15097 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15098 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15099 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15101 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15103 * doc/tm.texi: Rebuilt.
15104 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15105 get_stack_check_protect.
15106 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15107 * config/arm/arm.c (arm_expand_prologue): Likewise.
15108 (arm_frame_pointer_required): Likewise.
15109 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15110 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15111 * config/mips/mips.c (mips_expand_prologue): Likewise.
15112 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15113 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15114 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15115 (sparc_flat_expand_prologue): Likewise.
15117 * common.opt (-fstack-clash-protection): New option.
15118 * flag-types.h (enum stack_check_type): Note difference between
15119 -fstack-check= and -fstack-clash-protection.
15120 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15121 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15122 * toplev.c (process_options): Issue warnings/errors for cases
15123 not handled with -fstack-clash-protection.
15124 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15125 (-fstack-check): Note additional problem with -fstack-check=generic.
15126 Note that -fstack-check is primarily for Ada and refer users
15127 to -fstack-clash-protection for stack-clash-protection.
15128 Document new params for stack clash protection.
15130 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15132 * config/i386/i386.c (ix86_split_long_move): Do not handle
15133 address used for LEA in a special way.
15135 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15137 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15139 2017-09-19 Martin Sebor <msebor@redhat.com>
15142 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15143 of incompatible types.
15145 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15147 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15148 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15149 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15150 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15152 2017-09-19 Richard Biener <rguenther@suse.de>
15154 PR tree-optimization/82244
15155 * tree-vrp.c (remove_range_assertions): Do not propagate
15156 a constant to abnormals but replace the assert with a copy.
15158 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15160 PR rtl-optimization/57878
15161 PR rtl-optimization/68988
15162 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15163 avoidance test involving non_reload_pseudos. Move frequency test
15164 below the general fragmentation avoidance test.
15166 2017-09-19 Richard Biener <rguenther@suse.de>
15168 PR tree-optimization/69728
15169 * graphite-sese-to-poly.c (schedule_error): New global.
15170 (add_loop_schedule): Handle empty domain by failing the
15172 (build_original_schedule): Handle schedule_error.
15174 2017-09-19 Richard Biener <rguenther@suse.de>
15176 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15177 Do not iterate to sibling loops but only to siblings of inner
15180 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15183 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15184 that the registers are different.
15186 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15188 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15189 to processor_model and "amdfam17h" to arch_names_table.
15190 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15192 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15195 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15197 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15198 Alan Hayward <alan.hayward@arm.com>
15199 David Sherwood <david.sherwood@arm.com>
15201 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15203 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15205 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15206 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15207 vect_schedule_slp_instance.
15208 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15209 Update call to vect_slp_analyze_node_operations. Simplify return
15211 (vect_slp_analyze_bb_1): Update call accordingly.
15212 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15213 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15214 (vect_schedule_slp): Update call accordingly.
15216 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15217 Alan Hayward <alan.hayward@arm.com>
15218 David Sherwood <david.sherwood@arm.com>
15220 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15221 with types that aren't in fact scalar.
15223 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15225 * tree-vect-slp.c (vect_record_max_nunits): New function,
15227 (vect_build_slp_tree_1): ...here.
15228 (vect_build_slp_tree_2): Call it for phis too.
15230 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15232 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15233 to vect_get_vec_def_for_operand when getting the mask operand.
15235 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15236 Alan Hayward <alan.hayward@arm.com>
15237 David Sherwood <david.sherwood@arm.com>
15239 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15242 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15243 Alan Hayward <alan.hayward@arm.com>
15244 David Sherwood <david.sherwood@arm.com>
15246 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15247 calculation for vector booleans.
15249 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15250 Alan Hayward <alan.hayward@arm.com>
15251 David Sherwood <david.sherwood@arm.com>
15253 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15255 (vect_transform_stmt): ...here.
15256 (vect_analyze_stmt): Use it instead of calling
15257 vectorizable_live_operation directly.
15259 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15261 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15262 non-SIMT targets in acc vector loops.
15264 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15266 * configure.ac: Add arc and check if assembler supports gdwarf2.
15267 * configure: Regenerate.
15269 2017-09-18 Richard Biener <rguenther@suse.de>
15271 PR tree-optimization/82220
15272 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15273 epilogue niters from the min_profitable_iters compute.
15275 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15278 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15279 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15280 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15282 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15285 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15286 switching to a new text section.
15288 2017-09-18 Richard Biener <rguenther@suse.de>
15290 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15292 (build_alias_set): Reject aliases with no access function.
15294 2017-09-18 Richard Biener <rguenther@suse.de>
15296 PR tree-optimization/79622
15297 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15299 (build_cross_bb_scalars_use): Likewise.
15301 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15303 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15305 2017-09-18 Alan Modra <amodra@gmail.com>
15308 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15309 stack_pointer_rtx for count 0. Update comments. Break up
15310 large rtl expression.
15312 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15314 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15315 Increase to 20 bytes.
15316 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15317 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15318 or avx version of the stub.
15320 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15323 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15324 compute the minimum stack alignment. Also update preferred stack
15325 boundary for leaf functions.
15327 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15329 PR tree-optimization/82228
15330 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15333 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15335 * common/config/nds32/nds32-common.c
15336 (nds32_option_optimization_table): Refine formatting.
15337 (nds32_option_optimization_table): Use -fsched-pressure and
15338 -fomit-frame-pointer for specific optimization level.
15340 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15342 * config/nds32/nds32.c: Refine formatting and comments.
15343 * config/nds32/nds32.h: Likewise.
15344 * config/nds32/nds32.md: Likewise.
15345 * config/nds32/nds32-cost.c: Likewise.
15346 * config/nds32/nds32-isr.c: Likewise.
15347 * config/nds32/nds32-md-auxiliary.c: Likewise.
15348 * config/nds32/nds32-multiple.md: Likewise.
15349 * config/nds32/nds32-predicates.c: Likewise.
15351 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15352 Jakub Jelinek <jakub@redhat.com>
15354 Add support for -std=c++2a.
15355 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15357 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15359 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15362 * doc/extend.texi (Common Function Attributes): Add
15363 references to ARM, AArch64, and S/390 specific attributes.
15364 (Function Specific Option Pragmas): Add AArch64 and S/390
15365 to list of back ends that support the target pragma.
15367 2017-09-15 Nathan Sidwell <nathan@acm.org>
15369 * doc/standards.texi: Fix C++17 description. Update URLs for
15372 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15374 * common.opt (Wcast-align=strict): New warning option.
15375 * doc/invoke.texi: Document -Wcast-align=strict.
15377 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15379 * cgraph.h (cgraph_thunk_info): Add comments.
15380 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15381 assert for VIRTUAL_* arguments stricter.
15383 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15385 PR tree-optimization/71026
15386 * match.pd: Move RDIV patterns from fold-const.c
15387 * fold-const.c (distribute_real_division): Removed.
15388 (fold_binary_loc): Remove calls to distribute_real_divison.
15390 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15392 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15393 c++1z and gnu++1z as deprecated. Change other references to
15394 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15395 Change -Wc++1z-compat to -Wc++17-compat.
15396 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15397 * dwarf2out.c (highest_c_language): Handle C++17.
15398 (gen_compile_unit_die): Likewise.
15400 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15402 PR rtl-optimization/82192
15403 * combine.c (make_extraction): Don't look through non-paradoxical
15404 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15407 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15408 Alan Hayward <alan.hayward@arm.com>
15409 David Sherwood <david.sherwood@arm.com>
15411 * target.def (function_arg_offset): New hook.
15412 * targhooks.h (default_function_arg_offset): Declare.
15413 * targhooks.c (default_function_arg_offset): New function.
15414 * function.c (locate_and_pad_parm): Use
15415 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15416 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15417 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15418 * doc/tm.texi: Regenerate.
15419 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15420 * config/spu/spu.c (spu_function_arg_offset): New function.
15421 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15422 * system.h (FUNCTION_ARG_OFFSET): Poison.
15424 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15425 Alan Hayard <alan.hayward@arm.com>
15426 David Sherwood <david.sherwood@arm.com>
15428 * target.def (truly_noop_truncation): New hook.
15429 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15430 than TRULY_NOOP_TRUNCATION.
15431 * hooks.h (hook_bool_uint_uint_true): Declare.
15432 * hooks.c (hook_bool_uint_uint_true): New function.
15433 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15434 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15435 * doc/tm.texi: Regenerate.
15436 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15437 rather than TRULY_NOOP_TRUNCATION in comments.
15438 (simplify_comparison): Likewise.
15439 (record_truncated_value): Likewise.
15440 * expmed.c (extract_bit_field_1): Likewise.
15441 (extract_split_bit_field): Likewise.
15442 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15443 instead of TRULY_NOOP_TRUNCATION.
15444 * function.c (assign_parm_setup_block): Likewise.
15445 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15446 * rtlhooks.c: Include target.h.
15447 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15448 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15449 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15450 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15451 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15452 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15453 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15454 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15455 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15456 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15457 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15458 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15459 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15460 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15461 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15462 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15463 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15464 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15465 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15466 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15467 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15468 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15469 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15470 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15471 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15472 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15473 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15474 rather than TRULY_NOOP_TRUNCATION in comments.
15475 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15476 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15477 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15478 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15479 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15480 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15481 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15482 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15483 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15484 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15485 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15486 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15487 rather than TRULY_NOOP_TRUNCATION in comments.
15488 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15489 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15490 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15491 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15492 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15493 TRULY_NOOP_TRUNCATION condition.
15494 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15495 (TRULY_NOOP_TRUNCATION): Delete.
15496 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15497 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15498 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15499 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15500 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15501 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15502 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15503 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15504 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15505 rather than TRULY_NOOP_TRUNCATION in comments.
15506 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15507 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15508 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15509 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15510 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15511 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15513 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15516 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15517 (*cmp_ior): Likewise.
15518 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15519 (*ior_scc_scc_cmp): Likewise.
15520 (*and_scc_scc): Likewise.
15521 (*and_scc_scc_cmp): Likewise.
15523 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15524 Alan Hayard <alan.hayward@arm.com>
15525 David Sherwood <david.sherwood@arm.com>
15527 * target.def (can_change_mode_class): New hook.
15528 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15529 (hard_regno_nregs): Likewise.
15530 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15531 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15532 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15533 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15534 (LOAD_EXTEND_OP): Update accordingly.
15535 * doc/tm.texi: Regenerate.
15536 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15537 CANNOT_CHANGE_MODE_CLASS.
15538 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15539 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15540 * combine.c (simplify_set): Update accordingly.
15541 * emit-rtl.c (validate_subreg): Likewise.
15542 * recog.c (general_operand): Likewise.
15543 * regcprop.c (mode_change_ok): Likewise.
15544 * reload1.c (choose_reload_regs): Likewise.
15545 (inherit_piecemeal_p): Likewise.
15546 * rtlanal.c (simplify_subreg_regno): Likewise.
15547 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15548 instead of CANNOT_CHANGE_MODE_CLASS.
15549 (reload_cse_simplify_operands): Likewise.
15550 * reload.c (push_reload): Use targetm.can_change_mode_class
15551 instead of CANNOT_CHANGE_MODE_CLASS.
15552 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15553 REG_CANNOT_CHANGE_MODE_P.
15554 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15555 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15556 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15557 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15558 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15559 (arm_can_change_mode_class): New function.
15560 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15561 than CANNOT_CHANGE_MODE_CLASS in comments.
15562 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15563 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15564 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15565 (ix86_can_change_mode_class): ...this new function, inverting the
15566 sense of the return value.
15567 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15568 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15569 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15570 (ia64_can_change_mode_class): New function.
15571 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15572 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15573 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15574 (m32c_can_change_mode_class): ...this new function, inverting the
15575 sense of the return value.
15576 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15577 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15578 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15579 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15580 (mips_can_change_mode_class): ...this new function, inverting the
15581 sense of the return value.
15582 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15583 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15584 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15585 (msp430_can_change_mode_class): New function.
15586 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15587 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15588 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15589 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15590 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15591 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15592 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15593 (pa_cannot_change_mode_class): Replace with...
15594 (pa_can_change_mode_class): ...this new function, inverting the
15595 sense of the return value.
15596 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15597 than CANNOT_CHANGE_MODE_CLASS in comments.
15598 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15599 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15600 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15601 (pdp11_cannot_change_mode_class): Replace with...
15602 (pdp11_can_change_mode_class): ...this new function, inverting the
15603 sense of the return value.
15604 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15605 * config/powerpcspe/powerpcspe-protos.h
15606 (rs6000_cannot_change_mode_class_ptr): Delete.
15607 * config/powerpcspe/powerpcspe.c
15608 (rs6000_cannot_change_mode_class_ptr): Delete.
15609 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15610 (rs6000_option_override_internal): Assign to
15611 targetm.can_change_mode_class instead of
15612 rs6000_cannot_change_mode_class_ptr.
15613 (rs6000_cannot_change_mode_class): Replace with...
15614 (rs6000_can_change_mode_class): ...this new function, inverting the
15615 sense of the return value.
15616 (rs6000_debug_cannot_change_mode_class): Replace with...
15617 (rs6000_debug_can_change_mode_class): ...this new function.
15618 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15619 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15620 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15621 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15622 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15624 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15625 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15626 (rs6000_option_override_internal): Assign to
15627 targetm.can_change_mode_class instead of
15628 rs6000_cannot_change_mode_class_ptr.
15629 (rs6000_cannot_change_mode_class): Replace with...
15630 (rs6000_can_change_mode_class): ...this new function, inverting the
15631 sense of the return value.
15632 (rs6000_debug_cannot_change_mode_class): Replace with...
15633 (rs6000_debug_can_change_mode_class): ...this new function.
15634 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15635 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15636 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15637 (s390_can_change_mode_class): ...this new function, inverting the
15638 sense of the return value.
15639 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15640 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15641 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15642 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15643 (sh_cannot_change_mode_class): Replace with...
15644 (sh_can_change_mode_class): ...this new function, inverting the
15645 sense of the return value.
15646 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15647 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15648 (sparc_can_change_mode_class): New function.
15649 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15650 * config/spu/spu.c (spu_can_change_mode_class): New function.
15651 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15652 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15653 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15654 (visium_can_change_mode_class): New function.
15655 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15657 2017-09-15 Richard Biener <rguenther@suse.de>
15659 PR tree-optimization/82217
15660 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15661 but not undefined case.
15663 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15666 * postreload.c (reload_cse_simplify_operands): Skip
15667 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15669 2017-09-15 Richard Biener <rguenther@suse.de>
15671 PR tree-optimization/68823
15672 * graphite-scop-detection.c (build_alias_set): If we have a
15673 possible dependence check whether we can handle them by just
15674 looking at the DRs DR_ACCESS_FNs.
15675 (build_scops): If build_alias_set fails, fail the SCOP.
15677 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15679 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15680 to support float128 built-in functions that require the ISA 3.0
15682 (BU_FLOAT128_3_HW): Likewise.
15683 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15684 built-in functions.
15685 (FMAF128): Likewise.
15687 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15688 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15689 floating point instructions.
15690 (rs6000_invalid_builtin): Likewise.
15691 (rs6000_builtin_mask_names): Likewise.
15692 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15693 (RS6000_BTM_FLOAT128_HW): Likewise.
15694 (RS6000_BTM_COMMON): Likewise.
15695 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15697 * doc/extend.texi (RS/6000 built-in functions): Document the
15698 IEEE 128-bit floating point square root and fused multiply-add
15699 built-in functions.
15701 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15703 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15704 reg (r2) isn't in the set of registers defined in the prologue.
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 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15711 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15712 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15713 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15714 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15716 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15717 max_vectorization_factor.
15718 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15720 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15721 Alan Hayward <alan.hayward@arm.com>
15722 David Sherwood <david.sherwood@arm.com>
15724 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15725 (vect_worthwhile_without_simd_p): Declare.
15726 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15727 (vectorizable_reduction): Use it.
15728 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15729 (vectorizable_operation): Likewise.
15731 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15732 Alan Hayward <alan.hayward@arm.com>
15733 David Sherwood <david.sherwood@arm.com>
15735 * tree-vectorizer.h (vect_get_num_copies): New function.
15736 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15737 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15738 (vectorizable_induction): Likewise.
15739 (vectorizable_live_operation): Likewise.
15740 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15741 (vectorizable_bswap): Likewise.
15742 (vectorizable_call): Likewise.
15743 (vectorizable_conversion): Likewise.
15744 (vectorizable_assignment): Likewise.
15745 (vectorizable_shift): Likewise.
15746 (vectorizable_operation): Likewise.
15747 (vectorizable_store): Likewise.
15748 (vectorizable_load): Likewise.
15749 (vectorizable_condition): Likewise.
15750 (vectorizable_comparison): Likewise.
15751 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15753 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15754 Alan Hayward <alan.hayward@arm.com>
15755 David Sherwood <david.sherwood@arm.com>
15757 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15758 of vect_init_vector.
15760 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15761 Alan Hayward <alan.hayward@arm.com>
15762 David Sherwood <david.sherwood@arm.com>
15764 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15765 an inline wrapper that provides a location.
15766 (gimple_build_vector): Likewise.
15767 * gimple-fold.c (gimple_build_vector_from_val): New function.
15768 (gimple_build_vector): Likewise.
15769 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15770 functions to build the initial value. Always return a gimple value.
15771 (get_initial_defs_for_reduction): Likewise. Only compute
15773 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15774 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15775 (vectorizable_induction): Use gimple_build_vector rather than
15778 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15779 Alan Hayward <alan.hayward@arm.com>
15780 David Sherwood <david.sherwood@arm.com>
15782 * target.h (vec_perm_indices): New typedef.
15783 (auto_vec_perm_indices): Likewise.
15784 * optabs-query.h: Include target.h
15785 (can_vec_perm_p): Take a vec_perm_indices *.
15786 * optabs-query.c (can_vec_perm_p): Likewise.
15787 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15788 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15789 * tree-vect-generic.c (lower_vec_perm): Likewise.
15790 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15791 (vect_grouped_load_supported): Likewise.
15792 (vect_shift_permute_load_chain): Likewise.
15793 (vect_permute_store_chain): Use auto_vec_perm_indices.
15794 (vect_permute_load_chain): Likewise.
15795 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15796 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15797 Update uses of can_vec_perm_p.
15798 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15799 mode with a number of elements. Take a vec_perm_indices *.
15800 (vect_create_epilog_for_reduction): Update accordingly.
15801 Use auto_vec_perm_indices.
15802 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15803 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15804 (vect_transform_slp_perm_load): Likewise.
15805 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15806 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15807 (vect_gen_perm_mask_checked): Likewise.
15808 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15809 (vect_gen_perm_mask_checked): Likewise.
15810 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15811 (vectorizable_store): Likewise.
15812 (vectorizable_load): Likewise.
15813 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15814 (vectorizable_bswap): Likewise.
15816 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15817 Alan Hayward <alan.hayward@arm.com>
15818 David Sherwood <david.sherwood@arm.com>
15820 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15821 * tree.c (build_vector): Likewise.
15822 (build_vector_from_ctor): Update accordingly.
15823 (build_vector_from_val): Likewise.
15824 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15825 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15826 * tree-vect-generic.c (add_rshift): Likewise.
15827 (expand_vector_divmod): Likewise.
15828 (optimize_vector_constructor): Likewise.
15829 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15830 (vect_transform_slp_perm_load): Likewise.
15831 (vect_schedule_slp_instance): Likewise.
15832 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15833 (vectorizable_call): Likewise.
15834 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15835 * expmed.c (make_tree): Likewise.
15836 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15837 a vector passed to build_vector.
15838 (fold_convert_const): Likewise.
15839 (exact_inverse): Likewise.
15840 (fold_ternary_loc): Likewise.
15841 (fold_relational_const): Likewise.
15842 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15843 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15844 (const_unop): Likewise. Store the reduction accumulator in a
15845 variable rather than an array.
15846 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15847 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15848 the new vector, rather than constructing it after the input arrays.
15849 (native_interpret_vector): Use auto_vec<tree> when building
15850 a vector passed to build_vector. Add elements in order.
15851 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15852 auto_vec<tree> when building a vector passed to build_vector.
15853 (vect_create_epilog_for_reduction): Likewise.
15854 (vectorizable_induction): Likewise.
15855 (get_initial_def_for_reduction): Likewise. Fix indentation of
15857 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15859 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15860 passed to build_vector.
15862 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15863 Alan Hayward <alan.hayward@arm.com>
15864 David Sherwood <david.sherwood@arm.com>
15866 * tree-core.h (tree_base::u): Add an "nelts" field.
15867 (tree_vector): Use VECTOR_CST_NELTS as the length.
15868 * tree.c (tree_size): Likewise.
15869 (make_vector): Initialize VECTOR_CST_NELTS.
15870 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15871 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15872 TYPE_VECTOR_SUBPARTS.
15873 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15874 as the number of units, setting it from VECTOR_CST_NELTS.
15875 (const_vector_from_tree): Likewise.
15876 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15877 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15878 (fold_negate_expr_1): Likewise.
15879 (fold_convert_const): Likewise.
15880 (const_binop): Likewise. Differentiate the number of output and
15882 (const_unop): Likewise.
15883 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15884 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15885 in cases that did the opposite.
15887 2017-09-14 Richard Biener <rguenther@suse.de>
15889 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15892 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15894 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15896 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15899 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15900 if and where to split a bb, except for splitting before debug insn
15901 sequences followed by non-label real insn. Delete debug insns
15902 in between basic blocks.
15904 * combine.c (make_compound_operation_int): Formatting fixes.
15906 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15907 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15908 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15909 * config/sol2.h (LINK_EH_SPEC): Likewise.
15910 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15911 * config/s390/linux.h (LINK_SPEC): Likewise.
15912 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15913 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15914 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15915 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15916 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15917 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15918 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15919 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15921 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15923 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15925 (ENDFILE_LINUX_SPEC): Likewise.
15926 (LINK_EH_SPEC): Likewise.
15927 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15928 (LINK_OS_LINUX_SPEC32): Likewise.
15929 (LINK_OS_LINUX_SPEC64): Likewise.
15930 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15931 (LINK_OS_LINUX_SPEC): Likewise.
15933 2017-09-13 Martin Liska <mliska@suse.cz>
15935 PR middle-end/82154
15936 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15937 CASE_HIGH is NULL_TREE.
15939 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15940 Alan Hayward <alan.hayward@arm.com>
15941 David Sherwood <david.sherwood@arm.com>
15943 * target.def (secondary_memory_needed): New hook.
15944 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15945 instead of SECONDARY_MEMORY_NEEDED.
15946 (secondary_memory_needed_mode): Likewise.
15947 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15948 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15949 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15950 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15951 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15952 * doc/tm.texi: Regenerate.
15953 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15954 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15955 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15956 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15957 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15958 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15959 mode argument first and change the reg_class arguments to reg_class_t.
15960 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15961 Make static. Update the call to inline_secondary_memory_needed.
15962 (ix86_register_move_cost): Update the call to
15963 inline_secondary_memory_needed.
15964 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15965 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15967 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15968 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15970 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15971 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15972 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15973 and match hook interface. Add comment from mips.h.
15974 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15975 * config/mmix/mmix.md (truncdfsf2): Refer to
15976 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15978 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15979 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15980 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15981 (pa_secondary_memory_needed): New function.
15982 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15983 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15984 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15985 (pdp11_secondary_memory_needed): Make static and match hook interface.
15986 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15987 * config/powerpcspe/powerpcspe-protos.h
15988 (rs6000_secondary_memory_needed_ptr): Delete.
15989 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15991 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15992 (rs6000_option_override_internal): Assign to
15993 targetm.secondary_memory_needed rather than
15994 rs6000_secondary_memory_needed_ptr.
15995 (rs6000_secondary_memory_needed): Match hook interface.
15996 (rs6000_debug_secondary_memory_needed): Likewise.
15997 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15998 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15999 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
16000 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16001 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
16002 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
16004 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
16005 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16006 (rs6000_option_override_internal): Assign to
16007 targetm.secondary_memory_needed rather than
16008 rs6000_secondary_memory_needed_ptr.
16009 (rs6000_secondary_memory_needed): Match hook interface.
16010 (rs6000_debug_secondary_memory_needed): Likewise.
16011 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
16012 * config/s390/s390.c (s390_secondary_memory_needed): New function.
16013 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16014 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
16015 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16016 (sparc_secondary_memory_needed): New function.
16017 * lra-constraints.c (check_and_process_move): Refer to
16018 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16020 (curr_insn_transform): Likewise.
16021 (process_alt_operands): Use targetm.secondary_memory_needed
16022 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16023 (check_secondary_memory_needed_p): Likewise.
16024 (choose_split_class): Likewise.
16025 * reload.c: Unconditionally include code that was previously
16026 conditional on SECONDARY_MEMORY_NEEDED.
16027 (push_secondary_reload): Use targetm.secondary_memory_needed
16028 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16029 (push_reload): Likewise.
16030 * reload1.c: Unconditionally include code that was previously
16031 conditional on SECONDARY_MEMORY_NEEDED.
16032 (choose_reload_regs): Use targetm.secondary_memory_needed
16033 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16034 (gen_reload): Likewise.
16035 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
16037 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16038 Alan Hayward <alan.hayward@arm.com>
16039 David Sherwood <david.sherwood@arm.com>
16041 * target.def (secondary_memory_needed_mode): New hook:
16042 * targhooks.c (default_secondary_memory_needed_mode): Declare.
16043 * targhooks.h (default_secondary_memory_needed_mode): New function.
16044 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
16045 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
16046 * doc/tm.texi: Regenerate.
16047 * lra-constraints.c (check_and_process_move): Use
16048 targetm.secondary_memory_needed_mode instead of
16049 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
16050 (curr_insn_transform): Likewise.
16051 * reload.c (get_secondary_mem): Likewise.
16052 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16053 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
16055 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16056 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16057 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
16058 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16059 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16061 * config/powerpcspe/powerpcspe-protos.h
16062 (rs6000_secondary_memory_needed_mode): Delete.
16063 * config/powerpcspe/powerpcspe.c
16064 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16065 (rs6000_secondary_memory_needed_mode): Make static.
16066 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16067 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16069 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16071 (rs6000_secondary_memory_needed_mode): Make static.
16072 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16073 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16074 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16075 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16076 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16078 (sparc_secondary_memory_needed_mode): New function.
16079 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16081 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16083 * config/aarch64/constraints.md (Umq): New constraint.
16084 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16086 (mov<mode>): Update condition.
16088 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16090 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16091 when bitposition is the same.
16093 2017-09-13 Richard Biener <rguenther@suse.de>
16095 * dwarf2out.c (output_die_symbol): Remove.
16096 (output_die): Do not output a DIEs symbol.
16098 2017-09-13 Richard Biener <rguenther@suse.de>
16100 PR middle-end/82128
16101 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16102 default-def to avoid breaking iterator update with the weird
16103 interaction with cgraph_update_edges_for_call_stmt_node.
16105 2017-09-13 Richard Biener <rguenther@suse.de>
16107 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16108 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16109 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16110 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16112 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16114 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16115 Disable pc relative literal load irrespective of
16116 TARGET_FIX_ERR_A53_84341 for default.
16118 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16120 * config/sparc/sparc.c (output_return): Output the source location of
16121 the insn in the delay slot, if any.
16122 (output_sibcall): Likewise.
16124 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16127 * common.opt (-static-pie): New alias.
16128 (shared): Negate static-pie.
16129 (-no-pie): Update help text.
16131 (static-pie): New option.
16132 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16133 -static-pie support.
16134 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16135 (LINK_EH_SPEC): Likewise.
16136 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16137 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16138 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16139 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16140 (init_gcc_specs): Likewise.
16141 (init_spec): Likewise.
16142 (display_help): Update help message for -pie.
16143 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16146 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16148 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16149 (movdi_aarch64): Likewise.
16150 (movti_aarch64): Likewise.
16152 2017-09-12 Simon Wright <simon@pushface.org>
16155 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16156 calculation of the minor version, always output as 0.
16158 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16161 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16162 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16163 on it early, rather than manual conversion late. For
16164 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16165 instead of performing manual conversion.
16167 2017-09-12 Carl Love <cel@us.ibm.com>
16169 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16170 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16172 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16173 VMULOSW): Add definitions.
16174 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16175 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16176 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16177 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16178 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16180 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16182 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16184 (movti_aarch64): Likewise.
16185 (movdf_aarch64): Likewise.
16186 (movtf_aarch64): Likewise.
16187 (load_pairdi): Likewise.
16188 (store_pairdi): Likewise.
16189 (load_pairdf): Likewise.
16190 (store_pairdf): Likewise.
16191 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16192 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16193 (ldr_got_small_<mode>): Likewise.
16194 (ldr_got_small_28k_<mode>): Likewise.
16195 (ldr_got_tiny): Likewise.
16196 * config/aarch64/iterators.md (ldst_sz): New.
16197 (ldpstp_sz): Likewise.
16198 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16200 (thunderx_load): Split load_8 to load_16.
16201 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16203 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16204 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16205 (xgene1_store_pair): Split store_8 to store_16.
16206 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16207 (falkor_st_0_st_sd): Split store_8 to store_16.
16209 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16211 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16212 and store1/2/3/4 to store_4/8/12/16.
16213 * config/aarch64/aarch64.md: Update for rename.
16214 * config/arm/arm.md: Likewise.: Likewise.
16215 * config/arm/arm.c: Likewise.
16216 * config/arm/thumb1.md: Likewise.
16217 * config/arm/thumb2.md: Likewise.
16218 * config/arm/vfp.md: Likewise.
16219 * config/arm/arm-generic.md: Likewise.
16220 * config/arm/arm1020e.md: Likewise.
16221 * config/arm/arm1026ejs.md: Likewise.
16222 * config/arm/arm1136jfs.md: Likewise.
16223 * config/arm/arm926ejs.md: Likewise.
16224 * config/arm/cortex-a15.md: Likewise.
16225 * config/arm/cortex-a17.md: Likewise.
16226 * config/arm/cortex-a5.md: Likewise.
16227 * config/arm/cortex-a53.md: Likewise.
16228 * config/arm/cortex-a57.md: Likewise.
16229 * config/arm/cortex-a7.md: Likewise.
16230 * config/arm/cortex-a8.md: Likewise.
16231 * config/arm/cortex-a9.md: Likewise.
16232 * config/arm/cortex-m4.md: Likewise.
16233 * config/arm/cortex-m7.md: Likewise.
16234 * config/arm/cortex-r4.md: Likewise.
16235 * config/arm/exynos-m1.md: Likewise.
16236 * config/arm/fa526.md: Likewise.
16237 * config/arm/fa606te.md: Likewise.
16238 * config/arm/fa626te.md: Likewise.
16239 * config/arm/fa726te.md: Likewise.
16240 * config/arm/fmp626.md: Likewise.
16241 * config/arm/iwmmxt.md: Likewise.
16242 * config/arm/ldmstm.md: Likewise.
16243 * config/arm/marvell-pj4.md: Likewise.
16244 * config/arm/xgene1.md: Likewise.
16245 * config/aarch64/thunderx.md: Likewise.
16246 * config/aarch64/thunderx2t99.md: Likewise.
16247 * config/aarch64/falkor.md: Likewise.
16249 2017-09-12 Martin Liska <mliska@suse.cz>
16251 * attribs.c (private_lookup_attribute): New function.
16252 * attribs.h (private_lookup_attribute): Declared here.
16253 (lookup_attribute): Called from this place.
16255 2017-09-12 Richard Biener <rguenther@suse.de>
16257 PR tree-optimization/82157
16258 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16259 stmts with side-effects.
16261 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16262 Alan Hayward <alan.hayward@arm.com>
16263 David Sherwood <david.sherwood@arm.com>
16265 * target.def (hard_regno_nregs): New hook.
16266 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16267 * targhooks.h (default_hard_regno_nregs): Declare.
16268 * targhooks.c (default_hard_regno_nregs): New function.
16269 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16270 (TARGET_HARD_REGNO_NREGS): ...this hook.
16271 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16272 (CLASS_MAX_NREGS): Likewise.
16273 * doc/tm.texi: Regenerate.
16274 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16275 instead of HARD_REGNO_NREGS.
16276 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16277 HARD_REGNO_NREGS in the comment.
16278 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16279 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16280 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16281 Return an unsigned int.
16282 (TARGET_HARD_REGNO_NREGS): Redefine.
16283 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16284 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16285 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16286 (arc_hard_regno_nregs): New function.
16287 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16288 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16289 (arm_hard_regno_nregs): New function.
16290 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16291 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16292 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16293 (TARGET_HARD_REGNO_NREGS): Redefine.
16294 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16295 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16296 (HARD_REGNO_NREGS): Delete.
16297 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16298 (cr16_hard_regno_nregs): New function.
16299 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16300 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16301 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16302 (cris_hard_regno_nregs): New function.
16303 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16304 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16305 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16306 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16307 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16308 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16309 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16310 (frv_hard_regno_nregs): Make static. Take and return an
16312 (frv_class_max_nregs): Remove outdated copy of documentation.
16313 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16314 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16315 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16316 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16317 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16318 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16319 (TARGET_HARD_REGNO_NREGS): Redefine.
16320 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16321 (CLASS_MAX_NREGS): Update comment.
16322 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16323 (ia64_hard_regno_nregs): New function.
16324 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16325 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16326 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16327 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16328 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16330 (m32c_hard_regno_nregs): Likewise. Make static.
16331 (TARGET_HARD_REGNO_NREGS): Redefine.
16332 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16333 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16334 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16335 (m68k_hard_regno_nregs): New function.
16336 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16337 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16338 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16339 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16340 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16341 Take and return an unsigned int.
16342 (TARGET_HARD_REGNO_NREGS): Redefine.
16343 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16344 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16345 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16346 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16347 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16348 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16349 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16350 (msp430_hard_regno_nregs): Make static. Take and return an
16352 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16353 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16354 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16355 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16356 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16357 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16358 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16359 (TARGET_HARD_REGNO_NREGS): Redefine.
16360 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16361 (PA_HARD_REGNO_NREGS): ...this.
16362 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16363 (PA_HARD_REGNO_NREGS): ...this.
16364 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16365 (pa_hard_regno_nregs): New function.
16366 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16367 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16368 (pdp11_hard_regno_nregs): New function.
16369 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16370 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16371 (rs6000_hard_regno_nregs_hook): New function.
16372 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16373 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16374 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16375 Take and return an unsigned int. Move earlier in file.
16376 (TARGET_HARD_REGNO_NREGS): Redefine.
16377 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16378 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16379 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16380 (rl78_hard_regno_nregs): Make static. Take and return an
16382 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16383 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16384 (rs6000_hard_regno_nregs_hook): New function.
16385 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16386 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16387 (TARGET_HARD_REGNO_NREGS): Redefine.
16388 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16389 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16390 instead of HARD_REGNO_NREGS.
16391 (s390_hard_regno_nregs): New function.
16392 (s390_hard_regno_mode_ok): Add comment from s390.h.
16393 (TARGET_HARD_REGNO_NREGS): Redefine.
16394 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16395 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16396 (sh_hard_regno_nregs): New function.
16397 (sh_pass_in_reg_p): Use it.
16398 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16399 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16400 (sparc_hard_regno_nregs): New function.
16401 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16402 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16403 (spu_function_arg_advance): Use it, supplying a valid register number.
16404 (TARGET_HARD_REGNO_NREGS): Redefine.
16405 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16406 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16407 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16408 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16409 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16410 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16411 (CLASS_MAX_NREGS): Remove copy of old documentation.
16412 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16413 (visium_hard_regno_nregs): New function.
16414 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16415 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16416 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16417 xtensa_hard_regno_nregs): New function.
16418 * system.h (HARD_REGNO_NREGS): Poison.
16420 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16422 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16423 hard_regno_nregs instead of HARD_REGNO_NREGS.
16424 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16425 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16426 (c6x_expand_epilogue): Likewise.
16427 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16428 (frv_read_iacc_argument): Likewise.
16429 * config/sh/sh.c: Include regs.h.
16430 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16431 (regs_used): Likewise.
16432 (output_stack_adjust): Likewise.
16433 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16434 * expmed.c: Include regs.h.
16435 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16436 * ree.c: Include regs.h.
16437 (combine_reaching_defs): Use hard_regno_nregs instead of
16439 (add_removable_extension): Likewise.
16441 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16443 * regs.h (hard_regno_nregs): Turn into a function.
16444 (end_hard_regno): Update accordingly.
16445 * caller-save.c (setup_save_areas): Likewise.
16446 (save_call_clobbered_regs): Likewise.
16447 (replace_reg_with_saved_mem): Likewise.
16448 (insert_restore): Likewise.
16449 (insert_save): Likewise.
16450 * combine.c (can_change_dest_mode): Likewise.
16451 (move_deaths): Likewise.
16452 (distribute_notes): Likewise.
16453 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16454 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16455 (rs6000_split_multireg_move): Likewise.
16456 (rs6000_register_move_cost): Likewise.
16457 (rs6000_memory_move_cost): Likewise.
16458 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16459 (rs6000_split_multireg_move): Likewise.
16460 (rs6000_register_move_cost): Likewise.
16461 (rs6000_memory_move_cost): Likewise.
16462 * cselib.c (cselib_reset_table): Likewise.
16463 (cselib_lookup_1): Likewise.
16464 * emit-rtl.c (set_mode_and_regno): Likewise.
16465 * function.c (aggregate_value_p): Likewise.
16466 * ira-color.c (setup_profitable_hard_regs): Likewise.
16467 (check_hard_reg_p): Likewise.
16468 (calculate_saved_nregs): Likewise.
16469 (assign_hard_reg): Likewise.
16470 (improve_allocation): Likewise.
16471 (calculate_spill_cost): Likewise.
16472 * ira-emit.c (modify_move_list): Likewise.
16473 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16474 (ira_hard_reg_in_set_p): Likewise.
16475 * ira.c (setup_reg_mode_hard_regset): Likewise.
16476 (clarify_prohibited_class_mode_regs): Likewise.
16477 (check_allocation): Likewise.
16478 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16479 (lra_setup_reg_renumber): Likewise.
16480 (setup_try_hard_regno_pseudos): Likewise.
16481 (spill_for): Likewise.
16482 (assign_hard_regno): Likewise.
16483 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16484 * lra-constraints.c (in_class_p): Likewise.
16485 (lra_constraint_offset): Likewise.
16486 (simplify_operand_subreg): Likewise.
16487 (lra_constraints): Likewise.
16488 (split_reg): Likewise.
16489 (split_if_necessary): Likewise.
16490 (invariant_p): Likewise.
16491 (inherit_in_ebb): Likewise.
16492 * lra-lives.c (process_bb_lives): Likewise.
16493 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16494 (get_hard_regs): Likewise.
16495 (do_remat): Likewise.
16496 * lra-spills.c (assign_spill_hard_regs): Likewise.
16497 * mode-switching.c (create_pre_exit): Likewise.
16498 * postreload.c (reload_combine_recognize_pattern): Likewise.
16499 * recog.c (peep2_find_free_register): Likewise.
16500 * regcprop.c (kill_value_regno): Likewise.
16501 (set_value_regno): Likewise.
16502 (copy_value): Likewise.
16503 (maybe_mode_change): Likewise.
16504 (find_oldest_value_reg): Likewise.
16505 (copyprop_hardreg_forward_1): Likewise.
16506 * regrename.c (check_new_reg_p): Likewise.
16507 (regrename_do_replace): Likewise.
16508 * reload.c (push_reload): Likewise.
16509 (combine_reloads): Likewise.
16510 (find_dummy_reload): Likewise.
16511 (operands_match_p): Likewise.
16512 (find_reloads): Likewise.
16513 (find_equiv_reg): Likewise.
16514 (reload_adjust_reg_for_mode): Likewise.
16515 * reload1.c (count_pseudo): Likewise.
16516 (count_spilled_pseudo): Likewise.
16517 (find_reg): Likewise.
16518 (clear_reload_reg_in_use): Likewise.
16519 (free_for_value_p): Likewise.
16520 (allocate_reload_reg): Likewise.
16521 (choose_reload_regs): Likewise.
16522 (reload_adjust_reg_for_temp): Likewise.
16523 (emit_reload_insns): Likewise.
16524 (delete_output_reload): Likewise.
16525 * rtlanal.c (subreg_get_info): Likewise.
16526 * sched-deps.c (sched_analyze_reg): Likewise.
16527 * sel-sched.c (init_regs_for_mode): Likewise.
16528 (mark_unavailable_hard_regs): Likewise.
16529 (choose_best_reg_1): Likewise.
16530 (verify_target_availability): Likewise.
16531 * valtrack.c (dead_debug_insert_temp): Likewise.
16532 * var-tracking.c (track_loc_p): Likewise.
16533 (emit_note_insn_var_location): Likewise.
16534 * varasm.c (make_decl_rtl): Likewise.
16535 * reginfo.c (choose_hard_reg_mode): Likewise.
16536 (init_reg_modes_target): Refer directly to
16537 this_target_regs->x_hard_regno_nregs.
16539 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16541 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16542 instead of hard_regno_nregs.
16544 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16546 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16547 end_hard_regno instead of hard_regno_nregs.
16548 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16549 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16550 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16551 * ira-color.c (improve_allocation): Likewise.
16552 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16553 * lra-lives.c (mark_regno_live): Likewise.
16554 (mark_regno_dead): Likewise.
16555 * lra-remat.c (operand_to_remat): Likewise.
16556 * lra.c (collect_non_operand_hard_regs): Likewise.
16557 * postreload.c (reload_combine_note_store): Likewise.
16558 (move2add_valid_value_p): Likewise.
16559 * reload.c (regno_clobbered_p): Likewise.
16561 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16563 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16565 * config/v850/v850.c (v850_reorg): Likewise.
16566 * reload.c (refers_to_regno_for_reload_p): Likewise.
16567 (find_equiv_reg): Likewise.
16568 * reload1.c (reload_reg_reaches_end_p): Likewise.
16570 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16572 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16574 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16575 * config/arm/arm.c (output_move_neon): Likewise.
16576 (arm_attr_length_move_neon): Likewise.
16577 (neon_split_vcombine): Likewise.
16578 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16579 (c6x_mark_reg_written): Likewise.
16580 (c6x_dwarf_register_span): Likewise.
16581 * config/i386/i386.c (ix86_save_reg): Likewise.
16582 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16583 (rws_access_reg): Likewise.
16584 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16585 * mode-switching.c (create_pre_exit): Likewise.
16586 * ree.c (combine_reaching_defs): Likewise.
16587 (add_removable_extension): Likewise.
16588 * regcprop.c (find_oldest_value_reg): Likewise.
16589 (copyprop_hardreg_forward_1): Likewise.
16590 * reload.c (reload_inner_reg_of_subreg): Likewise.
16591 (push_reload): Likewise.
16592 (combine_reloads): Likewise.
16593 (find_dummy_reload): Likewise.
16594 (reload_adjust_reg_for_mode): Likewise.
16595 * reload1.c (find_reload_regs): Likewise.
16596 (forget_old_reloads_1): Likewise.
16597 (reload_reg_free_for_value_p): Likewise.
16598 (reload_adjust_reg_for_temp): Likewise.
16599 (emit_reload_insns): Likewise.
16600 (delete_output_reload): Likewise.
16601 * sel-sched.c (choose_best_reg_1): Likewise.
16602 (choose_best_pseudo_reg): Likewise.
16604 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16605 Alan Hayward <alan.hayward@arm.com>
16606 David Sherwood <david.sherwood@arm.com>
16608 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16609 * target.def (slow_unaligned_access): New hook.
16610 * targhooks.h (default_slow_unaligned_access): Declare.
16611 * targhooks.c (default_slow_unaligned_access): New function.
16612 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16613 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16614 * doc/tm.texi: Regenerate.
16615 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16616 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16617 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16619 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16620 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16622 (rs6000_slow_unaligned_access): New function.
16623 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16624 (expand_block_compare): Likewise.
16625 (expand_strn_compare): Likewise.
16626 (rs6000_rtx_costs): Likewise.
16627 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16628 (riscv_slow_unaligned_access): Likewise.
16629 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16630 (riscv_slow_unaligned_access_p): ...this and make static.
16631 (riscv_option_override): Update accordingly.
16632 (riscv_slow_unaligned_access): New function.
16633 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16634 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16635 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16636 (rs6000_slow_unaligned_access): New function.
16637 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16638 (rs6000_rtx_costs): Likewise.
16639 * config/rs6000/rs6000-string.c (expand_block_compare)
16640 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16641 of SLOW_UNALIGNED_ACCESS.
16642 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16643 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16644 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16645 of SLOW_UNALIGNED_ACCESS.
16646 * expmed.c (simple_mem_bitfield_p): Likewise.
16647 * expr.c (alignment_for_piecewise_move): Likewise.
16648 (emit_group_load_1): Likewise.
16649 (emit_group_store): Likewise.
16650 (copy_blkmode_from_reg): Likewise.
16651 (emit_push_insn): Likewise.
16652 (expand_assignment): Likewise.
16653 (store_field): Likewise.
16654 (expand_expr_real_1): Likewise.
16655 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16656 * lra-constraints.c (simplify_operand_subreg): Likewise.
16657 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16658 * gimple-ssa-store-merging.c: Likewise in block comment at start
16660 * tree-ssa-strlen.c: Include target.h.
16661 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16662 of SLOW_UNALIGNED_ACCESS.
16663 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16665 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16667 PR rtl-optimization/82185
16668 * expmed.c (emit_store_flag_int): Only test tem if it has been
16671 2017-09-12 Richard Biener <rguenther@suse.de>
16673 PR middle-end/82149
16674 * match.pd ((FTYPE) N CMP CST): Fix typo.
16676 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16678 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16680 (mips_near_type_p): Add 'short_call' attribute as a synonym
16682 * doc/extend.texi (short_call): Document new function attribute.
16684 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16687 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16688 assertion check that in the condition.
16689 (get_atomic_generic_size): Likewise. Before testing if parameter
16690 has pointer type, if it has array type, call for C++
16691 default_conversion to perform array-to-pointer conversion.
16693 2017-09-12 Richard Biener <rguenther@suse.de>
16695 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16696 for operations we cannot scalarize.
16698 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16700 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16701 vectors heap vectors. Clean up comments.
16702 Make visited_bbs a reference.
16703 (profitable_jump_thread_path): Make GC
16704 vectors heap vectors. Clean up comments.
16706 (convert_and_register_jump_thread_path): Make GC vectors heap
16708 (check_subpath_and_update_thread_path): Same. Clean up comments.
16709 Make visited_bbs a reference.
16710 (handle_phi): Abstract common code to to
16711 register_jump_thread_path_if_profitable.
16712 Rename VAR_BB to DEF_BB.
16714 Make GC vectors heap vectors.
16715 Make visited_bbs a reference.
16716 (handle_assignment): Same.
16717 (register_jump_thread_path_if_profitable): New.
16718 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16720 Make GC vectors heap vectors. Clean up comments.
16721 Make visited_bbs a reference.
16722 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16723 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16726 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16729 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16730 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16732 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16734 Revert r251800 and r251799.
16736 2017-09-11 Martin Jambor <mjambor@suse.cz>
16739 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16740 arguments in advance.
16741 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16742 use it to find predecessor edges.
16743 (naive_outof_ssa): Collect vector of predecessors.
16745 2017-09-08 Jason Merrill <jason@redhat.com>
16747 PR c++/70029 - ICE with ref-qualifier and -flto
16748 * langhooks.h (struct lang_hooks_for_types): Add
16749 copy_lang_qualifiers.
16750 * attribs.c (build_type_attribute_qual_variant): Use it.
16751 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16753 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16754 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16756 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16759 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16760 (*mulsi3_sp64): New instruction.
16761 (mulsi3): New expander.
16763 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16765 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16767 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16769 * sancov.c: Include memmodel.h.
16771 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16774 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16777 2017-09-07 Carl Love <cel@us.ibm.com>
16779 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16780 the sldi instruction.
16782 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16784 * sancov.c: Include tm_p.h.
16786 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16789 * output.h (switch_to_other_text_partition): New declaration.
16790 * varasm.c (switch_to_other_text_partition): New function.
16791 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16792 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16793 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16794 to the other text partition before emitting LCL label and switch back
16795 after emitting the word after it.
16797 2017-09-07 Richard Biener <rguenther@suse.de>
16799 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16800 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16801 critical edges here, after loop init.
16802 (pass_data_pre): Remove PROP_no_crit_edges flags.
16803 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16804 for valueization of call args to avoid leaking VN_TOP.
16805 (visit_use): Assert we do not visit default defs.
16806 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16807 Use error_mark_node to more easily detect leaking VN_TOP.
16808 All default-defs are varying, not VN_TOP. Mark them visited.
16809 (run_scc_vn): Make code match comment.
16811 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16813 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16814 OPTION_MASK_FLOAT128_KEYWORD.
16815 (POWERPC_MASKS): Likewise.
16816 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16817 support for the -mfloat128-type option, and make -mfloat128
16818 default on PowerPC Linux systems. Define or undefine
16819 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16820 Define __float128 to be __ieee128 if IEEE 128-bit support is
16821 enabled, or undefine it.
16822 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16823 Delete defining __FLOAT128_TYPE__.
16824 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16825 -mfloat128-type option and make -mfloat128 default on PowerPC
16827 (TARGET_FLOAT128_TYPE): Likewise.
16828 (-mfloat128-type): Likewise.
16829 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16830 Delete the -mfloat128-type option and make -mfloat128 default on
16831 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16832 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16833 128-bit floating point is enabled. Change tests from using
16834 -mfloat128-type to -mfloat128.
16835 (rs6000_mangle_type): Use the correct mangling for the __float128
16836 type even if normal long double is restricted to 64-bits.
16837 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16839 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16840 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16841 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16843 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16844 documentation for -mfloat128.
16846 2017-09-06 Olivier Hainque <hainque@adacore.com>
16848 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16850 2017-09-06 Wish Wu <wishwu007@gmail.com>
16851 Jakub Jelinek <jakub@redhat.com>
16853 * asan.c (initialize_sanitizer_builtins): Add
16854 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16855 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16856 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16857 BT_FN_VOID_UINT64_PTR variables.
16858 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16859 (BT_FN_VOID_UINT16_UINT16): Likewise.
16860 (BT_FN_VOID_UINT32_UINT32): Likewise.
16861 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16862 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16863 (BT_FN_VOID_UINT64_PTR): Likewise.
16864 * common.opt (flag_sanitize_coverage): New variable.
16865 (fsanitize-coverage=trace-pc): Remove.
16866 (fsanitize-coverage=): Add.
16867 * flag-types.h (enum sanitize_coverage_code): New enum.
16868 * fold-const.c (fold_range_test): Disable non-short-circuit
16869 optimization if flag_sanitize_coverage.
16870 (fold_truth_andor): Likewise.
16871 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16872 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16873 (coverage_sanitizer_opts): New array.
16874 (get_closest_sanitizer_option): Add OPTS argument, handle also
16875 OPT_fsanitize_coverage_.
16876 (parse_sanitizer_options): Adjusted to also handle
16877 OPT_fsanitize_coverage_.
16878 (common_handle_option): Add OPT_fsanitize_coverage_.
16879 * sancov.c (instrument_comparison, instrument_switch): New function.
16880 (sancov_pass): Add trace-cmp support.
16881 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16882 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16883 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16884 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16885 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16886 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16887 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16888 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16889 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16890 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16892 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16894 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16895 error. Only quit immediately if parsing is complete.
16896 (BEGIN): Initialize fatal_err and parse_done.
16897 (begin fpu, end fpu): Check number of arguments.
16898 (begin arch, end arch): Likewise.
16899 (begin cpu, end cpu): Likewise.
16900 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16901 (optalias): Likewise.
16903 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16905 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16906 * config/arm/arm-isa.h: Delete. Move definitions to ...
16907 * arm-cpus.in: ... here. Use new feature and fgroup values.
16908 * config/arm/arm.c (arm_option_override): Use lower case for feature
16910 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16911 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16912 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16914 (print_isa_bits_for): New function.
16915 (gen_isa): New function.
16916 (gen_comm_data): Use print_isa_bits_for.
16917 (define feature): New keyword.
16918 (define fgroup): New keyword.
16919 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16920 (arm-isa.h): Add rule to generate file.
16921 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16922 case for feature bit names.
16924 2017-09-06 Richard Biener <rguenther@suse.de>
16926 * tree-ssa-pre.c (NECESSARY): Remove.
16927 (create_expression_by_pieces): Do not touch pass-local flags.
16928 (insert_into_preds_of_block): Likewise.
16929 (do_pre_regular_insertion): Likewise.
16930 (eliminate_insert): Likewise.
16931 (eliminate_dom_walker::before_dom_children): Likewise.
16932 (fini_eliminate): Do not look at inserted_exprs.
16933 (mark_operand_necessary): Remove.
16934 (remove_dead_inserted_code): Replace with simple work-list
16935 algorithm based on inserted_exprs and SSA uses.
16936 (pass_pre::execute): Re-order fini_eliminate and
16937 remove_dead_inserted_code.
16939 2017-09-06 Olivier Hainque <hainque@adacore.com>
16941 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16942 for VxWorks 7. Adjust surrounding comments.
16944 2017-09-06 Richard Biener <rguenther@suse.de>
16946 * gimple-ssa-strength-reduction.c
16947 (find_candidates_dom_walker::before_dom_children): Also allow
16950 2017-09-06 Richard Biener <rguenther@suse.de>
16952 PR tree-optimization/82108
16953 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16954 for gap in the non-permutation SLP case.
16956 2017-09-06 Martin Jambor <mjambor@suse.cz>
16958 PR tree-optimization/82078
16959 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16960 add_access_to_work_queue...
16961 (build_accesses_from_assign): ...here.
16962 (propagate_all_subaccesses): Make sure racc is the group
16963 representative, if there is one.
16965 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16967 PR middle-end/82095
16968 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16971 2017-09-06 Richard Biener <rguenther@suse.de>
16973 * gimple-ssa-strength-reduction.c
16974 (find_candidates_dom_walker::before_doom_children): Use a
16975 type and not a mode check.
16977 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16980 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16982 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16984 2017-09-05 Jeff Law <law@redhat.com>
16986 PR tree-optimization/64910
16987 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16988 cases where we have 3 or more operands.
16990 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16992 PR middle-end/81768
16993 * omp-low.c (lower_omp_for): Recompute tree invariant if
16994 gimple_omp_for_initial/final is ADDR_EXPR.
16996 PR middle-end/81768
16997 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16998 into gimple val before gimplification fo the COND_EXPR.
17000 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
17002 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
17003 REGION_COPY argument.
17004 (thread_through_all_blocks): Remove unused argument to
17005 duplicate_thread_path.
17007 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17008 Alan Hayward <alan.hayward@arm.com>
17009 David Sherwood <david.sherwood@arm.com>
17011 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
17012 Take a scalar_mode rather than a machine_mode.
17013 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17014 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
17015 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17016 (aarch64_gen_adjusted_ldpstp): Likewise.
17017 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
17019 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17020 Alan Hayward <alan.hayward@arm.com>
17021 David Sherwood <david.sherwood@arm.com>
17023 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
17024 Take a scalar_int_mode instead of a machine_mode.
17025 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17026 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17027 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17028 (aarch64_simd_attr_length_rglist): Delete.
17029 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
17030 a scalar_int_mode instead of a machine_mode.
17031 (aarch64_add_offset): Likewise.
17032 (aarch64_internal_mov_immediate): Likewise
17033 (aarch64_add_constant_internal): Likewise.
17034 (aarch64_add_constant): Likewise.
17035 (aarch64_movw_imm): Likewise.
17036 (aarch64_rtx_arith_op_extract_p): Likewise.
17037 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17038 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17039 Remove assert that the mode isn't a vector.
17040 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17041 (aarch64_expand_mov_immediate): Update calls after above changes.
17042 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
17043 (aarch64_and_bitmask_imm): Check for scalar integer modes.
17044 (aarch64_move_imm): Likewise.
17045 (aarch64_can_const_movi_rtx_p): Likewise.
17046 (aarch64_strip_extend): Likewise.
17047 (aarch64_extr_rtx_p): Likewise.
17048 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
17049 a CONST_INT when the mode parameter is VOIDmode.
17050 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
17052 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17054 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
17055 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17056 (bitwise_type_for_mode): Update accordingly.
17058 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17060 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17061 * stor-layout.c (mode_for_size_tree): Likewise.
17062 (mode_for_array): Update accordingly.
17063 (layout_decl): Likewise.
17064 (compute_record_mode): Likewise. Only set the mode once.
17066 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17068 * target.def (get_mask_mode): Change return type to opt_mode.
17070 * doc/tm.texi: Regenerate.
17071 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17072 * targhooks.c (default_get_mask_mode): Likewise.
17073 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17074 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17075 targetm.get_mask_mode.
17076 * tree.c (build_truth_vector_type): Likewise.
17078 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17080 * machmode.h (mode_for_vector): Return an opt_mode.
17081 * stor-layout.c (mode_for_vector): Likewise.
17082 (mode_for_int_vector): Update accordingly.
17083 (layout_type): Likewise.
17084 * config/i386/i386.c (emit_memmov): Likewise.
17085 (ix86_expand_set_or_movmem): Likewise.
17086 (ix86_expand_vector_init): Likewise.
17087 (ix86_get_mask_mode): Likewise.
17088 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17090 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17091 * expmed.c (extract_bit_field_1): Likewise.
17092 * expr.c (expand_expr_real_2): Likewise.
17093 * optabs-query.c (can_vec_perm_p): Likewise.
17094 (can_vec_mask_load_store_p): Likewise.
17095 * optabs.c (expand_vec_perm): Likewise.
17096 * targhooks.c (default_get_mask_mode): Likewise.
17097 * tree-vect-stmts.c (vectorizable_store): Likewise.
17098 (vectorizable_load): Likewise.
17099 (get_vectype_for_scalar_type_and_size): Likewise.
17101 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17103 * machmode.h (mode_for_int_vector): New function.
17104 * stor-layout.c (mode_for_int_vector): Likewise.
17105 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17106 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17107 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17108 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17109 (s390_expand_vcond): Likewise.
17111 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17113 * machmode.h (opt_machine_mode): New type.
17114 (opt_mode<T>): Allow construction from anything that can be
17116 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17117 (mode_for_size): Return an opt_machine_mode.
17118 * stor-layout.c (mode_for_size): Likewise.
17119 (mode_for_size_tree): Update call accordingly.
17120 (bitwise_mode_for_mode): Likewise.
17121 (make_fract_type): Likewise.
17122 (make_accum_type): Likewise.
17123 * caller-save.c (replace_reg_with_saved_mem): Update call
17125 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17126 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17127 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17128 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17129 * expmed.c (extract_bit_field_1): Likewise.
17130 * reload.c (get_secondary_mem): Likewise.
17131 * varasm.c (assemble_integer): Likewise.
17132 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17135 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17137 * machmode.h (decimal_float_mode_for_size): New function.
17138 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17139 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17140 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17141 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17142 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17143 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17145 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17147 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17148 (get_builtin_sync_mode): Likewise.
17149 (expand_ifn_atomic_compare_exchange): Likewise.
17150 (expand_builtin_atomic_clear): Likewise.
17151 (expand_builtin_atomic_test_and_set): Likewise.
17152 (fold_builtin_atomic_always_lock_free): Likewise.
17153 * calls.c (compute_argument_addresses): Likewise.
17154 (emit_library_call_value_1): Likewise.
17155 (store_one_arg): Likewise.
17156 * combine.c (combine_instructions): Likewise.
17157 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17158 * config/arm/arm.c (arm_function_value): Likewise.
17159 (aapcs_allocate_return_reg): Likewise.
17160 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17161 * config/i386/i386.c (construct_container): Likewise.
17162 (ix86_gimplify_va_arg): Likewise.
17163 (ix86_expand_sse_cmp): Likewise.
17164 (emit_memmov): Likewise.
17165 (emit_memset): Likewise.
17166 (expand_small_movmem_or_setmem): Likewise.
17167 (ix86_expand_pextr): Likewise.
17168 (ix86_expand_pinsr): Likewise.
17169 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17170 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17172 * config/mips/mips.c (mips_function_value_1) Likewise.
17173 (mips_block_move_straight): Likewise.
17174 (mips_expand_ins_as_unaligned_store): Likewise.
17175 * config/powerpcspe/powerpcspe.c
17176 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17177 (rs6000_darwin64_record_arg_flush): Likewise.
17178 * config/rs6000/rs6000.c
17179 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17180 (rs6000_darwin64_record_arg_flush): Likewise.
17181 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17182 (sparc_function_value_1): Likewise.
17183 * config/spu/spu.c (adjust_operand): Likewise.
17184 (spu_emit_branch_or_set): Likewise.
17185 (arith_immediate_p): Likewise.
17186 * emit-rtl.c (gen_lowpart_common): Likewise.
17187 * expr.c (expand_expr_real_1): Likewise.
17188 * function.c (assign_parm_setup_block): Likewise.
17189 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17190 * reload1.c (alter_reg): Likewise.
17191 * stor-layout.c (mode_for_vector): Likewise.
17192 (layout_type): Likewise.
17194 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17196 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17197 (spu_convert_move): Likewise.
17198 * lower-subreg.c (resolve_simple_move): Likewise.
17200 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17203 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17204 define_insn to a define_expand.
17205 (altivec_vsum2sws_direct): New define_insn.
17206 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17208 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17210 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17213 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17218 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17219 realign stack if stack alignment needed is less than incoming
17222 2017-09-05 Marek Polacek <polacek@redhat.com>
17225 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17228 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17230 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17232 2017-09-05 Richard Biener <rguenther@suse.de>
17234 PR tree-optimization/82084
17235 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17237 2017-09-05 Richard Biener <rguenther@suse.de>
17239 PR tree-optimization/82102
17240 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17242 2017-09-05 Martin Liska <mliska@suse.cz>
17244 PR tree-optimization/82032
17245 * tree-cfg.c (generate_range_test): New function.
17246 * tree-cfg.h (generate_range_test): Declared here.
17247 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17248 (cleanup_control_expr_graph): Use it.
17249 * tree-switch-conversion.c (try_switch_expansion): Remove
17251 (emit_case_nodes): Use generate_range_test.
17253 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17256 * config/i386/i386.md (*<btsc><mode>_mask): Add
17257 TARGET_USE_BT to insn constraint.
17258 (*btr<mode>_mask): Ditto.
17260 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17262 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17263 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17265 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17268 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17269 TARGET_NEON and TARGET_IWMMXT.
17270 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17271 TARGET_NEON and TARGET_IWMMXT.
17272 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17274 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17276 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17277 (ix86_rewrite_tls_address): Ditto.
17278 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17279 (ix86_rewrite_tls_address_1): Ditto.
17280 (ix86_rewrite_tls_address): Ditto.
17281 * config/i386/predicates.md (tls_address_pattern): New predicate.
17282 * config/i386/i386.md (TLS address splitter): New splitter.
17284 2017-09-04 Richard Biener <rguenther@suse.de>
17286 PR tree-optimization/82084
17287 * fold-const.h (can_native_encode_string_p): Declare.
17288 * fold-const.c (can_native_encode_string_p): Factor out from ...
17289 (native_encode_string): ... here.
17290 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17291 vectorizing stores from constants we later cannot handle.
17293 2017-09-04 Marek Polacek <polacek@redhat.com>
17296 * doc/invoke.texi: Update -Wtautological-compare documentation.
17298 2017-09-04 Jeff Law <law@redhat.com>
17300 PR tree-optimization/64910
17301 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17302 swap the first and last operand if the last is a constant.
17304 2017-09-04 Marek Polacek <polacek@redhat.com>
17307 * convert.c (do_narrow): When sanitizing signed integer overflows,
17308 bail out for signed types.
17309 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17311 2017-09-04 Richard Biener <rguenther@suse.de>
17313 PR tree-optimization/82060
17314 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17315 Move devirtualization after stmt folding and before EH/AB/noreturn
17316 cleanup to get the stmt refs canonicalized. Use a bool instead
17317 of gimple_modified_p since that doesn't work for NOPs. Schedule
17318 NOPs generated by folding for removal.
17320 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17321 Alan Hayward <alan.hayward@arm.com>
17322 David Sherwood <david.sherwood@arm.com>
17324 * coretypes.h (pad_direction): New enum.
17325 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17326 (FUNCTION_ARG_PADDING): Likewise.
17327 * target.def (function_arg_padding): New hook.
17328 * targhooks.h (default_function_arg_padding): Declare.
17329 * targhooks.c (default_function_arg_padding): New function.
17330 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17331 (TARGET_FUNCTION_ARG_PADDING): ...this.
17332 * doc/tm.texi: Regenerate.
17333 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17334 instead of direction.
17335 (compute_argument_addresses): Likewise.
17336 (load_register_parameters): Likewise.
17337 (emit_library_call_value_1): Likewise.
17338 (store_one_arg): Use targetm.calls.function_arg_padding instead
17339 of FUNCTION_ARG_PADDING.
17340 (must_pass_in_stack_var_size_or_pad): Likewise.
17341 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17342 (emit_group_store): Likewise.
17343 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17344 instead of FUNCTION_ARG_PADDING.
17345 (emit_push_insn): Likewise, and propagate enum change throughout
17347 * function.h (direction): Delete.
17348 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17350 * function.c (assign_parm_find_stack_rtl): Likewise.
17351 (assign_parm_setup_block_p): Likewise.
17352 (assign_parm_setup_block): Likewise.
17353 (gimplify_parameters): Likewise.
17354 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17355 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17357 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17358 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17359 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17360 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17361 (aarch64_function_arg_padding): ...this new function.
17362 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17363 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17364 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17365 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17366 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17367 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17368 (arm_pad_arg_upward): Replace with...
17369 (arm_function_arg_padding): ...this new function.
17370 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17372 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17373 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17374 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17375 (ia64_hpux_function_arg_padding): Replace with...
17376 (ia64_function_arg_padding): ...this new function. Use pad_direction
17377 instead of direction. Check for TARGET_HPUX.
17378 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17379 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17380 (iq2000_function_arg_padding): New function.
17381 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17382 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17383 (mips_function_arg_padding): ...this new function.
17384 (mips_pad_reg_upward): Update accordingly.
17385 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17386 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17387 targetm.calls.function_arg_padding.
17388 (FUNCTION_ARG_PADDING): Delete.
17389 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17390 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17391 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17392 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17393 (nios2_block_reg_padding): Return pad_direction instead of direction.
17394 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17395 instead of direction.
17396 (nios2_function_arg_padding): Likewise. Make static.
17397 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17398 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17399 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17400 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17401 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17402 (pa_function_arg_padding): Make static. Return pad_direction instead
17404 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17405 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17406 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17407 instead of direction. Use targetm.calls.function_arg_padding.
17408 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17409 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17410 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17411 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17412 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17414 (function_arg_padding): Rename to...
17415 (rs6000_function_arg_padding): ...this. Make static. Return
17416 pad_direction instead of direction.
17417 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17418 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17419 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17420 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17421 instead of direction. Use targetm.calls.function_arg_padding.
17422 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17423 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17424 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17425 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17426 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17427 (function_arg_padding): Rename to...
17428 (rs6000_function_arg_padding): ...this. Make static. Return
17429 pad_direction instead of direction.
17430 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17431 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17432 * config/s390/s390.c (s390_function_arg_padding): New function.
17433 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17434 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17435 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17436 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17437 (function_arg_padding): Rename to...
17438 (sparc_function_arg_padding): ...this. Make static. Return
17439 pad_direction instead of direction.
17440 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17441 * config/spu/spu.c (spu_function_arg_padding): New function.
17442 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17443 * system.h (FUNCTION_ARG_PADDING): Poison.
17445 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17446 Alan Hayward <alan.hayward@arm.com>
17447 David Sherwood <david.sherwood@arm.com>
17449 * target.def (modes_tieable_p): New hook.
17450 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17451 (TARGET_MODES_TIEABLE_P): ...this.
17452 * doc/tm.texi.in: Regenerate.
17453 * hooks.h (hook_bool_mode_mode_true): Declare.
17454 * hooks.c (hook_bool_mode_mode_true): New function.
17455 * combine.c (subst): Use targetm.modes_tieable_p instead of
17457 * dse.c (find_shift_sequence): Likewise.
17458 * expmed.c (extract_low_bits): Likewise.
17459 * lower-subreg.c: Include target.h.
17460 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17462 * rtlanal.c (rtx_cost): Likewise.
17463 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17464 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17465 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17466 (TARGET_MODES_TIEABLE_P): Redefine.
17467 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17468 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17469 (TARGET_MODES_TIEABLE_P): Redefine.
17470 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17471 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17472 (arc_modes_tieable_p): New function.
17473 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17474 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17475 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17476 (arm_modes_tieable_p): Make static.
17477 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17478 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17479 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17480 (TARGET_MODES_TIEABLE_P): Redefine.
17481 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17482 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17483 (TARGET_MODES_TIEABLE_P): Redefine.
17484 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17485 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17486 (cr16_modes_tieable_p): New function.
17487 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17488 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17489 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17490 (TRULY_NOOP_TRUNCATION): Update comment.
17491 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17492 (TRULY_NOOP_TRUNCATION): Update comment.
17493 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17494 (frv_modes_tieable_p): New function.
17495 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17496 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17497 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17498 (TARGET_MODES_TIEABLE_P): Redefine.
17499 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17500 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17501 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17502 (TARGET_MODES_TIEABLE_P): Redefine.
17503 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17504 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17505 (ia64_modes_tieable_p): New function.
17506 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17507 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17508 (iq2000_modes_tieable_p): New function.
17509 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17510 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17511 (lm32_modes_tieable_p): New function.
17512 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17513 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17514 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17515 (TARGET_MODES_TIEABLE_P): Redefine.
17516 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17517 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17518 (m32r_modes_tieable_p): New function.
17519 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17520 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17521 (m68k_modes_tieable_p): New function.
17522 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17523 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17524 (mcore_modes_tieable_p): New function.
17525 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17526 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17528 (TARGET_MODES_TIEABLE_P): Redefine.
17529 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17530 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17531 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17532 (TARGET_MODES_TIEABLE_P): Redefine.
17533 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17534 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17535 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17536 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17537 (mn10300_modes_tieable_p): ...this and make static.
17538 (TARGET_MODES_TIEABLE_P): Redefine.
17539 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17540 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17541 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17542 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17543 (msp430_modes_tieable_p): Make static.
17544 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17545 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17546 (TARGET_MODES_TIEABLE_P): Redefine.
17547 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17548 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17549 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17550 (TARGET_MODES_TIEABLE_P): Redefine.
17551 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17552 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17553 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17554 (TARGET_MODES_TIEABLE_P): Redefine.
17555 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17556 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17557 (pdp11_modes_tieable_p): New function.
17558 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17559 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17560 (rs6000_modes_tieable_p): New function.
17561 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17562 * config/powerpcspe/powerpcspe.md: Update comment.
17563 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17564 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17565 (TARGET_MODES_TIEABLE_P): Redefine.
17566 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17567 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17568 (rl78_modes_tieable_p): New function.
17569 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17570 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17571 (rs6000_modes_tieable_p): New function.
17572 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17573 * config/rs6000/rs6000.md: Update comment.
17574 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17575 * config/rx/rx.c (rx_modes_tieable_p): New function.
17576 (TARGET_MODES_TIEABLE_P): Redefine.
17577 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17578 * config/s390/s390.c (s390_modes_tieable_p): New function.
17579 (TARGET_MODES_TIEABLE_P): Redefine.
17580 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17581 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17582 (sh_modes_tieable_p): New function.
17583 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17584 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17585 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17586 (sparc_modes_tieable_p): Make static.
17587 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17588 * config/spu/spu.c (spu_modes_tieable_p): New function.
17589 (TARGET_MODES_TIEABLE_P): Redefine.
17590 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17591 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17592 (TARGET_MODES_TIEABLE_P): Redefine.
17593 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17594 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17595 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17596 * config/v850/v850.c (v850_modes_tieable_p): New function.
17597 (TARGET_MODES_TIEABLE_P): Redefine.
17598 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17599 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17600 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17601 (visium_modes_tieable_p): New function.
17602 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17603 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17604 (xtensa_modes_tieable_p): New function.
17605 * system.h (MODES_TIEABLE_P): Poison.
17607 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17608 Alan Hayward <alan.hayward@arm.com>
17609 David Sherwood <david.sherwood@arm.com>
17611 * target.def (hard_regno_mode_ok): New hook.
17612 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17613 (TARGET_HARD_REGNO_MODE_OK): ...this.
17614 * doc/tm.texi.in: Regenerate.
17615 * hooks.h (hook_bool_uint_mode_true): Declare.
17616 * hooks.c (hook_bool_uint_mode_true): New function.
17617 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17618 HARD_REGNO_MODE_OK.
17619 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17620 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17621 instead of HARD_REGNO_MODE_OK.
17622 * caller-save.c: Include target.h.
17623 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17624 HARD_REGNO_MODE_OK.
17625 * combine.c (can_combine_p): Likewise.
17626 (combinable_i3pat): Likewise.
17627 (can_change_dest_mode): Likewise.
17628 * expr.c (init_expr_target): Likewise.
17629 (convert_move): Likewise.
17630 (convert_modes): Likewise.
17631 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17632 (setup_prohibited_mode_move_regs): Likewise.
17633 * ira.h (target_ira): Likewise.
17634 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17635 * lra-constraints.c (process_alt_operands): Likewise.
17636 (split_reg): Likewise.
17637 * recog.c (peep2_find_free_register): Likewise.
17638 * ree.c (combine_reaching_defs): Likewise.
17639 * regcprop.c (maybe_mode_change): Likewise.
17640 * reginfo.c (init_reg_sets_1): Likewise.
17641 (choose_hard_reg_mode): Likewise.
17642 (simplifiable_subregs): Likewise.
17643 * regrename.c (check_new_reg_p): Likewise.
17644 * reload.c (find_valid_class): Likewise.
17645 (find_valid_class_1): Likewise.
17646 (reload_inner_reg_of_subreg): Likewise.
17647 (push_reload): Likewise.
17648 (combine_reloads): Likewise.
17649 (find_dummy_reload): Likewise.
17650 (find_reloads): Likewise.
17651 * reload1.c (find_reg): Likewise.
17652 (set_reload_reg): Likewise.
17653 (allocate_reload_reg): Likewise.
17654 (choose_reload_regs): Likewise.
17655 (reload_adjust_reg_for_temp): Likewise.
17656 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17657 (simplify_subreg_regno): Likewise.
17658 * sel-sched.c (init_regs_for_mode): Likewise.
17659 * varasm.c (make_decl_rtl): Likewise.
17660 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17661 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17662 HARD_REGNO_MODE_OK.
17663 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17664 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17665 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17666 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17667 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17668 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17669 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17670 (arc_mode_class): Delete.
17671 (HARD_REGNO_MODE_OK): Delete.
17672 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17673 (arc_hard_regno_mode_ok): Rename old array to...
17674 (arc_hard_regno_mode_ok_modes): ...this.
17675 (arc_conditional_register_usage): Update accordingly.
17676 (arc_mode_class): Make static.
17677 (arc_hard_regno_mode_ok): New function.
17678 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17679 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17680 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17681 (arm_hard_regno_mode_ok): Make static.
17682 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17683 HARD_REGNO_MODE_OK.
17684 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17685 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17686 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17688 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17689 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17690 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17691 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17692 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17693 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17694 * config/bfin/predicates.md (valid_reg_operand): Use
17695 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17696 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17697 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17698 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17699 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17700 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17701 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17702 (cr16_hard_regno_mode_ok): Make static and return a bool.
17703 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17704 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17705 (cris_hard_regno_mode_ok): New function.
17706 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17707 (epiphany_mode_class): Delete.
17708 (HARD_REGNO_MODE_OK): Delete.
17709 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17710 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17711 (hard_regno_mode_ok): Rename to...
17712 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17713 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17714 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17715 HARD_REGNO_MODE_OK.
17716 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17717 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17718 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17719 (frv_hard_regno_mode_ok): Make static and return a bool.
17720 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17721 HARD_REGNO_MODE_OK.
17722 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17723 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17724 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17725 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17727 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17728 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17729 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17730 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17732 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17733 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17734 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17735 (ia64_hard_regno_mode_ok): New function.
17736 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17737 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17738 (iq2000_hard_regno_mode_ok): New function.
17739 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17740 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17741 (lm32_hard_regno_mode_ok): New function.
17742 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17743 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17744 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17745 instead of HARD_REGNO_MODE_OK.
17746 (m32c_hard_regno_ok): Rename to...
17747 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17748 (m32c_cannot_change_mode_class): Update accordingly.
17749 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17750 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17751 (m32r_mode_class): Delete.
17752 (HARD_REGNO_MODE_OK): Delete.
17753 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17754 (m32r_hard_regno_mode_ok): Rename to...
17755 (m32r_hard_regno_modes): ...this.
17756 (m32r_mode_class): Make static.
17757 (m32r_hard_regno_mode_ok): New function.
17758 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17759 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17760 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17761 (m68k_hard_regno_mode_ok): Make static.
17762 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17763 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17764 (mcore_hard_regno_mode_ok): New function.
17765 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17766 (HARD_REGNO_MODE_OK): Delete.
17767 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17769 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17770 (microblaze_hard_regno_mode_ok): New function.
17771 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17772 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17773 (mips_hard_regno_mode_ok): Delete.
17774 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17775 (mips_hard_regno_mode_ok_p): ...this and make static.
17776 (mips_hard_regno_mode_ok_p): Rename to...
17777 (mips_hard_regno_mode_ok_uncached): ...this.
17778 (mips_hard_regno_mode_ok): New function.
17779 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17780 of HARD_REGNO_MODE_OK.
17781 (mips_option_override): Update after above name changes.
17782 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17783 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17784 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17785 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17786 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17787 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17788 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17789 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17790 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17791 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17792 (msp430_hard_regno_mode_ok): Make static and return a bool.
17793 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17794 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17795 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17797 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17798 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17799 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17800 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17801 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17802 (PA_HARD_REGNO_MODE_OK): ...this
17803 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17804 (PA_HARD_REGNO_MODE_OK): ...this.
17805 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17806 (pa_hard_regno_mode_ok): New function.
17807 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17808 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17809 (pdp11_hard_regno_mode_ok): New function.
17810 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17811 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17813 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17815 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17816 (rs6000_hard_regno_mode_ok): Rename to...
17817 (rs6000_hard_regno_mode_ok_uncached): ...this.
17818 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17819 (rs6000_hard_regno_mode_ok): New function.
17820 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17821 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17822 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17823 (riscv_hard_regno_mode_ok): ...this and make static.
17824 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17825 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17826 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17827 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17828 (rl78_hard_regno_mode_ok): Make static and return bool.
17829 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17830 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17832 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17833 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17834 (rs6000_hard_regno_mode_ok): Rename to...
17835 (rs6000_hard_regno_mode_ok_uncached): ...this.
17836 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17837 (rs6000_hard_regno_mode_ok): New function.
17838 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17839 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17840 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17841 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17842 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17843 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17844 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17845 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17846 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17847 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17848 (sh_hard_regno_mode_ok): Make static.
17849 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17850 instead of HARD_REGNO_MODE_OK.
17851 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17852 (sparc_mode_class): Delete.
17853 (HARD_REGNO_MODE_OK): Delete.
17854 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17855 (hard_regno_mode_classes): Make static.
17856 (sparc_mode_class): Likewise.
17857 (sparc_hard_regno_mode_ok): New function.
17858 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17859 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17860 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17862 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17863 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17864 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17865 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17866 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17867 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17868 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17869 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17870 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17871 (visium_hard_regno_mode_ok): New function.
17872 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17873 instead of HARD_REGNO_MODE_OK.
17874 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17875 (HARD_REGNO_MODE_OK): Delete.
17876 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17877 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17878 (xtensa_option_override): Update accordingly.
17879 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17880 (xtensa_hard_regno_mode_ok): New function.
17881 * system.h (HARD_REGNO_MODE_OK): Poison.
17883 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17884 Alan Hayward <alan.hayward@arm.com>
17885 David Sherwood <david.sherwood@arm.com>
17887 * target.def (hard_regno_call_part_clobbered): New hook.
17888 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17889 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17890 * doc/tm.texi: Regenerate.
17891 * hooks.h (hook_bool_uint_mode_false): Declare.
17892 * hooks.c (hook_bool_uint_mode_false): New function.
17893 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17894 * cselib.c (cselib_process_insn): Use
17895 targetm.hard_regno_call_part_clobbered instead of
17896 HARD_REGNO_CALL_PART_CLOBBERED.
17897 * ira-conflicts.c (ira_build_conflicts): Likewise.
17898 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17899 * lra-constraints.c (need_for_call_save_p): Likewise.
17900 * lra-lives.c: Include target.h.
17901 (check_pseudos_live_through_calls): Use
17902 targetm.hard_regno_call_part_clobbered instead of
17903 HARD_REGNO_CALL_PART_CLOBBERED.
17904 * regcprop.c: Include target.h.
17905 (copyprop_hardreg_forward_1): Use
17906 targetm.hard_regno_call_part_clobbered instead of
17907 HARD_REGNO_CALL_PART_CLOBBERED.
17908 * reginfo.c (choose_hard_reg_mode): Likewise.
17909 * regrename.c (check_new_reg_p): Likewise.
17910 * reload.c (find_equiv_reg): Likewise.
17911 * reload1.c (emit_reload_insns): Likewise.
17912 * sched-deps.c (deps_analyze_insn): Likewise.
17913 * sel-sched.c (init_regs_for_mode): Likewise.
17914 (mark_unavailable_hard_regs): Likewise.
17915 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17916 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17917 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17919 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17920 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17921 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17923 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17925 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17926 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17927 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17929 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17930 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17931 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17933 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17934 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17936 * config/powerpcspe/powerpcspe.c
17937 (rs6000_hard_regno_call_part_clobbered): New function.
17938 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17939 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17940 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17942 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17943 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17944 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17946 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17947 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17948 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17950 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17951 Alan Hayward <alan.hayward@arm.com>
17952 David Sherwood <david.sherwood@arm.com>
17954 * rtl.h (subreg_memory_offset): Declare.
17955 * emit-rtl.c (subreg_memory_offset): New function.
17956 * expmed.c (store_bit_field_1): Use it.
17957 * expr.c (undefined_operand_subword_p): Likewise.
17958 * simplify-rtx.c (simplify_subreg): Likewise.
17960 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17962 PR rtl-optimization/57448
17965 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17966 using atomic_load pattern.
17967 (expand_atomic_store): Likewise.
17969 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17972 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17973 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17974 handling. Use replace_call_with_value with NULL instead of
17975 gsi_replace, unlink_stmt_vdef and release_defs.
17977 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17980 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17982 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17985 * rtl.h (emit_library_call_value_1): Declare.
17986 (emit_library_call): Replace declaration with a series of overloads.
17987 Remove the parameter count argument.
17988 (emit_library_call_value): Likewise.
17989 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17990 with an "rtx_mode_t *".
17991 (emit_library_call_value): Delete.
17992 (emit_library_call): Likewise.
17993 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17994 (asan_emit_allocas_unpoison): Likewise.
17995 * builtins.c (expand_builtin_powi): Likewise.
17996 (expand_asan_emit_allocas_unpoison): Likewise.
17997 * cfgexpand.c (expand_main_function): Likewise.
17998 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17999 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
18000 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
18001 * config/arm/arm.c (arm_trampoline_init): Likewise.
18002 (arm_call_tls_get_addr): Likewise.
18003 (arm_expand_divmod_libfunc): Likewise.
18004 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
18005 (smulsi3_highpart): Likewise.
18006 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
18007 (c6x_expand_compare): Likewise.
18008 (c6x_expand_movmem): Likewise.
18009 * config/frv/frv.c (frv_trampoline_init): Likewise.
18010 * config/i386/i386.c (ix86_trampoline_init): Likewise.
18011 (ix86_expand_divmod_libfunc): Likewise.
18012 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
18013 (ia64_expand_compare): Likewise.
18014 (ia64_profile_hook): Likewise.
18015 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
18016 (nonlocal_goto): Likewise.
18017 (restore_stack_nonlocal): Likewise.
18018 * config/m32r/m32r.c (block_move_call): Likewise.
18019 (m32r_trampoline_init): Likewise.
18020 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
18021 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
18022 (m68k_call_m68k_read_tp): Likewise.
18023 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
18024 (microblaze_expand_divide): Likewise.
18025 * config/mips/mips.h (mips_args): Likewise.
18026 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
18027 (MIPS_ICACHE_SYNC): Likewise.
18028 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
18029 (nios2_trampoline_init): Likewise.
18030 * config/pa/pa.c (hppa_tls_call): Likewise.
18031 (pa_trampoline_init): Likewise.
18032 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
18033 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
18034 (expand_strn_compare): Likewise.
18035 (rs6000_generate_compare): Likewise.
18036 (rs6000_expand_float128_convert): Likewise.
18037 (output_profile_hook): Likewise.
18038 (rs6000_trampoline_init): Likewise.
18039 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
18040 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
18041 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
18042 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
18043 (rs6000_generate_compare): Likewise.
18044 (rs6000_expand_float128_convert): Likewise.
18045 (output_profile_hook): Likewise.
18046 (rs6000_trampoline_init): Likewise.
18047 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
18048 * config/sh/sh.c (sh_trampoline_init): Likewise.
18049 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
18050 (sparc_emit_float_lib_cmp): Likewise.
18051 (sparc32_initialize_trampoline): Likewise.
18052 (sparc64_initialize_trampoline): Likewise.
18053 (sparc_profile_hook): Likewise.
18054 * config/spu/spu.c (ea_load_store): Likewise.
18055 * config/spu/spu.md (floatunssidf2): Likewise.
18056 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
18057 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
18058 * config/visium/visium.c (expand_block_move_4): Likewise.
18059 (expand_block_move_2): Likewise.
18060 (expand_block_move_1): Likewise.
18061 (expand_block_set_4): Likewise.
18062 (expand_block_set_2): Likewise.
18063 (expand_block_set_1): Likewise.
18064 (visium_trampoline_init): Likewise.
18065 (visium_profile_hook): Likewise.
18066 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18067 (xtensa_setup_frame_addresses): Likewise.
18068 (xtensa_trampoline_init): Likewise.
18069 * except.c (sjlj_emit_function_enter): Likewise.
18070 (sjlj_emit_function_exit): Likewise.
18071 * explow.c (allocate_dynamic_stack_space): Likewise.
18072 (probe_stack_range): Likewise.
18073 * expr.c (convert_mode_scalar): Likewise.
18074 * optabs.c (expand_binop): Likewise.
18075 (expand_twoval_binop_libfunc): Likewise.
18076 (expand_unop): Likewise.
18077 (prepare_cmp_insn): Likewise.
18078 (prepare_float_lib_cmp): Likewise.
18079 (expand_float): Likewise.
18080 (expand_fix): Likewise.
18081 (expand_fixed_convert): Likewise.
18082 (maybe_emit_sync_lock_test_and_set): Likewise.
18083 (expand_atomic_compare_and_swap): Likewise.
18084 (expand_mem_thread_fence): Likewise.
18085 (expand_atomic_fetch_op): Likewise.
18087 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18089 * doc/generic.texi (OpenACC): Adjust URL.
18090 * doc/invoke.texi (C Dialect Options): Ditto.
18092 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18094 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18095 predicate for operand 1. Add (m,<S>) constraint.
18096 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18097 Prevent memory operand 1 with register operand 2.
18099 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18101 PR rtl-optimization/82024
18102 * combine.c (try_combine): If the combination result is a PARALLEL,
18103 and we only need to retain the SET in there that would be placed
18104 at I2, check that we can place that at I3 instead, before doing so.
18106 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18109 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18111 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18112 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18115 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18116 Jeff Law <law@redhat.com>
18118 * varasm.c (bss_initializer_p): Do not put constants into .bss
18119 (categorize_decl_for_section): Handle bss_initializer_p returning
18120 false when DECL_INITIAL is NULL.
18122 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18125 * config/s390/s390.c (s390_can_inline_p): New function.
18127 2017-09-01 Jeff Law <law@redhat.com>
18129 PR tree-optimization/82052
18130 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18131 Always initialize the returned slot after a hash table miss
18132 when INSERT is true.
18134 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18136 * config/s390/s390.md (mem_signal_fence): Remove.
18137 * doc/md.texi (mem_signal_fence): Remove.
18138 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18140 * target-insns.def (mem_signal_fence): Remove.
18142 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18145 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18148 * asan.c (create_odr_indicator): Strip name encoding from assembler
18149 name before appending it after __odr_asan_.
18151 2017-09-01 Martin Liska <mliska@suse.cz>
18153 PR tree-optimization/82059
18154 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18155 frequency only when an edge is redirected.
18157 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18159 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18160 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18161 (arc_conditional_register_usage): Remove ARC600 lp_count
18163 (arc_file_start): Emit Tag_ARC_CPU_variation.
18164 (arc_can_use_doloop_p): New conditions to use ZOLs.
18165 (hwloop_fail): New function.
18166 (hwloop_optimize): Likewise.
18167 (hwloop_pattern_reg): Likewise.
18168 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18169 (arc_reorg_loops): New function, calls reorg_loops.
18170 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18171 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18173 (arc_loop_hazard): Remove function, functionality moved into
18175 (arc_hazard): Remove arc_loop_hazard call.
18176 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18177 into hwloop_optimize.
18178 (arc_label_align): Remove ZOL handling.
18179 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18180 * config/arc/arc.md (doloop_begin): Remove pattern.
18181 (doloop_begin_i): Likewise.
18182 (doloop_end_i): Likewise.
18183 (doloop_fallback): Likewise.
18184 (doloop_fallback_m): Likewise.
18185 (doloop_end): Reimplement expand.
18186 (arc_lp): New pattern for LP instruction.
18187 (loop_end): New pattern.
18188 (loop_fail): Likewise.
18189 (decrement_and_branch_until_zero): Likewise.
18190 * config/arc/arc.opt (mlpc-width): New option.
18191 * doc/invoke.texi (mlpc-width): Document option.
18193 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18195 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18196 (arc_ccfsm_advance): Fix checking for delay slots.
18197 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18199 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18201 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18203 (movhi_insn): Update store instruction constraint which are saving
18204 6-bit short immediates.
18205 (movsi_insn): Consider also short scaled load operations.
18206 (zero_extendhisi2_i): Use Usd constraint instead of T.
18207 (extendhisi2_i): Add q constraint.
18208 (arc_clzsi2): Add type and length attributes.
18209 (arc_ctzsi2): Likewise.
18210 * config/arc/constraints.md (Usc): Update constraint, the
18211 assembler can parse two relocations for a single instruction.
18213 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18215 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18216 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18218 2017-08-31 Olivier Hainque <hainque@adacore.com>
18220 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18221 match as powerpc-wrs-vxworks*.
18223 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18225 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18226 register constraint for by-element operand.
18227 (aarch64_mls_elt_merge<mode>): Likewise.
18229 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18231 * config/arc/arc.c (arc_can_follow_jump): Check for short
18234 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18236 * config.gcc: Use g.opt for arc.
18237 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18238 functionality moved to ...
18239 (legitimate_scaled_address_p): New function, ...here.
18240 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18241 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18242 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18244 (arc_override_options): Handle G option.
18245 (arc_output_pic_addr_const): Correct function definition.
18246 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18247 (arc_decl_anon_ns_mem_p): Delete.
18248 (arc_in_small_data_p): Overhaul this function to take into
18249 consideration the value given via G option.
18250 (arc_rewrite_small_data_1): Renamed and corrected old
18251 arc_rewrite_small_data function.
18252 (arc_rewrite_small_data): New function.
18253 (small_data_pattern): Don't use pic_offset_table_rtx.
18254 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18255 * config/arc/simdext.md (movmisalignv2hi): Use
18256 prepare_move_operands function.
18258 (movmisalign*): Likewise.
18259 * doc/invoke.texi (ARC options): Document -G option.
18261 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18263 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18265 * config/arc/arc.c (arc_print_operand): Output scalled address for
18266 sdata whenever is possible.
18267 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18268 load/stores are available.
18269 (compact_sda_memory_operand): Check for the alignment required by
18270 code density instructions.
18271 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18273 * config/arc/constraints.md (Usd): Update constraint.
18274 (Us0): New constraint.
18275 (Usc): Update constraint.
18277 2017-08-31 Richard Biener <rguenther@suse.de>
18279 PR middle-end/82054
18280 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18281 function only once.
18283 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18285 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18286 Resize type_signature.
18288 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18289 Alan Hayward <alan.hayward@arm.com>
18290 David Sherwood <david.sherwood@arm.com>
18292 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18293 subregs whose inner modes can be stored in GPRs.
18294 (aarch64_classify_index): Likewise.
18296 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18297 Alan Hayward <alan.hayward@arm.com>
18298 David Sherwood <david.sherwood@arm.com>
18300 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18301 (V_INT_EQUIV): ...this.
18302 (v_cmp_result): Rename to...
18303 (v_int_equiv): ...this.
18304 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18305 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18306 (copysign<mode>3): Likewise.
18307 (aarch64_simd_bsl<mode>_internal): Likewise.
18308 (aarch64_simd_bsl<mode>): Likewise.
18309 (vec_cmp<mode><mode>): Likewise.
18310 (vcond<mode><mode>): Likewise.
18311 (vcond<v_cmp_mixed><mode>): Likewise.
18312 (vcondu<mode><v_cmp_mixed>): Likewise.
18313 (aarch64_cm<optab><mode>): Likewise.
18314 (aarch64_cmtst<mode>): Likewise.
18315 (aarch64_fac<optab><mode>): Likewise.
18316 (vec_perm_const<mode>): Likewise.
18317 (vcond_mask_<mode><v_cmp_result>): Rename to...
18318 (vcond_mask_<mode><v_int_equiv>): ...this.
18319 (vec_cmp<mode><v_cmp_result>): Rename to...
18320 (vec_cmp<mode><v_int_equiv>): ...this.
18322 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18323 Alan Hayward <alan.hayward@arm.com>
18324 David Sherwood <david.sherwood@arm.com>
18326 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18328 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18329 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18330 (UNSPEC_LD4_DREG): New unspecs.
18331 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18332 (aarch64_ld2<mode>_dreg_be): Replace with...
18333 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18335 (aarch64_ld3<mode>_dreg_le)
18336 (aarch64_ld3<mode>_dreg_be): Replace with...
18337 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18339 (aarch64_ld4<mode>_dreg_le)
18340 (aarch64_ld4<mode>_dreg_be): Replace with...
18341 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18344 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18346 PR tree-optimization/81987
18347 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18348 insert an initializer in a location not dominated by the stride
18351 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18353 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18354 on the entire header of the finally block in the fallthru case.
18356 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18358 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18360 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18362 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18363 rs6000_emit_move_from_cr and call renamed function.
18364 (rs6000_emit_prologue): Call renamed functions.
18365 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18366 movesi_from_cr, remove volatile CRs.
18368 2017-08-30 Jon Beniston <jon@beniston.com>
18369 Richard Biener <rguenther@suse.de>
18371 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18372 of VECTOR_MODE_P check.
18373 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18374 element vector types.
18376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18378 * df.h (df_read_modify_subreg_p): Remove in favor of...
18379 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18380 const_rtx instead of an rtx.
18381 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18382 * df-problems.c (df_word_lr_mark_ref): Likewise.
18383 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18384 (mark_pseudo_reg_dead): Likewise.
18385 (mark_ref_dead): Likewise.
18386 * reginfo.c (init_subregs_of_mode): Likewise.
18387 * sched-deps.c (sched_analyze_1): Likewise.
18388 * df-scan.c (df_def_record_1): Likewise.
18389 (df_uses_record): Likewise.
18390 (df_read_modify_subreg_p): Remove in favor of...
18391 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18392 const_rtx instead of an rtx.
18394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18395 Alan Hayward <alan.hayward@arm.com>
18396 David Sherwood <david.sherwood@arm.com>
18398 * rtl.h (partial_subreg_p): New function.
18399 * caller-save.c (save_call_clobbered_regs): Use it.
18400 * calls.c (expand_call): Likewise.
18401 * combine.c (combinable_i3pat): Likewise.
18402 (simplify_set): Likewise.
18403 (make_extraction): Likewise.
18404 (make_compound_operation_int): Likewise.
18405 (gen_lowpart_or_truncate): Likewise.
18406 (force_to_mode): Likewise.
18407 (make_field_assignment): Likewise.
18408 (reg_truncated_to_mode): Likewise.
18409 (record_truncated_value): Likewise.
18410 (move_deaths): Likewise.
18411 * cse.c (record_jump_cond): Likewise.
18412 (cse_insn): Likewise.
18413 * cselib.c (cselib_lookup_1): Likewise.
18414 * expmed.c (extract_bit_field_using_extv): Likewise.
18415 * function.c (assign_parm_setup_reg): Likewise.
18416 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18417 * ira-build.c (create_insn_allocnos): Likewise.
18418 * lra-coalesce.c (merge_pseudos): Likewise.
18419 * lra-constraints.c (match_reload): Likewise.
18420 (simplify_operand_subreg): Likewise.
18421 (curr_insn_transform): Likewise.
18422 * lra-lives.c (process_bb_lives): Likewise.
18423 * lra.c (new_insn_reg): Likewise.
18424 (lra_substitute_pseudo): Likewise.
18425 * regcprop.c (mode_change_ok): Likewise.
18426 (maybe_mode_change): Likewise.
18427 (copyprop_hardreg_forward_1): Likewise.
18428 * reload.c (push_reload): Likewise.
18429 (find_reloads): Likewise.
18430 (find_reloads_subreg_address): Likewise.
18431 * reload1.c (alter_reg): Likewise.
18432 (eliminate_regs_1): Likewise.
18433 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18435 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18437 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18438 back to if statements, including unpack.
18440 2017-08-30 Martin Liska <mliska@suse.cz>
18442 PR inline-asm/82001
18443 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18445 (func_checker::compare_asm_inputs_outputs): ... this function.
18446 (func_checker::compare_gimple_asm): Use the function to compare
18447 also ASM constrains.
18448 * ipa-icf-gimple.h: Rename the function.
18450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18451 Alan Hayward <alan.hayward@arm.com>
18452 David Sherwood <david.sherwood@arm.com>
18454 * coretypes.h (complex_mode): New type.
18455 * gdbhooks.py (build_pretty_printer): Handle it.
18456 * machmode.h (complex_mode): New class.
18457 (complex_mode::includes_p): New function.
18458 (is_complex_int_mode): Likewise.
18459 (is_complex_float_mode): Likewise.
18460 * genmodes.c (get_mode_class): Handle complex mode classes.
18461 * function.c (expand_function_end): Use is_complex_int_mode.
18463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18464 Alan Hayward <alan.hayward@arm.com>
18465 David Sherwood <david.sherwood@arm.com>
18467 * coretypes.h (scalar_mode_pod): New typedef.
18468 * gdbhooks.py (build_pretty_printer): Handle it.
18469 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18470 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18471 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18472 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18473 as_a <scalar_mode>.
18475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18476 Alan Hayward <alan.hayward@arm.com>
18477 David Sherwood <david.sherwood@arm.com>
18479 * machmode.h (mode_for_vector): Take a scalar_mode instead
18481 * stor-layout.c (mode_for_vector): Likewise.
18482 * explow.c (promote_mode): Use as_a <scalar_mode>.
18483 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18486 Alan Hayward <alan.hayward@arm.com>
18487 David Sherwood <david.sherwood@arm.com>
18489 * target.def (preferred_simd_mode): Take a scalar_mode
18490 instead of a machine_mode.
18491 * targhooks.h (default_preferred_simd_mode): Likewise.
18492 * targhooks.c (default_preferred_simd_mode): Likewise.
18493 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18494 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18495 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18496 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18497 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18498 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18499 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18500 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18502 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18503 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18504 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18505 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18506 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18507 * doc/tm.texi: Regenerate.
18508 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18512 Alan Hayward <alan.hayward@arm.com>
18513 David Sherwood <david.sherwood@arm.com>
18515 * target.def (scalar_mode_supported_p): Take a scalar_mode
18516 instead of a machine_mode.
18517 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18518 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18519 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18520 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18521 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18522 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18523 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18524 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18525 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18526 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18527 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18528 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18529 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18530 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18532 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18533 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18534 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18535 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18536 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18538 * doc/tm.texi: Regenerate.
18540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18541 Alan Hayward <alan.hayward@arm.com>
18542 David Sherwood <david.sherwood@arm.com>
18544 * coretypes.h (opt_scalar_mode): New typedef.
18545 * gdbhooks.py (build_pretty_printers): Handle it.
18546 * machmode.h (mode_iterator::get_2xwider): Add overload for
18548 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18550 * expr.c (convert_mode_scalar): Likewise.
18551 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18552 * optabs.c (expand_float): Likewise.
18553 (expand_fix): Likewise.
18554 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18557 Alan Hayward <alan.hayward@arm.com>
18558 David Sherwood <david.sherwood@arm.com>
18560 * optabs.c (expand_float): Explicitly check for scalars before
18561 using a branching expansion.
18562 (expand_fix): Likewise.
18564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18565 Alan Hayward <alan.hayward@arm.com>
18566 David Sherwood <david.sherwood@arm.com>
18568 * expr.c (convert_mode): Split scalar handling out into...
18569 (convert_mode_scalar): ...this new function. Treat the modes
18572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18573 Alan Hayward <alan.hayward@arm.com>
18574 David Sherwood <david.sherwood@arm.com>
18576 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18578 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18581 Alan Hayward <alan.hayward@arm.com>
18582 David Sherwood <david.sherwood@arm.com>
18584 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18585 rather than a machine_mode.
18586 (fixed_from_string): Likewise.
18587 (fixed_convert): Likewise.
18588 (fixed_convert_from_int): Likewise.
18589 (fixed_convert_from_real): Likewise.
18590 (real_convert_from_fixed): Likewise.
18591 * fixed-value.c (fixed_from_double_int): Likewise.
18592 (fixed_from_string): Likewise.
18593 (fixed_convert): Likewise.
18594 (fixed_convert_from_int): Likewise.
18595 (fixed_convert_from_real): Likewise.
18596 (real_convert_from_fixed): Likewise.
18597 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18600 Alan Hayward <alan.hayward@arm.com>
18601 David Sherwood <david.sherwood@arm.com>
18603 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18604 of separate mode class checks. Do not allow vector modes here.
18605 (immed_wide_int_const): Use as_a <scalar_mode>.
18606 * explow.c (trunc_int_for_mode): Likewise.
18607 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18608 (wi::shwi): Likewise.
18609 (wi::min_value): Likewise.
18610 (wi::max_value): Likewise.
18611 * dwarf2out.c (loc_descriptor): Likewise.
18612 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18613 for CONST_WIDE_INT.
18615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18616 Alan Hayward <alan.hayward@arm.com>
18617 David Sherwood <david.sherwood@arm.com>
18619 * tree.h (SCALAR_TYPE_MODE): New macro.
18620 * expr.c (expand_expr_addr_expr_1): Use it.
18621 (expand_expr_real_2): Likewise.
18622 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18623 (fold_convert_const_fixed_from_int): Likewise.
18624 (fold_convert_const_fixed_from_real): Likewise.
18625 (native_encode_fixed): Likewise
18626 (native_encode_complex): Likewise
18627 (native_encode_vector): Likewise.
18628 (native_interpret_fixed): Likewise.
18629 (native_interpret_real): Likewise.
18630 (native_interpret_complex): Likewise.
18631 (native_interpret_vector): Likewise.
18632 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18633 (simd_clone_adjust_argument_types): Likewise.
18634 (simd_clone_init_simd_arrays): Likewise.
18635 (simd_clone_adjust): Likewise.
18636 * stor-layout.c (layout_type): Likewise.
18637 * tree.c (build_minus_one_cst): Likewise.
18638 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18639 * tree-inline.c (estimate_move_cost): Likewise.
18640 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18641 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18642 (vectorizable_reduction): Likewise.
18643 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18644 (vect_recog_mixed_size_cond_pattern): Likewise.
18645 (check_bool_pattern): Likewise.
18646 (adjust_bool_pattern): Likewise.
18647 (search_type_for_mask_1): Likewise.
18648 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18649 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18650 (vectorizable_load): Likewise.
18651 (vectorizable_store): Likewise.
18652 * ubsan.c (ubsan_encode_value): Likewise.
18653 * varasm.c (output_constant): Likewise.
18655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18656 Alan Hayward <alan.hayward@arm.com>
18657 David Sherwood <david.sherwood@arm.com>
18659 * coretypes.h (scalar_mode): New class.
18660 * machmode.h (scalar_mode): Likewise.
18661 (scalar_mode::includes_p): New function.
18662 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18663 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18664 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18665 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18666 * expmed.c (store_bit_field_1): Likewise.
18667 (extract_bit_field_1): Likewise.
18668 * expr.c (write_complex_part): Likewise.
18669 (read_complex_part): Likewise.
18670 (emit_move_complex_push): Likewise.
18671 (expand_expr_real_2): Likewise.
18672 * function.c (assign_parm_setup_reg): Likewise.
18673 (assign_parms_unsplit_complex): Likewise.
18674 * optabs.c (expand_binop): Likewise.
18675 * rtlanal.c (subreg_get_info): Likewise.
18676 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18677 * varasm.c (output_constant_pool_2): Likewise.
18679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18680 Alan Hayward <alan.hayward@arm.com>
18681 David Sherwood <david.sherwood@arm.com>
18683 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18685 (expmed_mult_highpart_optab): Likewise.
18686 (expmed_mult_highpart): Likewise.
18688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18689 Alan Hayward <alan.hayward@arm.com>
18690 David Sherwood <david.sherwood@arm.com>
18692 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18693 instead of a machine_mode.
18694 (builtin_memset_read_str): Likewise.
18695 * builtins.c (c_readstr): Likewise.
18696 (builtin_memcpy_read_str): Likewise.
18697 (builtin_strncpy_read_str): Likewise.
18698 (builtin_memset_read_str): Likewise.
18699 (builtin_memset_gen_str): Likewise.
18700 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18701 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18702 instead of a machine_mode.
18703 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18705 (make_extraction): Likewise.
18706 (try_widen_shift_mode): Take and return scalar_int_modes instead
18708 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18709 a scalar_int_mode instead of a machine_mode.
18710 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18711 (avr_addr_space_pointer_mode): Likewise.
18712 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18713 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18714 (msp430_unwind_word_mode): Likewise.
18715 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18716 (spu_addr_space_pointer_mode): Likewise.
18717 (spu_addr_space_address_mode): Likewise.
18718 (spu_libgcc_cmp_return_mode): Likewise.
18719 (spu_libgcc_shift_count_mode): Likewise.
18720 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18721 (rl78_addr_space_pointer_mode): Likewise.
18722 (fl78_unwind_word_mode): Likewise.
18723 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18725 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18726 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18727 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18728 (mips_valid_pointer_mode): Likewise.
18729 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18730 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18731 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18733 (ft32_addr_space_address_mode): Likewise.
18734 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18735 scalar_int_mode instead of a machine_mode.
18736 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18738 (m32c_addr_space_address_mode): Likewise.
18739 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18740 (rs6000_eh_return_filter_mode): Likewise.
18741 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18742 (rs6000_eh_return_filter_mode): Likewise.
18743 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18744 (s390_libgcc_shift_count_mode): Likewise.
18745 (s390_unwind_word_mode): Likewise.
18746 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18748 * target.def (mode_rep_extended): Likewise.
18749 (valid_pointer_mode): Likewise.
18750 (addr_space.valid_pointer_mode): Likewise.
18751 (eh_return_filter_mode): Return a scalar_int_mode rather than
18753 (libgcc_cmp_return_mode): Likewise.
18754 (libgcc_shift_count_mode): Likewise.
18755 (unwind_word_mode): Likewise.
18756 (addr_space.pointer_mode): Likewise.
18757 (addr_space.address_mode): Likewise.
18758 * doc/tm.texi: Regenerate.
18759 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18761 (do_jump): Use scalar_int_mode for local variables.
18762 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18763 rather than a machine_mode.
18764 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18765 (scompare_loc_descriptor_wide): Likewise.
18766 (scompare_loc_descriptor_narrow): Likewise.
18767 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18769 * except.c (sjlj_emit_dispatch_table): Likewise.
18770 (expand_builtin_eh_copy_values): Likewise.
18771 * explow.c (convert_memory_address_addr_space_1): Likewise.
18772 Take a scalar_int_mode rather than a machine_mode.
18773 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18774 than a machine_mode.
18775 (memory_address_addr_space): Use scalar_int_mode for local variables.
18776 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18777 rather than a machine_mode.
18778 * expmed.c (mask_rtx): Likewise.
18779 (init_expmed_one_conv): Likewise.
18780 (expand_mult_highpart_adjust): Likewise.
18781 (extract_high_half): Likewise.
18782 (expmed_mult_highpart_optab): Likewise.
18783 (expmed_mult_highpart): Likewise.
18784 (expand_smod_pow2): Likewise.
18785 (expand_sdiv_pow2): Likewise.
18786 (emit_store_flag_int): Likewise.
18787 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18789 (extract_low_bits): Likewise.
18790 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18792 * expr.c (pieces_addr::adjust): Likewise.
18793 (can_store_by_pieces): Likewise.
18794 (store_by_pieces): Likewise.
18795 (clear_by_pieces_1): Likewise.
18796 (expand_expr_addr_expr_1): Likewise.
18797 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18798 (expand_expr_real_1): Likewise.
18799 (try_casesi): Likewise.
18800 * final.c (shorten_branches): Likewise.
18801 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18802 type of "mode" to machine_mode.
18803 * internal-fn.c (expand_arith_overflow_result_store): Take a
18804 scalar_int_mode rather than a machine_mode.
18805 (expand_mul_overflow): Use scalar_int_mode for local variables.
18806 * loop-doloop.c (doloop_modify): Likewise.
18807 (doloop_optimize): Likewise.
18808 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18809 than a machine_mode.
18810 (expand_doubleword_shift_condmove): Likewise.
18811 (expand_doubleword_shift): Likewise.
18812 (expand_doubleword_clz): Likewise.
18813 (expand_doubleword_popcount): Likewise.
18814 (expand_doubleword_parity): Likewise.
18815 (expand_absneg_bit): Use scalar_int_mode for local variables.
18816 (prepare_float_lib_cmp): Likewise.
18817 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18818 rather than a machine_mode.
18819 (convert_memory_address_addr_space): Likewise.
18820 (get_mode_bounds): Likewise.
18821 (get_address_mode): Return a scalar_int_mode rather than a
18823 * rtlanal.c (get_address_mode): Likewise.
18824 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18825 than a machine_mode.
18826 * targhooks.c (default_mode_rep_extended): Likewise.
18827 (default_valid_pointer_mode): Likewise.
18828 (default_addr_space_valid_pointer_mode): Likewise.
18829 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18830 than a machine_mode.
18831 (default_libgcc_cmp_return_mode): Likewise.
18832 (default_libgcc_shift_count_mode): Likewise.
18833 (default_unwind_word_mode): Likewise.
18834 (default_addr_space_pointer_mode): Likewise.
18835 (default_addr_space_address_mode): Likewise.
18836 * targhooks.h (default_eh_return_filter_mode): Likewise.
18837 (default_libgcc_cmp_return_mode): Likewise.
18838 (default_libgcc_shift_count_mode): Likewise.
18839 (default_unwind_word_mode): Likewise.
18840 (default_addr_space_pointer_mode): Likewise.
18841 (default_addr_space_address_mode): Likewise.
18842 (default_mode_rep_extended): Take a scalar_int_mode rather than
18844 (default_valid_pointer_mode): Likewise.
18845 (default_addr_space_valid_pointer_mode): Likewise.
18846 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18848 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18849 rather than a machine_mode.
18850 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18851 for local variables.
18852 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18853 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18854 than a machine_mode.
18856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18857 Alan Hayward <alan.hayward@arm.com>
18858 David Sherwood <david.sherwood@arm.com>
18860 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18861 the mode argument to scalar_int_mode.
18862 (do_jump_by_parts_zero_rtx): Likewise.
18863 (do_jump_by_parts_equality_rtx): Likewise.
18864 (do_jump_by_parts_greater): Take a mode argument.
18865 (do_jump_by_parts_equality): Likewise.
18866 (do_jump_1): Update calls accordingly.
18868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18869 Alan Hayward <alan.hayward@arm.com>
18870 David Sherwood <david.sherwood@arm.com>
18872 * is-a.h (safe_dyn_cast): New function.
18873 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18874 (jump_table_for_label): Likewise.
18875 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18876 instead of an rtx_insn *.
18877 (shorten_branches): Use dyn_cast instead of LABEL_P and
18878 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18879 rtx_jump_table_data::get_data_mode.
18880 (final_scan_insn): Likewise.
18882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18883 Alan Hayward <alan.hayward@arm.com>
18884 David Sherwood <david.sherwood@arm.com>
18886 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18887 trying to combine a full-register integer set with a subreg
18890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18891 Alan Hayward <alan.hayward@arm.com>
18892 David Sherwood <david.sherwood@arm.com>
18894 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18895 that is always either address_mode or pointer_mode.
18897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18898 Alan Hayward <alan.hayward@arm.com>
18899 David Sherwood <david.sherwood@arm.com>
18901 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18902 when the two are known to be equal.
18904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18905 Alan Hayward <alan.hayward@arm.com>
18906 David Sherwood <david.sherwood@arm.com>
18908 * simplify-rtx.c (simplify_const_unary_operation): Use
18909 is_a <scalar_int_mode> instead of checking for a nonzero
18910 precision. Forcibly convert op_mode to a scalar_int_mode
18911 in that case. More clearly differentiate the operand and
18912 result modes and use the former when deciding what the value
18913 of a count-bits operation should be. Use is_int_mode instead
18914 of checking for a MODE_INT. Remove redundant check for whether
18915 this mode has a zero precision.
18917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18918 Alan Hayward <alan.hayward@arm.com>
18919 David Sherwood <david.sherwood@arm.com>
18921 * optabs.c (widen_leading): Change the type of the mode argument
18922 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18923 (widen_bswap): Likewise.
18924 (expand_parity): Likewise.
18925 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18926 (expand_ffs): Likewise.
18927 (epand_unop): Check for scalar integer modes before calling the
18930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18931 Alan Hayward <alan.hayward@arm.com>
18932 David Sherwood <david.sherwood@arm.com>
18934 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18936 (expand_expr_real_1): Update call accordingly.
18938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18939 Alan Hayward <alan.hayward@arm.com>
18940 David Sherwood <david.sherwood@arm.com>
18942 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18943 value_mode arguments. Use scalar_int_mode internally.
18944 (store_bit_field_1): Rename the new integer mode from imode
18945 to op0_mode and use it instead of GET_MODE (op0). Update calls
18946 to store_split_bit_field, store_bit_field_using_insv and
18947 store_fixed_bit_field.
18948 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18949 Use scalar_int_mode internally. Use a bit count rather than a mode
18950 when calculating the largest bit size for get_best_mode.
18951 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18952 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18953 Remove assertion that OP0 has a scalar integer mode.
18954 (store_split_bit_field): Add op0_mode and value_mode arguments.
18955 Update calls to extract_fixed_bit_field.
18956 (extract_bit_field_using_extv): Add an op0_mode argument.
18957 Use scalar_int_mode internally.
18958 (extract_bit_field_1): Rename the new integer mode from imode to
18959 op0_mode and use it instead of GET_MODE (op0). Update calls to
18960 extract_split_bit_field, extract_bit_field_using_extv and
18961 extract_fixed_bit_field.
18962 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18963 to extract_split_bit_field and extract_fixed_bit_field_1.
18964 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18965 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18966 on the target mode.
18967 (extract_split_bit_field): Add an op0_mode argument. Update call
18968 to extract_fixed_bit_field.
18970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18971 Alan Hayward <alan.hayward@arm.com>
18972 David Sherwood <david.sherwood@arm.com>
18974 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18975 * explow.c (hard_function_value): Likewise.
18976 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18977 convert_to_mode call outside the loop.
18978 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18979 for the mode iterator. Require the mode specified by max_pieces
18981 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18983 (copy_blkmode_to_reg): Likewise.
18984 (set_storage_via_setmem): Likewise.
18985 * optabs.c (prepare_cmp_insn): Likewise.
18986 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18987 * stor-layout.c (finish_bitfield_representative): Likewise.
18989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18990 Alan Hayward <alan.hayward@arm.com>
18991 David Sherwood <david.sherwood@arm.com>
18993 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18994 * expr.c (convert_move): Use them.
18995 (convert_modes): Likewise.
18996 (store_expr_with_bounds): Likewise.
18998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18999 Alan Hayward <alan.hayward@arm.com>
19000 David Sherwood <david.sherwood@arm.com>
19002 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
19003 parameter for the mode of "x". Remove the "known_x", "known_mode"
19004 and "known_ret" arguments. Change the type of the mode argument
19005 to scalar_int_mode.
19006 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
19007 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
19008 (reg_num_sign_bit_copies_for_combine): Likewise.
19009 * rtlanal.c (nonzero_bits1): Likewise.
19010 (num_sign_bit_copies1): Likewise.
19011 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
19012 (reg_num_sign_bit_copies_general): Likewise.
19013 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
19014 (reg_nonzero_bits_general): Likewise.
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 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
19021 than in subroutines. Return 1 for non-integer modes.
19022 (cached_num_sign_bit_copies): Change the type of the mode parameter
19023 to scalar_int_mode.
19024 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
19025 classes. Handle CONST_INT_P first and then check whether X also
19026 has a scalar integer mode. Check the same thing for inner registers
19027 of a SUBREG and for values that are being extended or truncated.
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 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
19034 in subroutines. Return the mode mask for non-integer modes.
19035 (cached_nonzero_bits): Change the type of the mode parameter
19036 to scalar_int_mode.
19037 (nonzero_bits1): Likewise. Remove early exit for other mode
19038 classes. Handle CONST_INT_P first and then check whether X
19039 also has a scalar integer mode.
19041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19042 Alan Hayward <alan.hayward@arm.com>
19043 David Sherwood <david.sherwood@arm.com>
19045 * expr.c (widest_int_mode_for_size): Make the comment match the code.
19046 Return a scalar_int_mode and assert that the size is greater than
19048 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
19049 (op_by_pieces_d::op_by_pieces_d): Likewise.
19050 (op_by_pieces_d::run): Likewise.
19051 (can_store_by_pieces): Likewise.
19053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19054 Alan Hayward <alan.hayward@arm.com>
19055 David Sherwood <david.sherwood@arm.com>
19057 * combine.c (extract_left_shift): Add a mode argument and update
19059 (make_compound_operation_int): Change the type of the mode parameter
19060 to scalar_int_mode and update the call to extract_left_shift.
19062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19063 Alan Hayward <alan.hayward@arm.com>
19064 David Sherwood <david.sherwood@arm.com>
19066 * combine.c (simplify_and_const_int): Change the type of the mode
19067 parameter to scalar_int_mode.
19068 (simplify_and_const_int_1): Likewise. Update recursive call.
19070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19071 Alan Hayward <alan.hayward@arm.com>
19072 David Sherwood <david.sherwood@arm.com>
19074 * combine.c (simplify_compare_const): Check that the mode is a
19075 scalar_int_mode (rather than VOIDmode) before testing its
19077 (simplify_comparison): Move COMPARISON_P handling out of the
19078 loop and restrict the latter part of the loop to scalar_int_modes.
19079 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19080 and when considering SUBREG_REGs. Use is_int_mode instead of
19081 checking GET_MODE_CLASS against MODE_INT.
19083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19084 Alan Hayward <alan.hayward@arm.com>
19085 David Sherwood <david.sherwood@arm.com>
19087 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19088 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19089 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
19096 (force_to_mode): ...here. Keep xmode up-to-date and use it
19097 instead of GET_MODE (x).
19099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19100 Alan Hayward <alan.hayward@arm.com>
19101 David Sherwood <david.sherwood@arm.com>
19103 * optabs-query.h (extraction_insn::struct_mode): Change type to
19104 opt_scalar_int_mode and update comment.
19105 (extraction_insn::field_mode): Change type to scalar_int_mode.
19106 (extraction_insn::pos_mode): Likewise.
19107 * combine.c (make_extraction): Update accordingly.
19108 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19109 (get_optab_extraction_insn): Likewise.
19110 * recog.c (simplify_while_replacing): Likewise.
19111 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19112 parameter to opt_scalar_int_mode.
19114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19115 Alan Hayward <alan.hayward@arm.com>
19116 David Sherwood <david.sherwood@arm.com>
19118 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19119 to a scalar_int_mode instead of a machine_mode.
19120 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19121 (get_best_mode): Return a boolean and use a pointer argument to store
19122 the selected mode. Replace the limit mode parameter with a bit limit.
19123 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19124 for the values returned by bit_field_mode_iterator::next_mode.
19125 (store_bit_field): Update call to get_best_mode.
19126 (store_fixed_bit_field): Likewise.
19127 (extract_fixed_bit_field): Likewise.
19128 * expr.c (optimize_bitfield_assignment_op): Likewise.
19129 * fold-const.c (optimize_bit_field_compare): Likewise.
19130 (fold_truth_andor_1): Likewise.
19131 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19132 Update for new type of m_mode.
19133 (get_best_mode): As above.
19135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19136 Alan Hayward <alan.hayward@arm.com>
19137 David Sherwood <david.sherwood@arm.com>
19139 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19140 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19141 (store_bit_field): Check is_a <scalar_int_mode> before calling
19142 strict_volatile_bitfield_p.
19143 (extract_bit_field): Likewise.
19145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19146 Alan Hayward <alan.hayward@arm.com>
19147 David Sherwood <david.sherwood@arm.com>
19149 * target.def (cstore_mode): Return a scalar_int_mode.
19150 * doc/tm.texi: Regenerate.
19151 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19152 * targhooks.h (default_cstore_mode): Likewise.
19153 * targhooks.c (default_cstore_mode): Likewise, using a forced
19155 * expmed.c (emit_cstore): Expect the target of the cstore to be
19158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19159 Alan Hayward <alan.hayward@arm.com>
19160 David Sherwood <david.sherwood@arm.com>
19162 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19164 (niter_desc): Likewise mode.
19165 (iv_analyze): Add a mode parameter.
19167 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19168 and change its type to scalar_int_mode.
19169 * loop-iv.c: Update commentary at head of file.
19170 (iv_constant): Pass the mode paraeter before the rtx it describes
19171 and change its type to scalar_int_mode. Remove VOIDmode handling.
19172 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19173 (iv_extend): Likewise.
19174 (shorten_into_mode): Likewise.
19175 (iv_add): Use scalar_int_mode.
19176 (iv_mult): Likewise.
19177 (iv_shift): Likewise.
19178 (canonicalize_iv_subregs): Likewise.
19179 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19180 it describes and change its mode to scalar_int_mode. Also change
19181 the type of the returned inner_mode to scalar_int_mode.
19182 (get_biv_step): Likewise, turning outer_mode from a pointer
19183 into a direct parameter. Update call to get_biv_step_1.
19184 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19185 iv_constant and get_biv_step.
19186 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19187 and change its type to scalar_int_mode. Don't initialise iv->mode
19188 to VOIDmode and remove later checks for its still being VOIDmode.
19189 Update calls to iv_analyze_op and iv_analyze_expr. Check
19190 is_a <scalar_int_mode> when changing the mode under consideration.
19191 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19192 Update call to iv_analyze_expr.
19193 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19194 inner register is not also a scalar_int_mode. Update call to
19196 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19197 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19198 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19199 separate mode class checks. Update calls to iv_analyze. Remove
19200 fix-up of VOIDmodes after iv_analyze_biv.
19201 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19202 don't have a scalar_int_mode. Update call to biv_p.
19204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19205 Alan Hayward <alan.hayward@arm.com>
19206 David Sherwood <david.sherwood@arm.com>
19208 * cfgexpand.c (convert_debug_memory_address): Use
19209 as_a <scalar_int_mode>.
19210 * combine.c (expand_compound_operation): Likewise.
19211 (make_extraction): Likewise.
19212 (change_zero_ext): Likewise.
19213 (simplify_comparison): Likewise.
19214 * cse.c (cse_insn): Likewise.
19215 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19216 (mem_loc_descriptor): Likewise.
19217 (loc_descriptor): Likewise.
19218 * expmed.c (init_expmed_one_mode): Likewise.
19219 (synth_mult): Likewise.
19220 (emit_store_flag_1): Likewise.
19221 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19222 of a comparison with size.
19223 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19224 (reduce_to_bit_field_precision): Likewise.
19225 * function.c (expand_function_end): Likewise.
19226 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19227 * loop-doloop.c (doloop_modify): Likewise.
19228 * optabs.c (expand_binop): Likewise.
19229 (expand_unop): Likewise.
19230 (expand_copysign_absneg): Likewise.
19231 (prepare_cmp_insn): Likewise.
19232 (maybe_legitimize_operand): Likewise.
19233 * recog.c (const_scalar_int_operand): Likewise.
19234 * rtlanal.c (get_address_mode): Likewise.
19235 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19236 (simplify_cond_clz_ctz): Likewise.
19237 * tree-nested.c (get_nl_goto_field): Likewise.
19238 * tree.c (build_vector_type_for_mode): Likewise.
19239 * var-tracking.c (use_narrower_mode): Likewise.
19241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19242 Alan Hayward <alan.hayward@arm.com>
19243 David Sherwood <david.sherwood@arm.com>
19245 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19246 * builtins.c (expand_builtin_signbit): Use it.
19247 * cfgexpand.c (expand_debug_expr): Likewise.
19248 * dojump.c (do_jump): Likewise.
19249 (do_compare_and_jump): Likewise.
19250 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19251 * expmed.c (make_tree): Likewise.
19252 * expr.c (expand_expr_real_2): Likewise.
19253 (expand_expr_real_1): Likewise.
19254 (try_casesi): Likewise.
19255 * fold-const-call.c (fold_const_call_ss): Likewise.
19256 * fold-const.c (unextend): Likewise.
19257 (extract_muldiv_1): Likewise.
19258 (fold_single_bit_test): Likewise.
19259 (native_encode_int): Likewise.
19260 (native_encode_string): Likewise.
19261 (native_interpret_int): Likewise.
19262 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19263 * internal-fn.c (expand_addsub_overflow): Likewise.
19264 (expand_neg_overflow): Likewise.
19265 (expand_mul_overflow): Likewise.
19266 (expand_arith_overflow): Likewise.
19267 * match.pd: Likewise.
19268 * stor-layout.c (layout_type): Likewise.
19269 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19270 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19271 * tree-ssanames.c (get_range_info): Likewise.
19272 * tree-switch-conversion.c (array_value_type) Likewise.
19273 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19274 (vect_recog_divmod_pattern): Likewise.
19275 (vect_recog_mixed_size_cond_pattern): Likewise.
19276 * tree-vrp.c (extract_range_basic): Likewise.
19277 (simplify_float_conversion_using_ranges): Likewise.
19278 * tree.c (int_fits_type_p): Likewise.
19279 * ubsan.c (instrument_bool_enum_load): Likewise.
19280 * varasm.c (mergeable_string_section): Likewise.
19281 (narrowing_initializer_constant_valid_p): Likewise.
19282 (output_constant): Likewise.
19284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19285 Alan Hayward <alan.hayward@arm.com>
19286 David Sherwood <david.sherwood@arm.com>
19288 * machmode.h (NARROWEST_INT_MODE): New macro.
19289 * expr.c (alignment_for_piecewise_move): Use it instead of
19290 GET_CLASS_NARROWEST_MODE (MODE_INT).
19291 (push_block): Likewise.
19292 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19294 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19297 Alan Hayward <alan.hayward@arm.com>
19298 David Sherwood <david.sherwood@arm.com>
19300 * postreload.c (move2add_valid_value_p): Change the type of the
19301 mode parameter to scalar_int_mode.
19302 (move2add_use_add2_insn): Add a mode parameter and use it instead
19304 (move2add_use_add3_insn): Likewise.
19305 (reload_cse_move2add): Update accordingly.
19307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19308 Alan Hayward <alan.hayward@arm.com>
19309 David Sherwood <david.sherwood@arm.com>
19311 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19313 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19314 * optabs.c (expand_unop): Likewise.
19316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19317 Alan Hayward <alan.hayward@arm.com>
19318 David Sherwood <david.sherwood@arm.com>
19320 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19321 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19322 (popcount_loc_descriptor): Likewise.
19323 (bswap_loc_descriptor): Likewise.
19324 (rotate_loc_descriptor): Likewise.
19325 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19326 calling the functions above.
19328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19329 Alan Hayward <alan.hayward@arm.com>
19330 David Sherwood <david.sherwood@arm.com>
19332 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19334 (try_combine): Likewise.
19335 (simplify_if_then_else): Likewise.
19336 * cse.c (cse_insn): Likewise.
19337 * dwarf2out.c (mem_loc_descriptor): Likewise.
19338 * emit-rtl.c (gen_lowpart_common): Likewise.
19339 * simplify-rtx.c (simplify_truncation): Likewise.
19340 (simplify_binary_operation_1): Likewise.
19341 (simplify_const_relational_operation): Likewise.
19342 (simplify_ternary_operation): Likewise.
19343 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19346 Alan Hayward <alan.hayward@arm.com>
19347 David Sherwood <david.sherwood@arm.com>
19349 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19350 * reload.c (push_reload): Likewise.
19351 (find_reloads): Likewise.
19353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19354 Alan Hayward <alan.hayward@arm.com>
19355 David Sherwood <david.sherwood@arm.com>
19357 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19358 (make_compound_operation_int): Likewise.
19359 (change_zero_ext): Likewise.
19360 * expr.c (convert_move): Likewise.
19361 (convert_modes): Likewise.
19362 * fwprop.c (forward_propagate_subreg): Likewise.
19363 * loop-iv.c (get_biv_step_1): Likewise.
19364 * optabs.c (widen_operand): Likewise.
19365 * postreload.c (move2add_valid_value_p): Likewise.
19366 * recog.c (simplify_while_replacing): Likewise.
19367 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19368 (simplify_binary_operation_1): Likewise. Remove redundant
19369 mode equality check.
19371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19372 Alan Hayward <alan.hayward@arm.com>
19373 David Sherwood <david.sherwood@arm.com>
19375 * combine.c (combine_simplify_rtx): Add checks for
19376 is_a <scalar_int_mode>.
19377 (simplify_if_then_else): Likewise.
19378 (make_field_assignment): Likewise.
19379 (simplify_comparison): Likewise.
19380 * ifcvt.c (noce_try_bitop): Likewise.
19381 * loop-invariant.c (canonicalize_address_mult): Likewise.
19382 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19385 Alan Hayward <alan.hayward@arm.com>
19386 David Sherwood <david.sherwood@arm.com>
19388 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19389 is_a <scalar_int_mode> instead of != BLKmode.
19391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19392 Alan Hayward <alan.hayward@arm.com>
19393 David Sherwood <david.sherwood@arm.com>
19395 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19396 instead of != VOIDmode.
19397 * combine.c (if_then_else_cond): Likewise.
19398 (change_zero_ext): Likewise.
19399 * dwarf2out.c (mem_loc_descriptor): Likewise.
19400 (loc_descriptor): Likewise.
19401 * rtlanal.c (canonicalize_condition): Likewise.
19402 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19405 Alan Hayward <alan.hayward@arm.com>
19406 David Sherwood <david.sherwood@arm.com>
19408 * simplify-rtx.c (simplify_binary_operation_1): Use
19409 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19412 Alan Hayward <alan.hayward@arm.com>
19413 David Sherwood <david.sherwood@arm.com>
19415 * wide-int.h (int_traits<unsigned char>) New class.
19416 (int_traits<unsigned short>) Likewise.
19417 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19418 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19420 * combine.c (set_nonzero_bits_and_sign_copies): Use
19421 is_a <scalar_int_mode>.
19422 (find_split_point): Likewise.
19423 (combine_simplify_rtx): Likewise.
19424 (simplify_logical): Likewise.
19425 (expand_compound_operation): Likewise.
19426 (expand_field_assignment): Likewise.
19427 (make_compound_operation): Likewise.
19428 (extended_count): Likewise.
19429 (change_zero_ext): Likewise.
19430 (simplify_comparison): Likewise.
19431 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19432 (ucompare_loc_descriptor): Likewise.
19433 (minmax_loc_descriptor): Likewise.
19434 (mem_loc_descriptor): Likewise.
19435 (loc_descriptor): Likewise.
19436 * expmed.c (init_expmed_one_mode): Likewise.
19437 * lra-constraints.c (lra_constraint_offset): Likewise.
19438 * optabs.c (prepare_libcall_arg): Likewise.
19439 * postreload.c (move2add_note_store): Likewise.
19440 * reload.c (operands_match_p): Likewise.
19441 * rtl.h (load_extend_op): Likewise.
19442 * rtlhooks.c (gen_lowpart_general): Likewise.
19443 * simplify-rtx.c (simplify_truncation): Likewise.
19444 (simplify_unary_operation_1): Likewise.
19445 (simplify_binary_operation_1): Likewise.
19446 (simplify_const_binary_operation): Likewise.
19447 (simplify_const_relational_operation): Likewise.
19448 (simplify_subreg): Likewise.
19449 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19450 * var-tracking.c (adjust_mems): Likewise.
19451 (prepare_call_arguments): Likewise.
19453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19454 Alan Hayward <alan.hayward@arm.com>
19455 David Sherwood <david.sherwood@arm.com>
19457 * machmode.h (is_int_mode): New fuction.
19458 * combine.c (find_split_point): Use it.
19459 (combine_simplify_rtx): Likewise.
19460 (simplify_if_then_else): Likewise.
19461 (simplify_set): Likewise.
19462 (simplify_shift_const_1): Likewise.
19463 (simplify_comparison): Likewise.
19464 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19465 * cse.c (notreg_cost): Likewise.
19466 (cse_insn): Likewise.
19467 * cselib.c (cselib_lookup_1): Likewise.
19468 * dojump.c (do_jump_1): Likewise.
19469 (do_compare_rtx_and_jump): Likewise.
19470 * dse.c (get_call_args): Likewise.
19471 * dwarf2out.c (rtl_for_decl_init): Likewise.
19472 (native_encode_initializer): Likewise.
19473 * expmed.c (emit_store_flag_1): Likewise.
19474 (emit_store_flag): Likewise.
19475 * expr.c (convert_modes): Likewise.
19476 (store_field): Likewise.
19477 (expand_expr_real_1): Likewise.
19478 * fold-const.c (fold_read_from_constant_string): Likewise.
19479 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19480 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19481 * optabs.c (expand_binop): Likewise.
19482 (expand_unop): Likewise.
19483 (expand_abs_nojump): Likewise.
19484 (expand_one_cmpl_abs_nojump): Likewise.
19485 * simplify-rtx.c (mode_signbit_p): Likewise.
19486 (val_signbit_p): Likewise.
19487 (val_signbit_known_set_p): Likewise.
19488 (val_signbit_known_clear_p): Likewise.
19489 (simplify_relational_operation_1): Likewise.
19490 * tree.c (vector_type_mode): Likewise.
19492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19493 Alan Hayward <alan.hayward@arm.com>
19494 David Sherwood <david.sherwood@arm.com>
19496 * machmode.h (smallest_mode_for_size): Fix formatting.
19497 (smallest_int_mode_for_size): New function.
19498 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19499 instead of smallest_mode_for_size.
19500 * combine.c (make_extraction): Likewise.
19501 * config/arc/arc.c (arc_expand_movmem): Likewise.
19502 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19503 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19504 * config/s390/s390.c (s390_expand_insv): Likewise.
19505 * config/sparc/sparc.c (assign_int_registers): Likewise.
19506 * config/spu/spu.c (spu_function_value): Likewise.
19507 (spu_function_arg): Likewise.
19508 * coverage.c (get_gcov_type): Likewise.
19509 (get_gcov_unsigned_t): Likewise.
19510 * dse.c (find_shift_sequence): Likewise.
19511 * expmed.c (store_bit_field_1): Likewise.
19512 * expr.c (convert_move): Likewise.
19513 (store_field): Likewise.
19514 * internal-fn.c (expand_arith_overflow): Likewise.
19515 * optabs-query.c (get_best_extraction_insn): Likewise.
19516 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19517 * stor-layout.c (layout_type): Likewise.
19518 (initialize_sizetypes): Likewise.
19519 * targhooks.c (default_get_mask_mode): Likewise.
19520 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19523 Alan Hayward <alan.hayward@arm.com>
19524 David Sherwood <david.sherwood@arm.com>
19526 * machmode.h (opt_mode::else_blk): New function.
19527 (int_mode_for_mode): Declare.
19528 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19529 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19531 * cfgexpand.c (expand_debug_expr): Likewise.
19532 * combine.c (gen_lowpart_or_truncate): Likewise.
19533 (gen_lowpart_for_combine): Likewise.
19534 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19535 * config/avr/avr.c (avr_to_int_mode): Likewise.
19536 (avr_out_plus_1): Likewise.
19537 (avr_out_plus): Likewise.
19538 (avr_out_round): Likewise.
19539 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19540 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19541 (s390_expand_vcond): Likewise.
19542 * config/spu/spu.c (spu_split_immediate): Likewise.
19543 (spu_expand_mov): Likewise.
19544 * dse.c (get_stored_val): Likewise.
19545 * expmed.c (store_bit_field_1): Likewise.
19546 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19548 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19549 (extract_low_bits): Likewise.
19550 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19551 handling rather than repeating the check.
19552 (emit_group_store): Likewise.
19553 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19554 * optabs.c (expand_absneg_bit): Likewise.
19555 (expand_copysign_absneg): Likewise.
19556 (expand_copysign_bit): Likewise.
19557 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19558 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19559 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19560 * var-tracking.c (prepare_call_arguments): Likewise.
19561 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19562 int_mode_for_mode instead of mode_for_size.
19563 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19566 Alan Hayward <alan.hayward@arm.com>
19567 David Sherwood <david.sherwood@arm.com>
19569 * machmode.h (int_mode_for_size): New function.
19570 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19571 instead of mode_for_size.
19572 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19574 * combine.c (expand_field_assignment): Use int_mode_for_size
19575 instead of mode_for_size.
19576 (make_extraction): Likewise.
19577 (simplify_shift_const_1): Likewise.
19578 (simplify_comparison): Likewise.
19579 * dojump.c (do_jump): Likewise.
19580 * dwarf2out.c (mem_loc_descriptor): Likewise.
19581 * emit-rtl.c (init_derived_machine_modes): Likewise.
19582 * expmed.c (flip_storage_order): Likewise.
19583 (convert_extracted_bit_field): Likewise.
19584 * expr.c (copy_blkmode_from_reg): Likewise.
19585 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19586 * internal-fn.c (expand_mul_overflow): Likewise.
19587 * lower-subreg.c (simple_move): Likewise.
19588 * optabs-libfuncs.c (init_optabs): Likewise.
19589 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19590 * tree.c (vector_type_mode): Likewise.
19591 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19592 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19593 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19594 * tree-vect-stmts.c (vectorizable_load): Likewise.
19595 (vectorizable_store): Likewise.
19597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19598 Alan Hayward <alan.hayward@arm.com>
19599 David Sherwood <david.sherwood@arm.com>
19601 * coretypes.h (pod_mode): New type.
19602 (scalar_int_mode_pod): New typedef.
19603 * machmode.h (pod_mode): New class.
19604 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19605 * genmodes.c (emit_mode_int_n): Update accordingly.
19606 * lower-subreg.h (target_lower_subreg): Change type to
19607 scalar_int_mode_pod.
19608 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19609 scalar_int_mode_pod.
19611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19612 Alan Hayward <alan.hayward@arm.com>
19613 David Sherwood <david.sherwood@arm.com>
19615 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19616 machine_mode to scalar_int_mode.
19617 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19618 (rs6000_option_override_internal): Remove cast to int.
19619 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19620 machine_mode to scalar_int_mode.
19621 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19622 (rs6000_option_override_internal): Remove cast to int.
19623 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19624 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19626 * config/s390/s390.c (s390_expand_builtin): Likewise.
19627 * coretypes.h (scalar_int_mode): New type.
19628 (opt_scalar_int_mode): New typedef.
19629 * machmode.h (scalar_int_mode): New class.
19630 (scalar_int_mode::includes_p): New function.
19631 (byte_mode): Change type to scalar_int_mode.
19632 (word_mode): Likewise.
19633 (ptr_mode): Likewise.
19634 * emit-rtl.c (byte_mode): Likewise.
19635 (word_mode): Likewise.
19636 (ptr_mode): Likewise.
19637 (init_derived_machine_modes): Update accordingly.
19638 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19639 and MODE_PARTIAL_INT.
19640 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19641 opt_scalar_int_mode.
19643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19644 Alan Hayward <alan.hayward@arm.com>
19645 David Sherwood <david.sherwood@arm.com>
19647 * target.def (libgcc_floating_mode_supported_p): Take a
19649 * doc/tm.texi: Regenerate.
19650 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19652 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19653 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19657 Alan Hayward <alan.hayward@arm.com>
19658 David Sherwood <david.sherwood@arm.com>
19660 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19661 * doc/tm.texi: Regenerate.
19662 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19663 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19664 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19665 * targhooks.h (default_floatn_mode): Likewise.
19666 * targhooks.c (default_floatn_mode): Likewise.
19667 * tree.c (build_common_tree_nodes): Update accordingly.
19669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19670 Alan Hayward <alan.hayward@arm.com>
19671 David Sherwood <david.sherwood@arm.com>
19673 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19674 (mode_iterator::iterate_p): Likewise.
19675 (mode_iterator::get_wider): Likewise.
19676 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19679 Alan Hayward <alan.hayward@arm.com>
19680 David Sherwood <david.sherwood@arm.com>
19682 * coretypes.h (opt_scalar_float_mode): New typedef.
19683 * machmode.h (float_mode_for_size): New function.
19684 * emit-rtl.c (double_mode): Delete.
19685 (init_emit_once): Use float_mode_for_size.
19686 * stor-layout.c (layout_type): Likewise.
19687 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19689 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19690 Alan Hayward <alan.hayward@arm.com>
19691 David Sherwood <david.sherwood@arm.com>
19693 * output.h (assemble_real): Take a scalar_float_mode.
19694 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19695 * config/arm/arm.md (consttable_4): Likewise.
19696 (consttable_8): Likewise.
19697 (consttable_16): Likewise.
19698 * config/mips/mips.md (consttable_float): Likewise.
19699 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19700 * varasm.c (assemble_real): Take a scalar_float_mode.
19701 (output_constant_pool_2): Update accordingly.
19702 (output_constant): Likewise.
19704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19705 Alan Hayward <alan.hayward@arm.com>
19706 David Sherwood <david.sherwood@arm.com>
19708 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19709 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19710 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19711 (native_encode_real): Likewise.
19712 (native_interpret_real): Likewise.
19713 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19714 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19717 Alan Hayward <alan.hayward@arm.com>
19718 David Sherwood <david.sherwood@arm.com>
19720 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19721 <scalar_float_mode>. Simplify.
19722 (gen_extend_conv_libfunc): Likewise.
19724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19725 Alan Hayward <alan.hayward@arm.com>
19726 David Sherwood <david.sherwood@arm.com>
19728 * coretypes.h (scalar_float_mode): New type.
19729 * machmode.h (mode_traits::from_int): Use machine_mode if
19730 USE_ENUM_MODES is defined.
19731 (is_a): New function.
19733 (dyn_cast): Likewise.
19734 (scalar_float_mode): New class.
19735 (scalar_float_mode::includes_p): New function.
19736 (is_float_mode): Likewise.
19737 * gdbhooks.py (MachineModePrinter): New class.
19738 (build_pretty_printer): Use it for scalar_float_mode.
19739 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19740 (format_helper::format_helper): Turn into a template.
19741 * genmodes.c (get_mode_class): New function.
19742 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19743 or machine_mode if none.
19744 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19745 as_a <scalar_float_mode>.
19746 * dwarf2out.c (mem_loc_descriptor): Likewise.
19747 (insert_float): Likewise.
19748 (add_const_value_attribute): Likewise.
19749 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19750 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19751 (expand_unop): Update accordingly.
19752 (expand_abs_nojump): Likewise.
19753 (expand_copysign_absneg): Take a scalar_float_mode.
19754 (expand_copysign_bit): Likewise.
19755 (expand_copysign): Update accordingly.
19757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19758 Alan Hayward <alan.hayward@arm.com>
19759 David Sherwood <david.sherwood@arm.com>
19761 * coretypes.h (opt_mode): New class.
19762 * machmode.h (opt_mode): Likewise.
19763 (opt_mode::else_void): New function.
19764 (opt_mode::require): Likewise.
19765 (opt_mode::exists): Likewise.
19766 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19767 (GET_MODE_2XWIDER_MODE): Likewise.
19768 (mode_iterator::get_wider): Update accordingly.
19769 (mode_iterator::get_2xwider): Likewise.
19770 (mode_iterator::get_known_wider): Likewise, turning into a template.
19771 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19772 forcing a wider mode to exist.
19773 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19774 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19775 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19776 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19777 * lower-subreg.c (init_lower_subreg): Likewise.
19778 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19779 on the final iteration.
19780 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19781 a wider mode exists before asking for a move pattern.
19782 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19783 forcing a wider mode to exist.
19784 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19785 returning false if no such mode exists.
19786 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19787 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19788 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19789 Avoid checking for a MODE_INT if we already know the mode is not a
19791 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19792 forcing a wider mode to exist.
19793 (expmed_mult_highpart_optab): Likewise.
19794 (expmed_mult_highpart): Likewise.
19795 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19797 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19798 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19799 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19800 * internal-fn.c (expand_mul_overflow): Update use of
19801 GET_MODE_2XWIDER_MODE.
19802 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19803 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19804 GET_MODE_WIDER_MODE.
19805 (convert_plusminus_to_widen): Likewise.
19806 * tree-switch-conversion.c (array_value_type): Likewise.
19807 * var-tracking.c (emit_note_insn_var_location): Likewise.
19808 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19809 Return false inside rather than outside the loop if no wider mode
19811 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19812 and GET_MODE_2XWIDER_MODE
19813 (can_compare_p): Use else_void.
19814 * gdbhooks.py (OptMachineModePrinter): New class.
19815 (build_pretty_printer): Use it for opt_mode.
19817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19818 Alan Hayward <alan.hayward@arm.com>
19819 David Sherwood <david.sherwood@arm.com>
19821 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19822 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19825 Alan Hayward <alan.hayward@arm.com>
19826 David Sherwood <david.sherwood@arm.com>
19828 * machmode.h (mode_traits): New structure.
19829 (get_narrowest_mode): New function.
19830 (mode_iterator::start): Likewise.
19831 (mode_iterator::iterate_p): Likewise.
19832 (mode_iterator::get_wider): Likewise.
19833 (mode_iterator::get_known_wider): Likewise.
19834 (mode_iterator::get_2xwider): Likewise.
19835 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19836 (FOR_EACH_MODE): Likewise.
19837 (FOR_EACH_MODE_FROM): Likewise.
19838 (FOR_EACH_MODE_UNTIL): Likewise.
19839 (FOR_EACH_WIDER_MODE): Likewise.
19840 (FOR_EACH_2XWIDER_MODE): Likewise.
19841 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19842 * combine.c (simplify_comparison): Likewise
19843 * config/i386/i386.c (type_natural_mode): Likewise.
19844 * cse.c (cse_insn): Likewise.
19845 * dse.c (find_shift_sequence): Likewise.
19846 * emit-rtl.c (init_derived_machine_modes): Likewise.
19847 (init_emit_once): Likewise.
19848 * explow.c (hard_function_value): Likewise.
19849 * expmed.c (extract_fixed_bit_field_1): Likewise.
19850 (extract_bit_field_1): Likewise.
19851 (expand_divmod): Likewise.
19852 (emit_store_flag_1): Likewise.
19853 * expr.c (init_expr_target): Likewise.
19854 (convert_move): Likewise.
19855 (alignment_for_piecewise_move): Likewise.
19856 (widest_int_mode_for_size): Likewise.
19857 (emit_block_move_via_movmem): Likewise.
19858 (copy_blkmode_to_reg): Likewise.
19859 (set_storage_via_setmem): Likewise.
19860 (compress_float_constant): Likewise.
19861 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19862 * optabs-query.c (get_best_extraction_insn): Likewise.
19863 * optabs.c (expand_binop): Likewise.
19864 (expand_twoval_unop): Likewise.
19865 (expand_twoval_binop): Likewise.
19866 (widen_leading): Likewise.
19867 (widen_bswap): Likewise.
19868 (expand_parity): Likewise.
19869 (expand_unop): Likewise.
19870 (prepare_cmp_insn): Likewise.
19871 (prepare_float_lib_cmp): Likewise.
19872 (expand_float): Likewise.
19873 (expand_fix): Likewise.
19874 (expand_sfix_optab): Likewise.
19875 * postreload.c (move2add_use_add2_insn): Likewise.
19876 * reg-stack.c (reg_to_stack): Likewise.
19877 * reginfo.c (choose_hard_reg_mode): Likewise.
19878 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19879 * stor-layout.c (mode_for_size): Likewise.
19880 (smallest_mode_for_size): Likewise.
19881 (mode_for_vector): Likewise.
19882 (finish_bitfield_representative): Likewise.
19883 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19884 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19885 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19886 * var-tracking.c (prepare_call_arguments): Likewise.
19888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19889 Alan Hayward <alan.hayward@arm.com>
19890 David Sherwood <david.sherwood@arm.com>
19892 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19893 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19894 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19895 * machmode.h (mode_size): Move earlier in file.
19896 (mode_precision): Likewise.
19897 (mode_inner): Likewise.
19898 (mode_nunits): Likewise.
19899 (mode_unit_size): Likewise.
19900 (unit_unit_precision): Likewise.
19901 (mode_wider): Likewise.
19902 (mode_2xwider): Likewise.
19903 (machine_mode): New class.
19904 (mode_to_bytes): New function.
19905 (mode_to_bits): Likewise.
19906 (mode_to_precision): Likewise.
19907 (mode_to_inner): Likewise.
19908 (mode_to_unit_size): Likewise.
19909 (mode_to_unit_precision): Likewise.
19910 (mode_to_nunits): Likewise.
19911 (GET_MODE_SIZE): Use mode_to_bytes.
19912 (GET_MODE_BITSIZE): Use mode_to_bits.
19913 (GET_MODE_PRECISION): Use mode_to_precision.
19914 (GET_MODE_INNER): Use mode_to_inner.
19915 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19916 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19917 (GET_MODE_NUNITS): Use mode_to_nunits.
19918 * system.h (ALWAYS_INLINE): New macro.
19919 * config/powerpcspe/powerpcspe-c.c
19920 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19921 int for arg1_mode and arg2_mode.
19923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19924 Alan Hayward <alan.hayward@arm.com>
19925 David Sherwood <david.sherwood@arm.com>
19927 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19928 Prefix mode names with E_ in case statements.
19929 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19930 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19931 (aarch64_split_simd_move): Likewise.
19932 (aarch64_gen_storewb_pair): Likewise.
19933 (aarch64_gen_loadwb_pair): Likewise.
19934 (aarch64_gen_store_pair): Likewise.
19935 (aarch64_gen_load_pair): Likewise.
19936 (aarch64_get_condition_code_1): Likewise.
19937 (aarch64_constant_pool_reload_icode): Likewise.
19938 (get_rsqrte_type): Likewise.
19939 (get_rsqrts_type): Likewise.
19940 (get_recpe_type): Likewise.
19941 (get_recps_type): Likewise.
19942 (aarch64_gimplify_va_arg_expr): Likewise.
19943 (aarch64_simd_container_mode): Likewise.
19944 (aarch64_emit_load_exclusive): Likewise.
19945 (aarch64_emit_store_exclusive): Likewise.
19946 (aarch64_expand_compare_and_swap): Likewise.
19947 (aarch64_gen_atomic_cas): Likewise.
19948 (aarch64_emit_bic): Likewise.
19949 (aarch64_emit_atomic_swap): Likewise.
19950 (aarch64_emit_atomic_load_op): Likewise.
19951 (aarch64_evpc_trn): Likewise.
19952 (aarch64_evpc_uzp): Likewise.
19953 (aarch64_evpc_zip): Likewise.
19954 (aarch64_evpc_ext): Likewise.
19955 (aarch64_evpc_rev): Likewise.
19956 (aarch64_evpc_dup): Likewise.
19957 (aarch64_gen_ccmp_first): Likewise.
19958 (aarch64_gen_ccmp_next): Likewise.
19959 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19960 (alpha_emit_xfloating_libcall): Likewise.
19961 (emit_insxl): Likewise.
19962 (alpha_arg_type): Likewise.
19963 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19964 (arc_preferred_simd_mode): Likewise.
19965 (arc_secondary_reload): Likewise.
19966 (get_arc_condition_code): Likewise.
19967 (arc_print_operand): Likewise.
19968 (arc_legitimate_constant_p): Likewise.
19969 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19970 * config/arc/arc.md (casesi_load): Likewise.
19971 (casesi_compact_jump): Likewise.
19972 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19973 (cc_use_register): Likewise.
19974 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19975 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19976 (arm_init_iwmmxt_builtins): Likewise.
19977 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19978 (neon_expand_vector_init): Likewise.
19979 (arm_attr_length_move_neon): Likewise.
19980 (maybe_get_arm_condition_code): Likewise.
19981 (arm_emit_vector_const): Likewise.
19982 (arm_preferred_simd_mode): Likewise.
19983 (arm_output_iwmmxt_tinsr): Likewise.
19984 (thumb1_output_casesi): Likewise.
19985 (thumb2_output_casesi): Likewise.
19986 (arm_emit_load_exclusive): Likewise.
19987 (arm_emit_store_exclusive): Likewise.
19988 (arm_expand_compare_and_swap): Likewise.
19989 (arm_evpc_neon_vuzp): Likewise.
19990 (arm_evpc_neon_vzip): Likewise.
19991 (arm_evpc_neon_vrev): Likewise.
19992 (arm_evpc_neon_vtrn): Likewise.
19993 (arm_evpc_neon_vext): Likewise.
19994 (arm_validize_comparison): Likewise.
19995 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19996 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19997 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19998 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19999 (c6x_preferred_simd_mode): Likewise.
20000 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
20001 (epiphany_rtx_costs): Likewise.
20002 * config/epiphany/predicates.md (proper_comparison_operator):
20004 * config/frv/frv.c (condexec_memory_operand): Likewise.
20005 (frv_emit_move): Likewise.
20006 (output_move_single): Likewise.
20007 (output_condmove_single): Likewise.
20008 (frv_hard_regno_mode_ok): Likewise.
20009 (frv_matching_accg_mode): Likewise.
20010 * config/h8300/h8300.c (split_adds_subs): Likewise.
20011 (h8300_rtx_costs): Likewise.
20012 (h8300_print_operand): Likewise.
20013 (compute_mov_length): Likewise.
20014 (output_logical_op): Likewise.
20015 (compute_logical_op_length): Likewise.
20016 (compute_logical_op_cc): Likewise.
20017 (h8300_shift_needs_scratch_p): Likewise.
20018 (output_a_shift): Likewise.
20019 (compute_a_shift_length): Likewise.
20020 (compute_a_shift_cc): Likewise.
20021 (expand_a_rotate): Likewise.
20022 (output_a_rotate): Likewise.
20023 * config/i386/i386.c (classify_argument): Likewise.
20024 (function_arg_advance_32): Likewise.
20025 (function_arg_32): Likewise.
20026 (function_arg_64): Likewise.
20027 (function_value_64): Likewise.
20028 (ix86_gimplify_va_arg): Likewise.
20029 (ix86_legitimate_constant_p): Likewise.
20030 (put_condition_code): Likewise.
20031 (split_double_mode): Likewise.
20032 (ix86_avx256_split_vector_move_misalign): Likewise.
20033 (ix86_expand_vector_logical_operator): Likewise.
20034 (ix86_split_idivmod): Likewise.
20035 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
20036 (ix86_build_const_vector): Likewise.
20037 (ix86_build_signbit_mask): Likewise.
20038 (ix86_match_ccmode): Likewise.
20039 (ix86_cc_modes_compatible): Likewise.
20040 (ix86_expand_branch): Likewise.
20041 (ix86_expand_sse_cmp): Likewise.
20042 (ix86_expand_sse_movcc): Likewise.
20043 (ix86_expand_int_sse_cmp): Likewise.
20044 (ix86_expand_vec_perm_vpermi2): Likewise.
20045 (ix86_expand_vec_perm): Likewise.
20046 (ix86_expand_sse_unpack): Likewise.
20047 (ix86_expand_int_addcc): Likewise.
20048 (ix86_split_to_parts): Likewise.
20049 (ix86_vectorize_builtin_gather): Likewise.
20050 (ix86_vectorize_builtin_scatter): Likewise.
20051 (avx_vpermilp_parallel): Likewise.
20052 (inline_memory_move_cost): Likewise.
20053 (ix86_tieable_integer_mode_p): Likewise.
20054 (x86_maybe_negate_const_int): Likewise.
20055 (ix86_expand_vector_init_duplicate): Likewise.
20056 (ix86_expand_vector_init_one_nonzero): Likewise.
20057 (ix86_expand_vector_init_one_var): Likewise.
20058 (ix86_expand_vector_init_concat): Likewise.
20059 (ix86_expand_vector_init_interleave): Likewise.
20060 (ix86_expand_vector_init_general): Likewise.
20061 (ix86_expand_vector_set): Likewise.
20062 (ix86_expand_vector_extract): Likewise.
20063 (emit_reduc_half): Likewise.
20064 (ix86_emit_i387_round): Likewise.
20065 (ix86_mangle_type): Likewise.
20066 (ix86_expand_round_sse4): Likewise.
20067 (expand_vec_perm_blend): Likewise.
20068 (canonicalize_vector_int_perm): Likewise.
20069 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20070 (expand_vec_perm_1): Likewise.
20071 (expand_vec_perm_interleave3): Likewise.
20072 (expand_vec_perm_even_odd_pack): Likewise.
20073 (expand_vec_perm_even_odd_1): Likewise.
20074 (expand_vec_perm_broadcast_1): Likewise.
20075 (ix86_vectorize_vec_perm_const_ok): Likewise.
20076 (ix86_expand_vecop_qihi): Likewise.
20077 (ix86_expand_mul_widen_hilo): Likewise.
20078 (ix86_expand_sse2_abs): Likewise.
20079 (ix86_expand_pextr): Likewise.
20080 (ix86_expand_pinsr): Likewise.
20081 (ix86_preferred_simd_mode): Likewise.
20082 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20083 * config/i386/sse.md (*andnot<mode>3): Likewise.
20084 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20085 (*<code><mode>3): Likewise.
20086 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20087 (ia64_expand_atomic_op): Likewise.
20088 (ia64_arg_type): Likewise.
20089 (ia64_mode_to_int): Likewise.
20090 (ia64_scalar_mode_supported_p): Likewise.
20091 (ia64_vector_mode_supported_p): Likewise.
20092 (expand_vec_perm_broadcast): Likewise.
20093 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20094 (iq2000_function_arg_advance): Likewise.
20095 (iq2000_function_arg): Likewise.
20096 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20097 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20098 (m68k_libcall_value): Likewise.
20099 (m68k_function_value): Likewise.
20100 (sched_attr_op_type): Likewise.
20101 * config/mcore/mcore.c (mcore_output_move): Likewise.
20102 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20104 (microblaze_function_arg): Likewise.
20105 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20106 (mips_print_operand): Likewise.
20107 (mips_mode_ok_for_mov_fmt_p): Likewise.
20108 (mips_vector_mode_supported_p): Likewise.
20109 (mips_preferred_simd_mode): Likewise.
20110 (mips_expand_vpc_loongson_even_odd): Likewise.
20111 (mips_expand_vec_unpack): Likewise.
20112 (mips_expand_vi_broadcast): Likewise.
20113 (mips_expand_vector_init): Likewise.
20114 (mips_expand_vec_reduc): Likewise.
20115 (mips_expand_msa_cmp): Likewise.
20116 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20117 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20118 (cc_flags_for_mode): Likewise.
20119 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20120 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20121 (nds32_output_casesi_pc_relative): Likewise.
20122 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20123 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20124 (nvptx_gen_unpack): Likewise.
20125 (nvptx_gen_pack): Likewise.
20126 (nvptx_gen_shuffle): Likewise.
20127 (nvptx_gen_wcast): Likewise.
20128 (nvptx_preferred_simd_mode): Likewise.
20129 * config/pa/pa.c (pa_secondary_reload): Likewise.
20130 * config/pa/predicates.md (base14_operand): Likewise.
20131 * config/powerpcspe/powerpcspe-c.c
20132 (altivec_resolve_overloaded_builtin): Likewise.
20133 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20135 (rs6000_preferred_simd_mode): Likewise.
20136 (output_vec_const_move): Likewise.
20137 (rs6000_expand_vector_extract): Likewise.
20138 (rs6000_split_vec_extract_var): Likewise.
20139 (reg_offset_addressing_ok_p): Likewise.
20140 (rs6000_legitimate_offset_address_p): Likewise.
20141 (rs6000_legitimize_address): Likewise.
20142 (rs6000_emit_set_const): Likewise.
20143 (rs6000_const_vec): Likewise.
20144 (rs6000_emit_move): Likewise.
20145 (spe_build_register_parallel): Likewise.
20146 (rs6000_darwin64_record_arg_recurse): Likewise.
20147 (swap_selector_for_mode): Likewise.
20148 (spe_init_builtins): Likewise.
20149 (paired_init_builtins): Likewise.
20150 (altivec_init_builtins): Likewise.
20151 (do_load_for_compare): Likewise.
20152 (rs6000_generate_compare): Likewise.
20153 (rs6000_expand_float128_convert): Likewise.
20154 (emit_load_locked): Likewise.
20155 (emit_store_conditional): Likewise.
20156 (rs6000_output_function_epilogue): Likewise.
20157 (rs6000_handle_altivec_attribute): Likewise.
20158 (rs6000_function_value): Likewise.
20159 (emit_fusion_gpr_load): Likewise.
20160 (emit_fusion_p9_load): Likewise.
20161 (emit_fusion_p9_store): Likewise.
20162 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20163 (fusion_gpr_mem_load): Likewise.
20164 (fusion_addis_mem_combo_load): Likewise.
20165 (fusion_addis_mem_combo_store): Likewise.
20166 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20167 (fusion_gpr_mem_load): Likewise.
20168 (fusion_addis_mem_combo_load): Likewise.
20169 (fusion_addis_mem_combo_store): Likewise.
20170 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20172 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20173 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20174 (rs6000_preferred_simd_mode): Likewise.
20175 (output_vec_const_move): Likewise.
20176 (rs6000_expand_vector_extract): Likewise.
20177 (rs6000_split_vec_extract_var): Likewise.
20178 (reg_offset_addressing_ok_p): Likewise.
20179 (rs6000_legitimate_offset_address_p): Likewise.
20180 (rs6000_legitimize_address): Likewise.
20181 (rs6000_emit_set_const): Likewise.
20182 (rs6000_const_vec): Likewise.
20183 (rs6000_emit_move): Likewise.
20184 (rs6000_darwin64_record_arg_recurse): Likewise.
20185 (swap_selector_for_mode): Likewise.
20186 (paired_init_builtins): Likewise.
20187 (altivec_init_builtins): Likewise.
20188 (rs6000_expand_float128_convert): Likewise.
20189 (emit_load_locked): Likewise.
20190 (emit_store_conditional): Likewise.
20191 (rs6000_output_function_epilogue): Likewise.
20192 (rs6000_handle_altivec_attribute): Likewise.
20193 (rs6000_function_value): Likewise.
20194 (emit_fusion_gpr_load): Likewise.
20195 (emit_fusion_p9_load): Likewise.
20196 (emit_fusion_p9_store): Likewise.
20197 * config/rx/rx.c (rx_gen_move_template): Likewise.
20198 (flags_from_mode): Likewise.
20199 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20200 (s390_slb_comparison): Likewise.
20201 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20202 (s390_vector_mode_supported_p): Likewise.
20203 (s390_cc_modes_compatible): Likewise.
20204 (s390_match_ccmode_set): Likewise.
20205 (s390_canonicalize_comparison): Likewise.
20206 (s390_emit_compare_and_swap): Likewise.
20207 (s390_branch_condition_mask): Likewise.
20208 (s390_rtx_costs): Likewise.
20209 (s390_secondary_reload): Likewise.
20210 (__SECONDARY_RELOAD_CASE): Likewise.
20211 (s390_expand_cs): Likewise.
20212 (s390_preferred_simd_mode): Likewise.
20213 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20214 * config/sh/sh.c (sh_print_operand): Likewise.
20215 (dump_table): Likewise.
20216 (sh_secondary_reload): Likewise.
20217 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20218 * config/sh/sh.md (casesi_worker_1): Likewise.
20219 (casesi_worker_2): Likewise.
20220 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20221 (fcc_comparison_operator): Likewise.
20222 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20223 (emit_soft_tfmode_cvt): Likewise.
20224 (sparc_preferred_simd_mode): Likewise.
20225 (output_cbranch): Likewise.
20226 (sparc_print_operand): Likewise.
20227 (sparc_expand_vec_perm_bmask): Likewise.
20228 (vector_init_bshuffle): Likewise.
20229 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20230 (spu_vector_mode_supported_p): Likewise.
20231 (spu_expand_insv): Likewise.
20232 (spu_emit_branch_or_set): Likewise.
20233 (spu_handle_vector_attribute): Likewise.
20234 (spu_builtin_splats): Likewise.
20235 (spu_builtin_extract): Likewise.
20236 (spu_builtin_promote): Likewise.
20237 (spu_expand_sign_extend): Likewise.
20238 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20239 (tilegx_simd_int): Likewise.
20240 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20241 (tilepro_simd_int): Likewise.
20242 * config/v850/v850.c (const_double_split): Likewise.
20243 (v850_print_operand): Likewise.
20244 (ep_memory_offset): Likewise.
20245 * config/vax/vax.c (vax_rtx_costs): Likewise.
20246 (vax_output_int_move): Likewise.
20247 (vax_output_int_add): Likewise.
20248 (vax_output_int_subtract): Likewise.
20249 * config/visium/predicates.md (visium_branch_operator): Likewise.
20250 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20251 (visium_print_operand_address): Likewise.
20252 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20253 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20254 (xtensa_expand_conditional_branch): Likewise.
20255 (xtensa_copy_incoming_a7): Likewise.
20256 (xtensa_output_literal): Likewise.
20257 * dfp.c (decimal_real_maxval): Likewise.
20258 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20261 Alan Hayward <alan.hayward@arm.com>
20262 David Sherwood <david.sherwood@arm.com>
20264 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20265 (mode_nunits_inline): Likewise.
20266 (mode_inner_inline): Likewise.
20267 (mode_unit_size_inline): Likewise.
20268 (mode_unit_precision_inline): Likewise.
20269 (emit_insn_modes_h): Likewise. Also emit a #define of the
20271 (emit_mode_wider): Add an E_ prefix to mode names.
20272 (emit_mode_complex): Likewise.
20273 (emit_mode_inner): Likewise.
20274 (emit_mode_adjustments): Likewise.
20275 (emit_mode_int_n): Likewise.
20276 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20277 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20278 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20279 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20280 (CRC32_BUILTIN, ENTRY): Likewise.
20281 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20282 (aarch64_pop_regs): Likewise.
20283 (aarch64_process_components): Likewise.
20284 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20285 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20286 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20287 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20288 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20289 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20291 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20292 (pcmpistr): Likewise.
20293 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20294 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20295 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20297 * config/rl78/rl78.c (mduc_regs): Likewise.
20298 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20299 (htm_expand_builtin): Likewise.
20300 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20301 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20302 * config/xtensa/xtensa.c (print_operand): Likewise.
20303 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20304 (NUM_MODE_VECTOR_INT): Likewise.
20305 * genoutput.c (null_operand): Likewise.
20306 (output_operand_data): Likewise.
20307 * genrecog.c (print_parameter_value): Likewise.
20308 * lra.c (debug_operand_data): Likewise.
20310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20312 * dwarf2out.c (scompare_loc_descriptor_wide)
20313 (scompare_loc_descriptor_narrow): New functions, split out from...
20314 (scompare_loc_descriptor): ...here.
20315 * expmed.c (emit_store_flag_int): New function, split out from...
20316 (emit_store_flag): ...here.
20318 2017-08-30 Richard Biener <rguenther@suse.de>
20320 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20321 (dwarf2out_early_finish): Move setting of AT_pubnames from
20322 early debug output to early finish.
20324 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20326 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20327 and -mdata-region to the assembler.
20329 2017-08-30 Richard Biener <rguenther@suse.de>
20331 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20333 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20334 (dwarf2out_early_global_decl): Only generate a DIE for the
20335 abstract origin if it doesn't already exist or is a declaration DIE.
20336 (resolve_addr): Do not add the linkage name twice when
20337 generating a stub DIE for the DW_TAG_GNU_call_site target.
20339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20341 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20342 Use machine_mode rather than int for arg1_mode.
20344 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20347 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20348 that the second argument of the built-in functions to unpack
20349 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20350 switch statement instead a lot of if statements.
20351 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20352 Allow 64-bit values to be in Altivec registers as well as
20353 traditional floating point registers.
20354 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20356 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20358 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20359 MAX_REGS_PER_ADDRESS == 1.
20361 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20363 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20364 * config/i386/i386.c (ix86_option_override_internal): Simplify
20365 setting of opts->x_flag_entry.
20367 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20368 Jakub Jelinek <jakub@redhat.com>
20369 Richard Biener <rguenther@suse.de>
20371 PR tree-optimization/81503
20372 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20373 folded constant fits in the target type; reorder tests for clarity.
20375 2017-08-29 Martin Liska <mliska@suse.cz>
20377 * passes.def: Include pass_lower_switch.
20378 * stmt.c (dump_case_nodes): Remove and move to
20379 tree-switch-conversion.
20380 (case_values_threshold): Likewise.
20381 (expand_switch_as_decision_tree_p): Likewise.
20382 (emit_case_decision_tree): Likewise.
20383 (expand_case): Likewise.
20384 (balance_case_nodes): Likewise.
20385 (node_has_low_bound): Likewise.
20386 (node_has_high_bound): Likewise.
20387 (node_is_bounded): Likewise.
20388 (emit_case_nodes): Likewise.
20389 (struct simple_case_node): New struct.
20390 (add_case_node): Remove.
20391 (emit_case_dispatch_table): Use vector instead of case_list.
20392 (reset_out_edges_aux): Remove.
20393 (compute_cases_per_edge): Likewise.
20394 (expand_case): Build list of simple_case_node.
20395 (expand_sjlj_dispatch_table): Use it.
20396 * tree-switch-conversion.c (struct case_node): Moved from
20397 stmt.c and adjusted.
20398 (emit_case_nodes): Likewise.
20399 (node_has_low_bound): Likewise.
20400 (node_has_high_bound): Likewise.
20401 (node_is_bounded): Likewise.
20402 (case_values_threshold): Likewise.
20403 (reset_out_edges_aux): Likewise.
20404 (compute_cases_per_edge): Likewise.
20405 (add_case_node): Likewise.
20406 (dump_case_nodes): Likewise.
20407 (balance_case_nodes): Likewise.
20408 (expand_switch_as_decision_tree_p): Likewise.
20409 (emit_jump): Likewise.
20410 (emit_case_decision_tree): Likewise.
20411 (try_switch_expansion): Likewise.
20412 (do_jump_if_equal): Likewise.
20413 (emit_cmp_and_jump_insns): Likewise.
20414 (fix_phi_operands_for_edge): New function.
20415 (record_phi_operand_mapping): Likewise.
20416 (class pass_lower_switch): New pass.
20417 (pass_lower_switch::execute): New function.
20418 (make_pass_lower_switch): Likewise.
20419 (conditional_probability):
20420 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20421 * tree-pass.h: Add make_pass_lower_switch.
20423 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20426 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20429 2017-08-29 Richard Biener <rguenther@suse.de>
20431 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20432 we do not add a DW_AT_inline attribute twice.
20433 (gen_subprogram_die): Remove code setting DW_AT_inline on
20434 DECL_ABSTRACT_P nodes.
20436 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20438 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20439 calls to internal functions.
20440 (gimplify_modify_expr): Likewise.
20441 * tree-call-cdce.c (use_internal_fn): Likewise.
20442 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20443 (convert_to_divmod): Set the nothrow flag.
20444 * tree-if-conv.c (predicate_mem_writes): Likewise.
20445 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20446 (vectorizable_call): Likewise.
20447 (vectorizable_store): Likewise.
20448 (vectorizable_load): Likewise.
20449 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20450 (vect_recog_mask_conversion_pattern): Likewise.
20452 2017-08-29 Martin Liska <mliska@suse.cz>
20455 * gcc.c (driver_handle_option): Add new argument.
20456 * opts-common.c (handle_option): Pass
20457 target_option_override_hook.
20458 * opts-global.c (lang_handle_option): Add new option.
20459 (set_default_handlers): Add new argument.
20460 (decode_options): Likewise.
20461 * opts.c (target_handle_option): Likewise.
20462 (common_handle_option): Call target_option_override_hook.
20463 * opts.h (struct cl_option_handler_func): Add hook for
20464 target option override.
20465 (struct cl_option_handlers): Likewise.
20466 (set_default_handlers): Add new argument.
20467 (decode_options): Likewise.
20468 (common_handle_option): Likewise.
20469 (target_handle_option): Likewise.
20470 * toplev.c (toplev::main): Pass targetm.target_option.override
20473 2017-08-29 Richard Biener <rguenther@suse.de>
20474 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20476 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20477 life to the active subtree.
20479 2017-08-28 Jeff Law <law@redhat.com>
20481 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20482 derive_equivalences.
20483 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20484 Code moved into....
20485 (edge_info::derive_equivalences): New private member function
20487 * tree-ssa-dom.c (class edge_info): Changed from a struct
20488 to a class. Add ctor/dtor, methods and data members.
20489 (edge_info::edge_info): Renamed from allocate_edge_info.
20490 Initialize additional members.
20491 (edge_info::~edge_info): New.
20492 (free_dom_edge_info): Delete the edge info.
20493 (record_edge_info): Use new class & associated member functions.
20494 Tighten forms for testing for edge equivalences.
20495 (record_temporary_equivalences): Iterate over the simple
20496 equivalences rather than assuming there's only one per edge.
20497 (cprop_into_successor_phis): Iterate over the simple
20498 equivalences rather than assuming there's only one per edge.
20499 (optimize_stmt): Use operand_equal_p rather than pointer
20500 equality for mini-DSE code.
20502 2017-08-28 Nathan Sidwell <nathan@acm.org>
20504 * gcc.c (execute): Fold SIGPIPE handling into switch
20505 statement. Adjust internal error message.
20507 2017-08-28 Richard Biener <rguenther@suse.de>
20510 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20511 Do nothing for removed DIEs.
20513 2017-08-28 Richard Biener <rguenther@suse.de>
20515 PR tree-optimization/81977
20516 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20519 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20522 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20523 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20524 using targetm.gen_mem_thread_fence.
20526 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20529 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20530 predicate to register_operand. Reorder operands.
20531 (*btr<mode>): Ditto.
20532 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20533 (*btr<mode>_mask): Ditto.
20535 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20537 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20538 * config/rs6000/xmmintrin.h: New file.
20539 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20541 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20544 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20545 parameter and_insn and return it.
20546 (recombine_lvx_pattern): Insert a copy to ensure availability of
20547 the base register of the copied masking operation at the point of
20548 the instruction replacement.
20549 (recombine_stvx_pattern): Likewise.
20551 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20553 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20554 undocumented switches.
20555 (-mpower9-dform-vector): Likewise.
20556 (-mpower9-dform): Likewise.
20557 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20558 comments to delete references to -mpower9-dform* switches.
20559 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20560 Delete reference to -mpower9-dform* switches, test for
20561 -mpower9-vector instead.
20562 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20563 (OTHER_P9_VECTOR_MASKS): Likewise.
20564 (POWERPC_MASKS): Likewise.
20565 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20566 tests against -mpower9-dform* to -mpower9-vector. Delete code
20567 that checked for -mpower9-dform* consistancy with other options.
20568 Add test for -mpower9-misc to enable other power9 switches.
20569 (rs6000_init_hard_regno_mode_ok): Likewise.
20570 (rs6000_option_override_internal): Likewise.
20571 (rs6000_emit_prologue): Likewise.
20572 (rs6000_emit_epilogue): Likewise.
20573 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20574 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20575 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20577 (emit_fusion_p9_store): Likewise.
20578 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20579 resetting these macros if the assembler does not support ISA 3.0
20581 (TARGET_P9_DFORM_VECTOR): Likewise.
20582 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20583 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20585 2017-08-25 Alan Modra <amodra@gmail.com>
20587 PR rtl-optimization/81747
20588 * cse.c (cse_extended_basic_block): Don't attempt to record
20589 equivalences for degenerate conditional branches that branch
20590 to their fall-through.
20592 2017-08-24 Martin Sebor <msebor@redhat.com>
20594 PR middle-end/81908
20595 * gimple-fold.c (size_must_be_zero_p): New function.
20596 (gimple_fold_builtin_memory_op): Call it.
20598 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20600 * config/rs6000/mm_malloc.h: New file.
20602 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20604 PR tree-optimization/81913
20605 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20606 analysis when either IVs in condition can wrap.
20608 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20610 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20611 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20613 2017-08-24 Richard Biener <rguenther@suse.de>
20616 * targhooks.c (default_target_can_inline_p): Properly
20617 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20618 is present and always compare.
20619 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20620 infer -mfpmath=sse from TARGET_SSE_P.
20621 (ix86_can_inline_p): Properly use target_option_default_node when
20622 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20624 2017-08-24 Richard Biener <rguenther@suse.de>
20627 * dwarf2out.c (output_die): Handle flag_generate_offload like
20629 (output_comp_unit): Likewise.
20630 (gen_array_type_die): Likewise.
20631 (dwarf2out_early_finish): Likewise.
20632 (note_variable_value_in_expr): Likewise.
20633 (dwarf2out_finish): Likewise. Adjust assert.
20634 * cgraphunit.c (symbol_table::compile): Move setting of
20635 flag_generate_offload earlier ...
20636 (symbol_table::finalize_compilation_unit): ... here, before
20637 early debug finalization.
20639 2017-08-24 Richard Biener <rguenther@suse.de>
20641 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20642 and ipa-fnsummary.h.
20643 (ix86_can_inline_p): When ix86_fpmath flags do not match
20644 check whether the callee uses FP math at all.
20646 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20648 PR middle-end/81931
20649 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20650 instead of TYPE_PRECISION.
20652 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20653 Alan Hayward <alan.hayward@arm.com>
20654 David Sherwood <david.sherwood@arm.com>
20656 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20658 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20660 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20662 (ix86_frame::stack_realign_allocate): New field.
20663 (struct machine_frame_state): Modify comments.
20664 (machine_frame_state::sp_realigned_fp_end): New field.
20665 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20666 layout calculation.
20667 (sp_valid_at): Add assertion to assure no attempt to access invalid
20668 offset of a realigned stack.
20669 (fp_valid_at): Likewise.
20670 (choose_baseaddr): Modify comments.
20671 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20672 ix86_expand_prologue.
20673 (ix86_expand_prologue): Modify stack realignment and allocation.
20674 (ix86_expand_epilogue): Modify comments.
20675 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20676 avx2_runtime, avx512f, and avx512f_runtime.
20678 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20680 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20681 (mstackrealign): Do not init to -1.
20682 * config/i386/i386.c (ix86_option_override_internal):
20683 Check opts_set, not opts when setting default value of
20684 opts->x_ix86_force_align_arg_pointer.
20686 2017-08-23 Richard Biener <rguenther@suse.de>
20688 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20689 lang_hooks.decl_printable_name.
20690 * print-rtl-function.c (print_rtx_function): Likewise.
20691 * tree-pretty-print.c (dump_function_header): Likewise.
20693 2017-08-23 Richard Biener <rguenther@suse.de>
20696 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20697 -g0 at compile-time.
20699 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20701 PR middle-end/19706
20702 * doc/sourcebuild.texi (Other hardware attributes):
20705 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20707 PR middle-end/19706
20708 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20709 Fix single-use check.
20711 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20713 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20715 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20717 * doc/install.texi: Modify to add more details on running selected
20720 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20722 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20723 is combined with -mabi=ms.
20724 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20727 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20729 PR tree-optimization/81488
20730 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20731 and cached_basis fields.
20732 (MAX_SPREAD): New constant.
20733 (alloc_cand_and_find_basis): Initialize new fields.
20734 (clear_visited): New function.
20735 (create_phi_basis_1): Rename from create_phi_basis, set visited
20736 and cached_basis fields.
20737 (create_phi_basis): New wrapper function.
20738 (phi_add_costs_1): Rename from phi_add_costs, add spread
20739 parameter, set visited field, short-circuit when limits reached.
20740 (phi_add_costs): New wrapper function.
20741 (record_phi_increments_1): Rename from record_phi_increments, set
20743 (record_phi_increments): New wrapper function.
20744 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20745 (phi_incr_cost): New wrapper function.
20746 (all_phi_incrs_profitable_1): Rename from
20747 all_phi_incrs_profitable, set visited field.
20748 (all_phi_incrs_profitable): New wrapper function.
20750 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20751 Alan Hayward <alan.hayward@arm.com>
20752 David Sherwood <david.sherwood@arm.com>
20754 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20755 that takes the outer and inner modes.
20756 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20757 comparison as the canonical test for a paradoxical subreg.
20758 * combine.c (simplify_set): Use paradoxical_subreg_p.
20759 (make_extraction): Likewise.
20760 (force_to_mode): Likewise.
20761 (rtx_equal_for_field_assignment_p): Likewise.
20762 (gen_lowpart_for_combine): Likewise.
20763 (simplify_comparison): Likewise.
20764 * cse.c (equiv_constant): Likewise.
20765 * expmed.c (store_bit_field_1): Likewise.
20766 * final.c (alter_subreg): Likewise.
20767 * fwprop.c (propagate_rtx): Likewise.
20768 (forward_propagate_subreg): Likewise.
20769 * ira-conflicts.c (ira_build_conflicts): Likewise.
20770 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20771 * lra-constraints.c (curr_insn_transform): Likewise.
20772 (split_reg): Likewise.
20773 * lra-eliminations.c (move_plus_up): Likewise.
20774 (lra_eliminate_regs_1): Likewise.
20775 * recog.c (general_operand): Likewise.
20776 * ree.c (combine_reaching_defs): Likewise.
20777 * reload.c (push_reload): Likewise.
20778 (find_reloads): Likewise.
20779 * reload1.c (elimination_effects): Likewise.
20780 (compute_reload_subreg_offset): Likewise.
20781 (choose_reload_regs): Likewise.
20782 * rtlanal.c (subreg_lsb_1): Likewise.
20783 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20784 (simplify_subreg): Likewise.
20785 * var-tracking.c (track_loc_p): Likewise.
20786 * emit-rtl.c (byte_lowpart_offset): Likewise.
20787 (paradoxical_subreg_p): Delete out-of-line definition.
20789 2017-08-22 Jeff Law <law@redhat.com>
20791 PR tree-optimization/81741
20792 PR tree-optimization/71947
20793 * tree-ssa-dom.c: Include tree-inline.h.
20794 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20795 equivalences if one is more expensive to compute than the other.
20796 * tree-ssa-scopedtables.h (class const_or_copies): Make
20797 record_const_or_copy_raw method private.
20798 (class avail_exprs_stack): New method simplify_binary_operation.
20799 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20800 avail_exprs_stack::simplify_binary_operation as needed.
20801 (avail_exprs_stack::simplify_binary_operation): New function.
20803 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20805 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20806 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20807 (DOT_SYMBOLS): Likewise.
20808 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20809 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20810 (RS6000_ABI_NAME): Likewise.
20811 (TARGET_CMODEL): Likewise.
20812 (TOC_SECTION_ASM_OP): Likewise.
20813 (SET_CMODEL): New macro.
20814 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20816 2017-08-22 Richard Biener <rguenther@suse.de>
20818 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20819 to allow for free-lang-data replacements similar to verify_type_variant.
20821 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20822 Alan Hayward <alan.hayward@arm.com>
20823 David Sherwood <david.sherwood@arm.com>
20825 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20826 VOIDmode for the LABEL_REF.
20828 2017-08-22 Richard Biener <rguenther@suse.de>
20830 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20832 2017-08-22 Richard Biener <rguenther@suse.de>
20834 * common.opt (feliminate-dwarf2-dups): Ignore.
20835 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20836 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20837 same_die_p_wrap, compute_section_prefix,
20838 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20839 (comdat_symbol_id, comdat_symbol_number): Likewise.
20840 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20842 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20843 (output_die): Mark unreachable path unreachable.
20844 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20845 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20846 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20847 (dwarf2out_early_finish): Likewise.
20849 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20851 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20853 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20856 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20857 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20858 (avr_attribute_table) <io, io_low, address>: Initialize
20859 .decl_required with true.
20861 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20863 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20864 undocumented debugging options.
20865 (-mvsx-scalar-double): Likewise.
20866 (-mallow-df-permute): Likewise.
20867 (-mvectorize-builtins): Likewise.
20868 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20869 (rs6000_builtin_vectorized_function): Likewise.
20870 (rs6000_builtin_md_vectorized_function): Likewise.
20871 (rs6000_opt_vars): Likewise.
20873 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20876 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20877 (*btrq_imm): Rename from *btrq.
20878 (*btcq_imm): Rename from *btcq.
20879 (btsc): New code attribute.
20880 (*<btsc><mode>): New insn pattern.
20881 (*btr<mode>): Ditto.
20882 (*<btsc><mode>_mask): New insn_and_split pattern.
20883 (*btr<mode>_mask): Ditto.
20885 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20886 Alan Hayward <alan.hayward@arm.com>
20887 David Sherwood <david.sherwood@arm.com>
20889 * function.c (pad_below): Simplify padding calculation.
20891 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20892 Alan Hayward <alan.hayward@arm.com>
20893 David Sherwood <david.sherwood@arm.com>
20895 * target.def (function_prologue): Remove frame size argument.
20896 (function_epilogue): Likewise.
20897 * doc/tm.texi: Regenerate.
20898 * final.c (final_start_function): Update call to function_prologue.
20899 (final_end_function): Update call to function_epilogue.
20900 (default_function_pro_epilogue): Remove frame size argument.
20901 * output.h (default_function_pro_epilogue): Likewise.
20902 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20903 (arm_output_function_prologue): Likewise.
20904 * config/frv/frv.c (frv_function_prologue): Likewise.
20905 (frv_function_epilogue): Likewise.
20906 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20907 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20908 (ia64_output_function_epilogue): Likewise.
20909 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20910 (m32r_output_function_epilogue): Likewise.
20911 * config/microblaze/microblaze.c (microblaze_function_prologue)
20912 (microblaze_function_epilogue): Likewise.
20913 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20914 (mips_output_function_epilogue): Likewise.
20915 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20916 (mmix_target_asm_function_epilogue): Likewise.
20917 * config/msp430/msp430.c (msp430_start_function): Likewise.
20918 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20919 (nds32_asm_function_epilogue): Likewise.
20920 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20921 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20922 (pa_output_function_epilogue): Likewise.
20923 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20924 (rs6000_output_function_epilogue): Likewise.
20925 * config/rl78/rl78.c (rl78_start_function): Likewise.
20926 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20927 (rs6000_output_function_epilogue): Likewise.
20928 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20929 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20930 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20931 (sparc_asm_function_epilogue): Likewise.
20933 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20935 * tree.h (type_has_mode_precision_p): New function.
20936 * convert.c (convert_to_integer_1): Use it.
20937 * expr.c (expand_expr_real_2): Likewise.
20938 (expand_expr_real_1): Likewise.
20939 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20940 * match.pd: Likewise.
20941 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20942 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20943 * tree-tailcall.c (process_assignment): Likewise.
20944 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20945 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20946 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20947 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20948 (vectorizable_assignment): Likewise.
20949 (vectorizable_shift): Likewise.
20950 (vectorizable_operation): Likewise.
20951 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20953 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20955 * match.pd: Add pow (C, x) simplification.
20957 2017-08-21 Richard Biener <rguenther@suse.de>
20959 PR tree-optimization/81900
20960 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20961 for blocks with abnormal predecessors.
20962 (compute_antic): Do not set visited flag prematurely.
20964 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20967 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20969 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20971 * stor-layout.h (vector_type_mode): Move to...
20972 * tree.h (vector_type_mode): ...here.
20973 * stor-layout.c (vector_type_mode): Move to...
20974 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20976 2017-08-21 Richard Biener <rguenther@suse.de>
20978 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20979 register_external_die hooks.
20980 (debug_false_tree_charstarstar_uhwistar): Declare.
20981 (debug_nothing_tree_charstar_uhwi): Likewise.
20982 * debug.c (do_nothing_debug_hooks): Adjust.
20983 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20984 (debug_nothing_tree_charstar_uhwi): Likewise.
20985 * dbxout.c (dbx_debug_hooks): Adjust.
20986 (xcoff_debug_hooks): Likewise.
20987 * sdbout.c (sdb_debug_hooks): Likewise.
20988 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20989 * dwarf2out.c (macinfo_label_base): New global.
20990 (dwarf2out_register_external_die): New function for the
20991 register_external_die hook.
20992 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20993 (dwarf2_debug_hooks): Use them.
20994 (dwarf2_lineno_debug_hooks): Adjust.
20995 (struct die_struct): Add with_offset flag.
20996 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20997 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20998 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20999 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
21000 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
21001 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
21002 defining section names for the early LTO debug variants.
21003 (reset_indirect_string): New helper.
21004 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
21005 (print_dw_val): Add support for offsetted symbol references.
21006 (get_ultimate_context): Split out from is_cxx.
21007 (is_cxx): Use get_ultimate_context.
21008 (is_fortran): Add decl overload.
21009 (compute_comp_unit_symbol): Split out worker from
21010 compute_section_prefix.
21011 (compute_section_prefix): Call compute_comp_unit_symbol and
21012 set comdat_type_p here.
21013 (output_die): Skip DIE symbol output for the LTO added one.
21014 Handle DIE symbol references with offset.
21015 (output_comp_unit): Guard section name mangling properly.
21016 For LTO debug sections emit a symbol at the section beginning
21017 which we use to refer to its DIEs.
21018 (add_abstract_origin_attribute): For DIEs registered via
21019 dwarf2out_register_external_die directly refer to the early
21020 DIE rather than indirectly through the shadow one we created.
21021 Remove obsolete call to dwarf2out_abstract_function for
21022 non-function/block origins.
21023 (gen_array_type_die): When generating early LTO debug do
21024 not emit DW_AT_string_length.
21025 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
21026 late when in LTO. As suggested place a gcc_unreachable for
21027 the DECL_ABSTRACT_P case.
21028 (gen_subprogram_die): Avoid another specification DIE
21029 for early built declarations/definitions for the late LTO case.
21030 (gen_variable_die): Add type references for late duplicated VLA dies
21032 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
21033 we have the abstract instance already.
21034 (process_scope_var): Adjust decl DIE contexts in LTO which
21035 first puts them in limbo.
21036 (gen_decl_die): Do not generate type DIEs late apart from
21037 types for VLAs or for decls we do not yet have a DIE. Do not
21038 call dwarf2out_abstract_function late.
21039 (dwarf2out_early_global_decl): Make sure to create DIEs
21040 for abstract instances of a decl first.
21041 (dwarf2out_late_global_decl): Adjust comment.
21042 (output_macinfo_op): With multiple macro sections use
21043 macinfo_label_base to distinguish labels.
21044 (output_macinfo): Likewise. Update macinfo_label_base.
21045 Pass in the line info label.
21046 (note_variable_value_in_expr): When generating LTO resolve
21047 all variable values here by generating DIEs as needed.
21048 (init_sections_and_labels): Add early LTO debug flag parameter
21049 and generate different sections and names if set. Add generation
21050 counter for the labels so we can have multiple of them.
21051 (reset_dies): Helper to allow DIEs to be output multiple times.
21052 (dwarf2out_finish): When outputting DIEs to the fat part of an
21053 LTO object first reset DIEs.
21054 (dwarf2out_early_finish): Output early DIEs when generating LTO.
21055 (modified_type_die): Check for decl_ultimate_origin being self
21057 (gen_type_die_with_usage): Likewise.
21058 (gen_typedef_die): Allow decl_ultimate_origin being self.
21059 (set_decl_abstract_flags): Remove.
21060 (set_block_abstract_flags): Likewise.
21061 (dwarf2out_abstract_function): Treat the early generated DIEs
21062 as the abstract copy and only add DW_AT_inline and
21063 DW_AT_artificial here and call set_decl_origin_self.
21064 If the DIE has an abstract origin don't do anything.
21065 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21066 if we have none yet (Go fails to build one, PR78628).
21067 (variably_modified_type_p): Prevent endless recursion for Ada
21068 cyclic pointer types.
21069 * lto-streamer-in.c: Include debug.h.
21070 (dref_queue): New global.
21071 (lto_read_tree_1): Stream in DIE references.
21072 (lto_input_tree): Register DIE references.
21073 (input_function): Stream DECL_DEBUG_ARGS.
21074 * lto-streamer-out.c: Include debug.h.
21075 (lto_write_tree_1): Output DIE references.
21076 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21077 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21078 (output_function): Stream DECL_DEBUG_ARGS.
21079 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21080 Stream DECL_ABSTRACT_ORIGIN.
21081 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21082 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21083 DECL_CONTEXT for file-scope decls.
21084 * lto-streamer.h (struct dref_entry): Declare.
21085 (dref_queue): Likewise.
21086 * cfgexpand.c (pass_expand::execute): Do not call the
21087 outlining_inline_function hook here.
21088 * lto-wrapper.c (debug_obj): New global.
21089 (tool_cleanup): Unlink it if required.
21090 (debug_objcopy): New function.
21091 (run_gcc): Handle early debug sections in the IL files by
21092 extracting them to separate files, partially linkin them and
21093 feeding the result back as result to the linker.
21094 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21095 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21096 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21097 sections into a separate segment.
21098 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21100 (darwin_asm_dwarf_section): Likewise.
21101 (darwin_asm_output_dwarf_offset): Likewise.
21102 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
21104 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21105 Alan Hayward <alan.hayward@arm.com>
21106 David Sherwood <david.sherwood@arm.com>
21108 * read-md.h (md_reader::record_potential_iterator_use): Replace
21109 pointer argument with an rtx and an index.
21110 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21111 (apply_mode_iterator): Likewise.
21112 (apply_code_iterator): Likewise.
21113 (apply_int_iterator): Likewise.
21114 (apply_subst_iterator): Likewise.
21115 (record_iterator_use): Likewise.
21116 (record_attribute_use): Likewise.
21117 (md_reader::record_potential_iterator_use): Likewise. Update calls
21118 to record_iterator_use and apply_iterator.
21119 (iterator_use): Replace ptr with x and index.
21120 (attribute_use): Likewise.
21121 (apply_attribute_uses): Update calls to apply_iterator.
21122 (apply_iterators): Likewise. Update initialization of iterator_use.
21123 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21124 and record_potential_iterator_use.
21125 (rtx_reader::read_rtx_operand): Likewise.
21127 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21128 Alan Hayward <alan.hayward@arm.com>
21129 David Sherwood <david.sherwood@arm.com>
21131 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21134 2017-08-21 Richard Biener <rguenther@suse.de>
21136 PR middle-end/81884
21137 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21138 at struct end conservatively when comparing common bases.
21140 2017-08-21 Richard Biener <rguenther@suse.de>
21142 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21143 (mem_ref_in_stmt): Remove.
21144 (determine_max_movement): Use ref index to get at the reference.
21145 (invariantness_dom_walker::before_dom_children): Deal with
21146 lim data already initialized.
21147 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21149 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21151 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21152 (TARGET_ISA_ROUND): Ditto.
21153 (TARGET_ROUND): Ditto.
21154 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21155 * config/i386/i386.md: Ditto.
21156 * config/i386/sse.md: Ditto.
21157 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21158 with OPTION_MASK_ISA_SSE4_1.
21160 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21163 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21164 __builtin_ia32_lzcnt_u16.
21166 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21169 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21170 (rs6000_set_current_function): Rewrite function to use it.
21172 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21175 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21176 and TYPE_WARN_IF_NOT_ALIGN.
21177 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21178 (handle_warn_if_not_align): New.
21179 (place_union_field): Call handle_warn_if_not_align.
21180 (place_field): Call handle_warn_if_not_align.
21181 Copy TYPE_WARN_IF_NOT_ALIGN.
21182 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21183 (layout_type): Likewise.
21184 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21186 (tree_decl_common): Add warn_if_not_align.
21187 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21188 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21189 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21190 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21191 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21192 * doc/extend.texi: Document warn_if_not_aligned attribute.
21193 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21195 2017-08-17 Martin Liska <mliska@suse.cz>
21198 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21199 (get_data_dependence): Use it as pointer type.
21200 (distribute_loop): Likewise.
21202 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21204 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21205 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21206 (p8_vmrgow_<mode>_direct): New define_insn.
21207 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21208 handle endianness for vmrgew and vmrgow permute patterns.
21210 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21212 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21213 * config/rs6000/rs6000-cpus.def: Remove comment.
21214 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21215 (POWERPC_MASKS): Likewise.
21216 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21217 use of TARGET_VSX_TIMODE.
21218 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21219 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21220 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21221 (rs6000_option_override_internal): Remove dead code.
21222 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21223 (rs6000_legitimize_reload_address): Likewise.
21224 (rs6000_legitimate_address_p): Likewise.
21225 (rs6000_opt_masks): Delete "vsx-timode".
21226 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21227 from function comment.
21228 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21229 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21230 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21232 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21233 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21234 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21235 (VSX_TI): Likewise.
21237 (define_peephole2): Likewise.
21239 2017-08-17 Martin Sebor <msebor@redhat.com>
21242 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21243 past the end of an array.
21244 (test_pp_format): Add test cases.
21246 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21248 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21249 missing ECF_NOTHROW flags.
21251 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21254 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21255 operands residing in integer registers.
21256 (*vsx_le_perm_load_<mode>): Likewise.
21257 (*vsx_le_perm_store_<mode>): Likewise.
21258 (define_peephole2): Add peepholes to optimize the above.
21260 2017-08-17 Marek Polacek <polacek@redhat.com>
21262 PR middle-end/81814
21263 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21264 to mimic what shorten_compare did. Change the return type to bool.
21265 (fold_cond_expr_with_comparison): Update call to
21266 operand_equal_for_comparison_p.
21267 (fold_ternary_loc): Likewise.
21269 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21271 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21273 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21275 2017-08-17 Richard Biener <rguenther@suse.de>
21277 * tree-ssa-structalias.c (solve_graph): When propagating
21278 to successors update the graphs succ edges and avoid duplicate work.
21280 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21283 * config/i386/i386.c (ix86_option_override_internal): Save target
21284 specific options after ix86_stack_protector_guard_reg was changed.
21286 2017-08-17 Richard Biener <rguenther@suse.de>
21288 PR tree-optimization/81827
21289 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21290 (new_var_info): Initialize it conservatively.
21291 (get_call_vi): Mark register vars.
21292 (new_scalar_tmp_constraint_exp): Likewise.
21293 (handle_rhs_call): Likewise.
21294 (handle_const_call): Likewise.
21295 (create_function_info_for): Likewise.
21296 (solve_constraints): Sort varinfos to separate register from
21297 non-register vars to pack points-to solution bitmaps during
21300 2017-08-17 Marek Polacek <polacek@redhat.com>
21302 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21304 2017-08-17 Richard Biener <rguenther@suse.de>
21306 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21307 to true when overflow is undefined and we saturated the result.
21309 2017-08-17 Alan Modra <amodra@gmail.com>
21312 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21313 Don't use store multiple if only one reg needs saving.
21314 (interesting_frame_related_regno): New function.
21315 (rs6000_frame_related): Don't emit frame info for regs that
21317 (rs6000_emit_epilogue): Likewise.
21319 2017-08-16 Nathan Sidwell <nathan@acm.org>
21321 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21322 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21323 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21324 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21325 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21326 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21327 (verify_type): Adjust for TYPE_BINFO move.
21328 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21329 process TYPE_BINFO directly.
21330 (hash_tree): Likewise.
21331 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21333 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21336 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21338 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21340 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21343 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21344 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21345 Add ix86_binary_operator_ok to insn constraint.
21346 (*iordi_1_bts): Ditto.
21347 (*xordi_1_btc): Ditto.
21348 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21349 Update corresponding peephole2 pattern.
21353 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21355 PR tree-optimization/81832
21356 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21357 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21359 2017-08-16 Marek Polacek <polacek@redhat.com>
21362 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21363 regarding size_zero_node.
21365 2017-08-16 Martin Liska <mliska@suse.cz>
21368 * config.gcc: Respect previously set extra_objs in case
21371 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21373 PR tree-optimization/81835
21374 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21375 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21376 not depend on the phi.
21378 2017-08-16 Alan Modra <amodra@gmail.com>
21380 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21383 2017-08-16 Alan Modra <amodra@gmail.com>
21385 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21386 (save_reg_p): ..into this. Update all callers.
21387 (first_reg_to_save): Simplify.
21389 2017-08-16 Alan Modra <amodra@gmail.com>
21391 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21394 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21398 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21399 constant count if that count is less than 32.
21401 2017-08-15 Nathan Sidwell <nathan@acm.org>
21403 * gcc.c (execute): Emit friendlier message if inferior is killed
21404 by an external cause.
21406 2017-08-15 Richard Biener <rguenther@suse.de>
21408 PR tree-optimization/81790
21409 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21410 CONSTRUCTORs from simplifying and VN.
21412 2017-08-14 Martin Sebor <msebor@redhat.com>
21414 * builtin-attrs.def: Add comments.
21416 2017-08-14 Martin Sebor <msebor@redhat.com>
21419 * doc/extend.texi (attribute nonstring): Document new attribute.
21421 2017-08-14 Martin Sebor <msebor@redhat.com>
21424 * tree-diagnostic.c (default_tree_printer): Handle %G.
21425 * gimple-pretty-print.h (percent_G_format): Declare new function.
21426 * gimple-pretty-print.c (percent_G_format): Define.
21427 * tree-pretty-print.c (percent_K_format): Add argument.
21429 2017-08-14 Martin Sebor <msebor@redhat.com>
21431 PR translation/79998
21432 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21433 Remove a stray space.
21435 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21438 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21439 (*iordi_1_bts): Ditto.
21440 (*xordi_1_btc): Ditto.
21442 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21445 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21446 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21448 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21449 quoted strings, and make more translator-friendly.
21450 (darwin_rs6000_override_options): Likewise.
21451 (rs6000_option_override_internal): Likewise.
21452 (rs6000_return_in_memory): Fix overlong line.
21453 (init_cmulative_args): Use quoted strings, and make more
21454 translator-friendly.
21455 (rs6000_pass_by_reference): Fix overlong line.
21456 (def_builtin): Use quoted strings.
21457 (altivec_expand_predicate_builtin): Use quoted strings, and make
21458 more translator-friendly.
21459 (htm_expand_builtin): Use quoted strings.
21460 (cpu_expand_builtin): Use quoted strings, and make more
21461 translator-friendly.
21462 (altivec_expand_builtin): Likewise.
21463 (paired_expand_predicate_builtin): Likewise.
21464 (rs6000_invalid_builtin): Likewise.
21465 (builtin_function_type): Use quoted strings.
21466 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21467 more translator-friendly.
21468 (rs6000_trampoline_init): Likewise.
21469 (rs6000_handle_altivec_attribute): Likewise.
21470 (rs6000_inner_target_options): Use quoted strings.
21471 (rs6000_disable_incompatible_switches): Likewise.
21472 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21473 strings, and make more translator-friendly.
21474 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21476 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21478 PR tree-optimization/81799
21479 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21480 cond_expr to simple gimple operand.
21482 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21484 PR middle-end/46932
21485 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21487 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21491 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21492 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21493 TARGET_GASISR_PROLOGUES.
21494 * config/avr/avr.c (avr_option_override): Same.
21495 (avr_pass_pre_proep::execute): Same.
21497 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21500 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21501 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21503 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21505 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21506 define_insn_and_split. Split to a memory load from 0 in
21507 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21508 using PTR mode iterator.
21509 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21510 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21511 (*add_tp_<mode>): Redefine as define_insn_and_split.
21512 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21513 address space. Merge with *add_tp_x32 using PTR mode iterator.
21514 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21515 Split to an add with a memory load from 0 in
21516 DEFAULT_TLS_SEG_REG address space.
21518 2017-08-12 Andrew Pinski <apinski@cavium.com>
21520 * config/aarch64/aarch64-option-extensions.def (rdma):
21521 Fix feature string to what Linux prints out in /proc/cpuinfo.
21523 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21526 * dwarf2out.c (modified_type_die): For C typedef types that have
21527 an ultimate origin, process the ultimate origin instead of the
21529 (gen_typedef_die): Assert that input DECLs have no ultimate
21531 (gen_type_die_with_usage): For typedef variants that have an
21532 ultimate origin, just call gen_decl_die on the original DECL.
21533 (process_scope_var): Avoid creating DIEs for local typedefs and
21534 concrete static variables.
21536 2017-08-12 Alan Modra <amodra@gmail.com>
21540 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21541 match gnu-user.h startfile.
21542 (ENDFILE_LINUX_SPEC): Similarly.
21544 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21547 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21549 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21551 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21552 * config/aarch64/aarch64.md (mov<mode>): Change.
21553 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21554 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21555 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21557 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21559 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21560 for storage order barriers.
21562 2017-08-11 Martin Liska <mliska@suse.cz>
21565 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21566 variables of void type.
21568 2017-08-11 Martin Liska <mliska@suse.cz>
21570 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21571 TARGET_SUPPORTS_ALIASES.
21572 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21573 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21574 (optimize_weakref): Likewise.
21575 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21576 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21577 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21579 2017-08-11 Martin Liska <mliska@suse.cz>
21582 * config/i386/i386.c (make_resolver_func): Do complete
21583 refactoring of the function.
21585 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21588 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21589 * config/i386/i386.c (ix86_stack_protect_guard): Use
21590 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21591 * doc/invoke.texi (x86 Options): Document
21592 -mstack-protector-guard-symbol= option.
21594 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21596 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21597 * config/i386/i386.c (ix86_split_stack_guard): New function.
21598 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21599 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21600 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21601 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21602 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21603 (split_stack_space_check): Call ix86_split_stack_guard.
21605 2017-08-10 Martin Sebor <msebor@redhat.com>
21607 * print-tree.c (print_node): Print location using the established
21609 Replace spaces with colons.
21610 (debug_raw, debug): Ditto.
21612 2017-08-10 Martin Sebor <msebor@redhat.com>
21615 * pretty-print.c (pp_format): Correct the handling of %s precision.
21617 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21620 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21622 (ix86_finalize_stack_frame_flags): This. Also clear
21623 frame_pointer_needed if -fno-omit-frame-pointer is used without
21625 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21626 with ix86_finalize_stack_frame_flags.
21627 (ix86_expand_epilogue): Likewise.
21628 (ix86_expand_split_stack_prologue): Likewise.
21629 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21631 2017-08-10 Martin Liska <mliska@suse.cz>
21634 * c-attribs.c (handle_target_attribute):
21635 Report warning for an empty string argument of target attribute.
21637 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21640 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21642 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21643 or DECL_NONLOCAL labels.
21644 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21645 or DECL_NONLOCAL labels here.
21647 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21649 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21650 to indicate when early gimple folding has been disabled.
21651 (rs6000_gimple_fold_builtin): Add debug content.
21652 (rs6000_invalid_builtin): Fix whitespace.
21653 (rs6000_expand_builtin): Fix whitespace.
21654 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21656 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21659 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21660 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21662 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21664 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21665 (qdf24xx): Likewise.
21666 * config/aarch64/aarch64.md: Include falkor.md.
21667 * config/aarch64/falkor.md: New.
21669 2017-08-09 Marek Polacek <polacek@redhat.com>
21672 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21673 * diagnostic.c (emit_diagnostic): Add a comment.
21674 (emit_diagnostic_valist): New function.
21676 2017-08-09 Marek Polacek <polacek@redhat.com>
21679 * input.c (make_location): New overload.
21680 * input.h (make_location): Declare.
21682 2017-08-08 Alan Modra <amodra@gmail.com>
21683 H.J. Lu <hongjiu.lu@intel.com>
21686 * gcc.c (NO_PIE_SPEC): Delete.
21687 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21689 (LINK_PIE_SPEC): ..to here.
21690 (LINK_COMMAND_SPEC): Support -no-pie.
21691 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21692 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21694 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21695 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21696 (ENDFILE_CRTEND_SPEC): Similarly.
21698 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21701 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21702 (mstack-protector-guard-offset=): Ditto.
21703 * config/i386/i386.c (ix86_option_override): Handle
21704 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21706 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21707 ix86_stack_protect_guard_offset variables.
21708 (TARGET_STACK_PROTECT_GUARD): Always define.
21709 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21710 and -mstack-protector-guard-offset= options.
21712 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21714 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21715 boundary case for the last candidate.
21717 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21719 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21720 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21722 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21724 PR middle-end/19706
21725 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21726 * config/aarch64/aarch64-builtins.c
21727 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21728 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21729 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21731 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21732 Andrew Pinski <pinskia@gmail.com>
21734 PR middle-end/19706
21735 * internal-fn.def (XORSIGN): New.
21736 * optabs.def (xorsign_optab): New.
21737 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21738 (convert_expand_mult_copysign): New.
21739 (pass_optimize_widening_mul::execute): Call
21740 convert_expand_mult_copysign.
21742 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21744 PR tree-optimization/81354
21745 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21746 Insert on edges rather than explicitly creating landing pads.
21747 (analyze_candidates_and_replace): Commit edge inserts.
21749 2017-08-08 Richard Biener <rguenther@suse.de>
21751 PR middle-end/81719
21752 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21753 (expand_simple_operations): Also look through ADDR_EXPRs with
21754 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21756 2017-08-08 Richard Biener <rguenther@suse.de>
21758 PR tree-optimization/81723
21759 * tree-vect-slp.c (struct bst_traits): New hash traits.
21760 (bst_fail): New global.
21761 (vect_build_slp_tree_2): New worker, split out from ...
21762 (vect_build_slp_tree): ... this now wrapping it with using
21763 bst_fail set to cache SLP tree build fails. Properly handle
21765 (vect_analyze_slp_instance): Allocate and free bst_fail.
21767 2017-08-08 Martin Liska <mliska@suse.cz>
21770 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21771 LABEL_DECLs that can be from a different function.
21773 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21775 PR tree-optimization/81744
21776 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21777 loop's number of iterations.
21779 2017-08-08 Martin Liska <mliska@suse.cz>
21781 * asan.c: Include header files.
21782 * attribs.c (build_decl_attribute_variant): New function moved
21784 (build_type_attribute_qual_variant): Likewise.
21785 (cmp_attrib_identifiers): Likewise.
21786 (simple_cst_list_equal): Likewise.
21787 (omp_declare_simd_clauses_equal): Likewise.
21788 (attribute_value_equal): Likewise.
21789 (comp_type_attributes): Likewise.
21790 (build_type_attribute_variant): Likewise.
21791 (lookup_ident_attribute): Likewise.
21792 (remove_attribute): Likewise.
21793 (merge_attributes): Likewise.
21794 (merge_type_attributes): Likewise.
21795 (merge_decl_attributes): Likewise.
21796 (merge_dllimport_decl_attributes): Likewise.
21797 (handle_dll_attribute): Likewise.
21798 (attribute_list_equal): Likewise.
21799 (attribute_list_contained): Likewise.
21800 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21801 (lookup_attribute_by_prefix): Likewise.
21802 * bb-reorder.c: Include header files.
21803 * builtins.c: Likewise.
21804 * calls.c: Likewise.
21805 * cfgexpand.c: Likewise.
21806 * cgraph.c: Likewise.
21807 * cgraphunit.c: Likewise.
21808 * convert.c: Likewise.
21809 * dwarf2out.c: Likewise.
21810 * final.c: Likewise.
21811 * fold-const.c: Likewise.
21812 * function.c: Likewise.
21813 * gimple-expr.c: Likewise.
21814 * gimple-fold.c: Likewise.
21815 * gimple-pretty-print.c: Likewise.
21816 * gimple.c: Likewise.
21817 * gimplify.c: Likewise.
21818 * hsa-common.c: Likewise.
21819 * hsa-gen.c: Likewise.
21820 * internal-fn.c: Likewise.
21821 * ipa-chkp.c: Likewise.
21822 * ipa-cp.c: Likewise.
21823 * ipa-devirt.c: Likewise.
21824 * ipa-fnsummary.c: Likewise.
21825 * ipa-inline.c: Likewise.
21826 * ipa-visibility.c: Likewise.
21828 * lto-cgraph.c: Likewise.
21829 * omp-expand.c: Likewise.
21830 * omp-general.c: Likewise.
21831 * omp-low.c: Likewise.
21832 * omp-offload.c: Likewise.
21833 * omp-simd-clone.c: Likewise.
21834 * opts-global.c: Likewise.
21835 * passes.c: Likewise.
21836 * predict.c: Likewise.
21837 * sancov.c: Likewise.
21838 * sanopt.c: Likewise.
21839 * symtab.c: Likewise.
21840 * toplev.c: Likewise.
21841 * trans-mem.c: Likewise.
21842 * tree-chkp.c: Likewise.
21843 * tree-eh.c: Likewise.
21844 * tree-into-ssa.c: Likewise.
21845 * tree-object-size.c: Likewise.
21846 * tree-parloops.c: Likewise.
21847 * tree-profile.c: Likewise.
21848 * tree-ssa-ccp.c: Likewise.
21849 * tree-ssa-live.c: Likewise.
21850 * tree-ssa-loop.c: Likewise.
21851 * tree-ssa-sccvn.c: Likewise.
21852 * tree-ssa-structalias.c: Likewise.
21853 * tree-ssa.c: Likewise.
21854 * tree-streamer-in.c: Likewise.
21855 * tree-vectorizer.c: Likewise.
21856 * tree-vrp.c: Likewise.
21857 * tsan.c: Likewise.
21858 * ubsan.c: Likewise.
21859 * varasm.c: Likewise.
21860 * varpool.c: Likewise.
21861 * tree.c: Remove functions moved to attribs.[ch].
21862 * tree.h: Likewise.
21863 * config/aarch64/aarch64.c: Add attrs.h header file.
21864 * config/alpha/alpha.c: Likewise.
21865 * config/arc/arc.c: Likewise.
21866 * config/arm/arm.c: Likewise.
21867 * config/avr/avr.c: Likewise.
21868 * config/bfin/bfin.c: Likewise.
21869 * config/c6x/c6x.c: Likewise.
21870 * config/cr16/cr16.c: Likewise.
21871 * config/cris/cris.c: Likewise.
21872 * config/darwin.c: Likewise.
21873 * config/epiphany/epiphany.c: Likewise.
21874 * config/fr30/fr30.c: Likewise.
21875 * config/frv/frv.c: Likewise.
21876 * config/ft32/ft32.c: Likewise.
21877 * config/h8300/h8300.c: Likewise.
21878 * config/i386/winnt.c: Likewise.
21879 * config/ia64/ia64.c: Likewise.
21880 * config/iq2000/iq2000.c: Likewise.
21881 * config/lm32/lm32.c: Likewise.
21882 * config/m32c/m32c.c: Likewise.
21883 * config/m32r/m32r.c: Likewise.
21884 * config/m68k/m68k.c: Likewise.
21885 * config/mcore/mcore.c: Likewise.
21886 * config/microblaze/microblaze.c: Likewise.
21887 * config/mips/mips.c: Likewise.
21888 * config/mmix/mmix.c: Likewise.
21889 * config/mn10300/mn10300.c: Likewise.
21890 * config/moxie/moxie.c: Likewise.
21891 * config/msp430/msp430.c: Likewise.
21892 * config/nds32/nds32-isr.c: Likewise.
21893 * config/nds32/nds32.c: Likewise.
21894 * config/nios2/nios2.c: Likewise.
21895 * config/nvptx/nvptx.c: Likewise.
21896 * config/pa/pa.c: Likewise.
21897 * config/pdp11/pdp11.c: Likewise.
21898 * config/powerpcspe/powerpcspe.c: Likewise.
21899 * config/riscv/riscv.c: Likewise.
21900 * config/rl78/rl78.c: Likewise.
21901 * config/rx/rx.c: Likewise.
21902 * config/s390/s390.c: Likewise.
21903 * config/sh/sh.c: Likewise.
21904 * config/sol2.c: Likewise.
21905 * config/sparc/sparc.c: Likewise.
21906 * config/spu/spu.c: Likewise.
21907 * config/stormy16/stormy16.c: Likewise.
21908 * config/tilegx/tilegx.c: Likewise.
21909 * config/tilepro/tilepro.c: Likewise.
21910 * config/v850/v850.c: Likewise.
21911 * config/vax/vax.c: Likewise.
21912 * config/visium/visium.c: Likewise.
21913 * config/xtensa/xtensa.c: Likewise.
21915 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21918 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21919 constraints since the -mupper-regs-* switches have been
21921 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21922 into a vector from a double word element that was extracted from
21923 another vector, and eliminate extra XXPERMDI instructions.
21924 (vsx_concat_<mode>_2): Likewise.
21925 (vsx_concat_<mode>_3): Likewise.
21926 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21927 concat to allow optimizing inserts from previous extracts.
21929 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21931 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21932 memory reference to a SSP offset in TLS address space.
21933 (ix86_print_operand) <case '@'>: Remove.
21934 (ix86_print_operand_punct_valid_p): Remove '@' code.
21935 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21936 UNSPEC_SP_TLS_TEST.
21937 (stack_tls_protect_set_<mode>): Remove.
21938 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21939 (stack_tls_protect_test_<mode>): Remove.
21940 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21942 2017-08-07 Olivier Hainque <hainque@adacore.com>
21945 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21947 2017-08-07 Douglas Rupp <rupp@adacore.com>
21949 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21950 variable was referenced as multidir in command.
21952 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21955 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21958 2017-08-07 Martin Liska <mliska@suse.cz>
21960 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21961 * config/rl78/rl78.c: Add include of attribs.h.
21962 * config/sh/sh.c: Likewise.
21963 * config/v850/v850.c: Likewise.
21965 2017-08-07 Tom de Vries <tom@codesourcery.com>
21967 PR middle-end/78266
21968 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21970 2017-08-07 Martin Liska <mliska@suse.cz>
21972 * config/mips/mips.c: Include attribs.h.
21974 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21977 * doc/invoke.texi: Document change in behvaior for -Ofast for
21980 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21982 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21984 (aarch64_pop_regs): Likewise.
21985 (aarch64_gen_load_pair): Likewise.
21986 (aarch64_save_callee_saves): Likewise.
21987 (aarch64_restore_callee_saves): Likewise.
21989 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21991 * config/i386/i386.c: Revert the last change.
21993 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21996 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21998 (ix86_finalize_stack_frame_flags): This. Also clear
21999 frame_pointer_needed if -fno-omit-frame-pointer is used without
22001 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
22002 with ix86_finalize_stack_frame_flags.
22003 (ix86_expand_epilogue): Likewise.
22004 (ix86_expand_split_stack_prologue): Likewise.
22006 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22009 * config/i386/i386.c (get_builtin_code_for_version): Set priority
22010 to P_AES for Westmere.
22012 2017-08-07 Jonathan Yong <10walls@gmail.com>
22014 * config/i386/mingw.opt (fset-stack-executable): Removed.
22015 * config/i386/cygming.opt (fset-stack-executable): Moved
22017 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
22019 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
22021 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
22023 2017-08-07 Marek Polacek <polacek@redhat.com>
22025 PR middle-end/81737
22026 * fold-const.c (fold_indirect_ref_1): Check type_domain.
22028 2017-08-07 Martin Liska <mliska@suse.cz>
22030 * attribs.h (canonicalize_attr_name): New function.
22031 (cmp_attribs): Move from c-format.c and adjusted.
22032 (is_attribute_p): Moved from tree.h.
22033 * tree-inline.c: Add new includes.
22034 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
22035 (private_is_attribute_p): Remove.
22036 (private_lookup_attribute): Likewise.
22037 (private_lookup_attribute_by_prefix): Simplify.
22038 (remove_attribute): Use is_attribute_p.
22039 * tree.h: Remove removed declarations.
22041 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22043 PR middle-end/81698
22044 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
22045 instead of computing it in the function. Formatting fix.
22046 (expand_case): Don't rely on default_edge being the first edge,
22047 clear it if removing it, pass default_edge to
22048 emit_case_dispatch_table.
22049 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
22052 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
22054 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
22055 insn in the function, emit NOP after the insn.
22057 2017-08-06 Tom de Vries <tom@codesourcery.com>
22059 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22062 2017-08-06 Tom de Vries <tom@codesourcery.com>
22064 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22067 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22069 PR tree-optimization/57371
22070 * match.pd: New pattern.
22072 2017-08-04 Marek Polacek <polacek@redhat.com>
22074 PR middle-end/81695
22075 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22076 perform the computation in offset_int.
22078 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22080 PR tree-optimization/81136
22081 * tree-vectorizer.h: Include tree-hash-traits.h.
22082 (vec_base_alignments): New typedef.
22083 (vec_info): Add a base_alignments field.
22084 (vect_record_base_alignments): Declare.
22085 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22087 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22088 (create_data_ref): Add an is_conditional_in_stmt argument.
22089 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22090 the is_conditional_in_stmt field.
22091 (data_ref_loc): Add an is_conditional_in_stmt field.
22092 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22093 (find_data_references_in_stmt): Update call to create_data_ref.
22094 (graphite_find_data_references_in_stmt): Likewise.
22095 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22096 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22097 (vect_record_base_alignment): New function.
22098 (vect_record_base_alignments): Likewise.
22099 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22100 for nested statements even if we fail to compute a misalignment.
22101 Use pooled base alignments for unconditional references.
22102 (vect_find_same_alignment_drs): Compare base addresses instead
22104 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22105 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22107 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22109 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22110 Add an explicit name for the enum. Use auto_vec for slp_instances
22111 and grouped_stores.
22112 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22114 (_bb_vec_info): Add a constructor and destructor.
22115 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22116 (destroy_loop_vec_info): Delete.
22117 (vect_destroy_datarefs): Likewise.
22118 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22119 (vec_info::vec_info): New function.
22120 (vec_info::~vec_info): Likewise.
22121 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22122 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22123 destroy_loop_vec_info.
22124 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22125 (_loop_vec_info::_loop_vec_info): ...this.
22126 (destroy_loop_vec_info): Replace with...
22127 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22128 the stmt_vec_infos. Leave handling of vec_info information to its
22129 destructor. Remove explicit vector releases.
22130 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22131 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22132 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22133 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22134 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22135 (destroy_bb_vec_info): Replace with...
22136 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22137 information to its destructor.
22138 (vect_slp_analyze_bb_1): Use new and delete instead of
22139 new_bb_vec_info and destroy_bb_vec_info.
22140 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22143 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22145 * tree-data-ref.h (subscript): Add access_fn field.
22146 (data_dependence_relation): Add could_be_independent_p.
22147 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22148 (same_access_functions): Move to tree-data-ref.c.
22149 * tree-data-ref.c (ref_contains_union_access_p): New function.
22150 (access_fn_component_p): Likewise.
22151 (access_fn_components_comparable_p): Likewise.
22152 (dr_analyze_indices): Add a reference to access_fn_component_p.
22153 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22155 (constant_access_functions): Likewise.
22156 (add_other_self_distances): Likewise.
22157 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22158 (initialize_data_dependence_relation): Use XCNEW and remove
22159 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22160 of access functions that have the same type. Allow the
22161 subsequence to end with different bases in some circumstances.
22162 Record the chosen access functions in SUB_ACCESS_FN.
22163 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22164 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22165 (subscript_dependence_tester_1): Likewise dra and drb.
22166 (build_classic_dist_vector): Update calls accordingly.
22167 (subscript_dependence_tester): Likewise.
22168 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22169 DDR_COULD_BE_INDEPENDENT_P.
22170 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22171 comp_alias_ddrs instead of may_alias_ddrs.
22172 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22174 (vect_analyze_data_ref_dependence): Use it if
22175 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22176 distance vectors if that fails.
22177 (dependence_distance_ge_vf): New function.
22178 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22179 LOOP_VINFO_MAY_ALIAS_DDRS.
22181 2017-08-04 Richard Biener <rguenther@suse.de>
22183 PR middle-end/81705
22184 * fold-const.c (fold_binary_loc): Properly restrict
22185 minus_var0 && minus_var1 case when associating undefined overflow
22188 2017-08-04 Tom de Vries <tom@codesourcery.com>
22190 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22192 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22194 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22195 Don't start diagnostic messages with a capital letter.
22196 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22198 (rs6000_invalid_builtin): Likewise.
22199 (rs6000_trampoline_init): Likewise.
22201 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22204 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22205 after setting changeable df flags.
22207 2017-08-03 Richard Biener <rguenther@suse.de>
22209 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22210 up if the use is in USE - X.
22212 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22214 * toplev.c (dumpfile.h): New include.
22215 (internal_error_reentered): New static function. Use it...
22216 (internal_error_function): ...here to handle reentered internal_error.
22218 2017-08-03 Richard Biener <rguenther@suse.de>
22220 PR middle-end/81148
22221 * fold-const.c (split_tree): Add minus_var and minus_con
22222 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22223 here but always use minus_*.
22224 (associate_trees): Assert we never associate with MINUS_EXPR
22225 and NULL first operand. Do not recurse for PLUS_EXPR operands
22226 when associating as MINUS_EXPR either.
22227 (fold_binary_loc): Track minus_var and minus_con.
22229 2017-08-03 Tom de Vries <tom@codesourcery.com>
22232 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22233 ACCEL_COMPILER, apply finish_options on
22234 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22236 2017-08-03 Tom de Vries <tom@codesourcery.com>
22239 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22240 function_entry_patch_area_size > 0.
22242 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22245 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22246 instead of 10??LU, perform unit multiplication in wide_int,
22247 don't change alloc_object_size_limit if the limit is larger
22250 PR tree-optimization/81655
22251 PR tree-optimization/81588
22252 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22253 the case when ranges[i].low and high are 1 for unsigned type with
22256 PR middle-end/81052
22257 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22258 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22260 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22262 * tree-vrp.h: Add include guard.
22264 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22267 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22268 (ud2): New insn pattern.
22269 * config/i386/i386.c (ix86_expand_epilogue):
22270 For naked functions, generate ud2 instead of trap insn.
22272 2017-08-02 Marek Polacek <polacek@redhat.com>
22275 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22277 2017-08-02 Tom de Vries <tom@codesourcery.com>
22278 Cesar Philippidis <cesar@codesourcery.com>
22280 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22281 Add missing edge probabilities.
22283 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22285 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22286 Correct endianness.
22288 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22290 PR middle-end/79499
22291 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22292 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22293 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22294 prologue_seq sequences - if any.
22296 2017-08-02 Richard Biener <rguenther@suse.de>
22298 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22299 via vectors if supported, integer extracts via punning if supported
22300 or otherwise vector extracts.
22302 2017-08-02 Richard Biener <rguenther@suse.de>
22304 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22306 (bitmap_insert_into_set): ... this.
22308 2017-08-02 Richard Biener <rguenther@suse.de>
22310 PR tree-optimization/81633
22312 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22314 PR tree-optimization/71752
22315 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22317 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22319 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22320 (machine_function::call_ms2sysv_pad_out): Remove field.
22321 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22322 (ix86_compute_frame_layout): Likewise.
22324 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22327 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22328 attribute with interrupt attribute.
22330 2017-08-01 Andrew Pinski <apinski@cavium.com>
22332 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22333 BIT_INSERT_EXPR's operand 1
22334 to see if the types precision matches.
22336 2017-08-01 Martin Liska <mliska@suse.cz>
22338 PR middle-end/70140
22339 * builtins.c (expand_builtin_memcpy_args): Remove.
22340 (expand_builtin_memcpy): Call newly added function
22341 expand_builtin_memory_copy_args.
22342 (expand_builtin_memcpy_with_bounds): Likewise.
22343 (expand_builtin_mempcpy): Remove last argument.
22344 (expand_builtin_mempcpy_with_bounds): Likewise.
22345 (expand_builtin_memory_copy_args): New function created from
22346 expand_builtin_mempcpy_args with small modifications.
22347 (expand_builtin_mempcpy_args): Remove.
22348 (expand_builtin_stpcpy): Remove unused argument.
22349 (expand_builtin): Likewise.
22350 (expand_builtin_with_bounds): Likewise.
22352 2017-08-01 Martin Liska <mliska@suse.cz>
22355 Make mempcpy more optimal (PR middle-end/70140).
22357 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22360 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22361 __builtin_vec_cmpne verify both arguments are compatible vectors
22362 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22363 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22364 move computation of aligned to after checking the argument types.
22368 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22369 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22372 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22373 Doug Rupp <rupp@adacore.com>
22374 Olivier Hainque <hainque@adacore.com>
22376 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22377 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22379 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22380 for TARGET_OS_CPP_BUILTIN.
22381 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22382 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22384 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22385 passing required abi options to the assembler for EABI configurations.
22386 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22387 of .text.hot and .text.unlikely sections for kernel modules when
22388 using ARM style exceptions.
22389 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22390 options. Add EXTRA_CC1_SPEC.
22391 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22393 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22395 (ARM_TARGET2_DWARF_FORMAT): Define.
22396 * config/arm/t-vxworks: Adjust multilib control to removal of the
22397 Diab command line options.
22399 2017-08-01 Martin Liska <mliska@suse.cz>
22401 PR gcov-profile/81561
22402 * gcov.c (unblock): Make unblocking safe as we need to preserve
22403 index correspondence of blocks and block_lists.
22405 2017-08-01 Richard Biener <rguenther@suse.de>
22407 PR tree-optimization/81181
22408 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22409 (compute_antic): ... end of iteration here.
22411 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22413 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22414 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22415 (ftree-slp-vectorize): Likewise.
22416 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22417 can no longer be set independent of flag_tree_loop_vectorize.
22418 * omp-general.c (emp_max_vf): Likewise.
22419 * opts.c (enable_fdo_optimizations): Remove references to
22420 flag_tree_vectorize, these are now implicit.
22421 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22422 and leave it for the options machinery.
22424 2017-08-01 Martin Liska <mliska@suse.cz>
22426 PR middle-end/70140
22427 * builtins.c (expand_builtin_memcpy_args): Remove.
22428 (expand_builtin_memcpy): Call newly added function
22429 expand_builtin_memory_copy_args.
22430 (expand_builtin_memcpy_with_bounds): Likewise.
22431 (expand_builtin_mempcpy): Remove last argument.
22432 (expand_builtin_mempcpy_with_bounds): Likewise.
22433 (expand_builtin_memory_copy_args): New function created from
22434 expand_builtin_mempcpy_args with small modifications.
22435 (expand_builtin_mempcpy_args): Remove.
22436 (expand_builtin_stpcpy): Remove unused argument.
22437 (expand_builtin): Likewise.
22438 (expand_builtin_with_bounds): Likewise.
22440 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22443 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22444 print "ds:" only for immediates in generic address space.
22446 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22449 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22450 (ix86_function_ok_for_sibcall): Return false for naked functions.
22452 2017-08-01 Richard Biener <rguenther@suse.de>
22454 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22455 (compute_antic): Seed worklist with exit block predecessors.
22456 * cfganal.c (dfs_find_deadend): For a cycle return the source
22457 of the edge closing it.
22459 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22461 * config/aarch64/aarch64.c
22462 (aarch64_can_const_movi_rtx_p): Move 0 check.
22464 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22466 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22467 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22469 * match.pd: Ditto in address comparison pattern.
22471 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22473 PR tree-optimization/81627
22474 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22475 closed ssa form for store-store chain.
22477 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22479 PR tree-optimization/81620
22480 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22481 for store-store chain.
22483 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22485 PR tree-optimization/81588
22486 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22487 ranges[i].in_p, invert comparison code ccode. For >/>=,
22488 swap rhs1 and rhs2 and comparison code unconditionally,
22489 for </<= don't do that. Don't swap rhs1/rhs2 again if
22490 ranges[i].in_p, instead invert comparison code ccode if
22491 opcode or oe->rank is BIT_IOR_EXPR.
22494 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22495 a direct optab to conversion optab.
22496 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22497 with GET_MODE_INNER as last argument instead of optab_handler.
22498 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22499 vector extraction if possible and optab is available.
22500 * expr.c (store_constructor): Use convert_optab_handler instead
22501 of optab_handler. Use vector initialization from smaller
22502 vectors if possible and optab is available.
22503 * tree-vect-stmts.c (vectorizable_load): Likewise.
22504 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22505 now have two modes.
22506 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22507 of vec_init from half-sized vectors with the same element mode.
22508 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22509 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22510 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22511 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22512 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22513 after mode in gen_vec_extract* calls.
22514 (vec_extract<mode>): Renamed to ...
22515 (vec_extract<mode><ssescalarmodelower>): ... this.
22516 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22517 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22518 element mode after mode in gen_vec_init* calls.
22519 (VEC_INIT_HALF_MODE): New mode iterator.
22520 (vec_init<mode>): Renamed to ...
22521 (vec_init<mode><ssescalarmodelower>): ... this.
22522 (vec_init<mode><ssehalfvecmodelower>): New expander.
22523 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22524 (vec_extractv2sfsf): ... this.
22525 (vec_initv2sf): Renamed to ...
22526 (vec_initv2sfsf): ... this.
22527 (vec_extractv2si): Renamed to ...
22528 (vec_extractv2sisi): ... this.
22529 (vec_initv2si): Renamed to ...
22530 (vec_initv2sisi): ... this.
22531 (vec_extractv4hi): Renamed to ...
22532 (vec_extractv4hihi): ... this.
22533 (vec_initv4hi): Renamed to ...
22534 (vec_initv4hihi): ... this.
22535 (vec_extractv8qi): Renamed to ...
22536 (vec_extractv8qiqi): ... this.
22537 (vec_initv8qi): Renamed to ...
22538 (vec_initv8qiqi): ... this.
22539 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22540 (vec_init<mode>): Renamed to ...
22541 (vec_init<mode><VEC_base_l>): ... this.
22542 (vec_extract<mode>): Renamed to ...
22543 (vec_extract<mode><VEC_base_l>): ... this.
22544 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22545 (vec_initv2sfsf): ... this.
22546 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22547 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22548 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22549 element mode after mode in gen_vec_init* calls.
22550 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22551 (vec_init<mode><Vel>): ... this.
22552 (vec_extract<mode>): Renamed to ...
22553 (vec_extract<mode><Vel>): ... this.
22554 * config/aarch64/iterators.md (Vel): New mode attribute.
22555 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22556 Add element mode after mode in gen_vec_extract* calls.
22557 * config/s390/vector.md (non_vec_l): New mode attribute.
22558 (vec_extract<mode>): Renamed to ...
22559 (vec_extract<mode><non_vec_l>): ... this.
22560 (vec_init<mode>): Renamed to ...
22561 (vec_init<mode><non_vec_l>): ... this.
22562 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22563 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22565 * config/arm/iterators.md (V_elem_l): New mode attribute.
22566 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22567 (vec_extract<mode><V_elem_l>): ... this.
22568 (vec_extractv2di): Renamed to ...
22569 (vec_extractv2didi): ... this.
22570 (vec_init<mode>): Renamed to ...
22571 (vec_init<mode><V_elem_l>): ... this.
22572 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22573 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22574 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22575 Add element mode after gen_vec_extract* calls.
22576 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22577 (vec_init<mode><unitmode>): ... this.
22578 (vec_extract<mode>): Renamed to ...
22579 (vec_extract<mode><unitmode>): ... this.
22580 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22581 (vec_init<mode><unitmode>): ... this.
22582 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22583 (vec_initv2sfsf): ... this.
22584 (vec_extractv2sf): Renamed to ...
22585 (vec_extractv2sfsf): ... this.
22586 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22587 Add element mode after gen_vec_extract* calls.
22588 * config/mips/mips.md (unitmode): New mode iterator.
22589 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22590 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22591 * config/spu/spu.md (inner_l): New mode attribute.
22592 (vec_init<mode>): Renamed to ...
22593 (vec_init<mode><inner_l>): ... this.
22594 (vec_extract<mode>): Renamed to ...
22595 (vec_extract<mode><inner_l>): ... this.
22596 * config/sparc/sparc.md (veltmode): New mode iterator.
22597 (vec_init<VMALL:mode>): Renamed to ...
22598 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22599 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22600 (vec_initv2sisi): ... this.
22601 (vec_initv2sf): Renamed to ...
22602 (vec_initv2sfsf): ... this.
22603 (vec_extractv2sf): Renamed to ...
22604 (vec_extractv2sfsf): ... this.
22605 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22606 (vec_init<mode>): Renamed to ...
22607 (vec_init<mode><VEC_base_l>): ... this.
22608 (vec_extract<mode>): Renamed to ...
22609 (vec_extract<mode><VEC_base_l>): ... this.
22610 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22611 (vec_initv2sfsf): ... this.
22612 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22613 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22614 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22615 gen_vec_init* calls.
22617 2017-08-01 Richard Biener <rguenther@suse.de>
22619 PR tree-optimization/81297
22620 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22621 TREE_OVERFLOW from INTEGER_CSTs.
22623 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22625 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22627 2017-07-31 Carl Love <cel@us.ibm.com>
22629 * config/rs6000/rs6000-c: Add support for built-in functions
22630 vector signed char vec_xl_be (signed long long, signed char *);
22631 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22632 vector signed int vec_xl_be (signed long long, signed int *);
22633 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22634 vector signed long long vec_xl_be (signed long long, signed long long *);
22635 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22636 vector signed short vec_xl_be (signed long long, signed short *);
22637 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22638 vector double vec_xl_be (signed long long, double *);
22639 vector float vec_xl_be (signed long long, float *);
22640 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22641 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22642 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22644 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22645 (altivec_expand_builtin): Add switch statement to call
22646 altivec_expand_xl_be for each builtin.
22647 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22648 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22649 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22650 __builtin_vsx_le_be_v16qi.
22651 * doc/extend.texi: Update the built-in documentation file for the
22652 new built-in functions.
22654 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22657 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22659 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22661 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22663 * config.gcc: Add z14.
22664 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22665 CPU model numbers for z13s and z14.
22666 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22668 * config/s390/s390-opts.h (enum processor_type): Rename
22669 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22670 * config/s390/s390.c (processor_table): Add field for CPU name to
22671 be passed to Binutils.
22672 (s390_asm_output_machine_for_arch): Use the new field in
22673 processor_table for Binutils.
22674 (s390_expand_builtin): Replace arch12 with z14.
22675 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22676 (s390_get_sched_attrmask): Likewise.
22677 (s390_get_unit_mask): Likewise.
22678 * config/s390/s390.opt: Add z14 to processor_type enum.
22680 2017-07-31 Martin Jambor <mjambor@suse.cz>
22683 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22684 regardless of optimization level.
22686 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22687 Martin Liska <mliska@suse.cz>
22689 * predict.def: Remove old comment and adjust probability.
22690 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22691 PREDICT statements.
22693 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22696 * config/i386/i386.c (ix86_function_naked): New function.
22697 (ix86_can_use_return_insn_p): Return false for naked functions.
22698 (ix86_expand_prologue): Skip prologue for naked functions.
22699 (ix86_expand_epilogue): Skip epilogue for naked functions
22700 and emit trap instruction.
22701 (ix86_warn_func_return): New function.
22702 (ix86_attribute_table): Add "naked" attribute specification.
22703 (TARGET_WARN_FUNC_RETURN): Define.
22704 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22706 2017-07-31 Martin Liska <mliska@suse.cz>
22708 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22709 (dump_gimple_bb_header): Always dump BB info.
22710 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22712 2017-07-31 Martin Liska <mliska@suse.cz>
22715 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22716 also with current_function_decl non-null equality.
22718 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22721 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22722 change type to the element type, instead add eltype variable and
22723 use it where we are interested in the element type.
22725 PR tree-optimization/81603
22726 * ipa-polymorphic-call.c
22727 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22728 offset arithmetic in offset_int, bail out if the resulting bit offset
22729 doesn't fit into shwi.
22731 2017-07-31 Martin Liska <mliska@suse.cz>
22733 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22734 (gimplify_save_expr): Fix comment.
22736 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22739 * config/i386/i386.c (ix86_function_arg): Update arguments for
22741 (ix86_compute_frame_layout): Set the initial stack offset to
22742 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22743 INCOMING_FRAME_SP_OFFSET.
22744 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22745 stack before exception handler returns.
22746 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22747 the 'ERROR_CODE' for exception handler.
22749 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22751 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22752 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22753 (ASM_OUTPUT_REG_POP): Ditto.
22754 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22755 instead of asm_fprintf to output pure string.
22757 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22759 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22760 to imported_module_or_decl hook.
22761 (debug_nothing_tree_tree_tree_bool): Remove.
22762 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22763 * debug.c (do_nothing_debug_hooks): Use
22764 debug_nothing_tree_tree_tree_bool_bool instead of
22765 debug_nothing_tree_tree_tree_bool.
22766 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22767 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22768 * sdbout.c (sdb_debug_hooks): Likewise.
22769 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22770 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22772 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22773 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22774 attribute, don't add anything.
22776 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22778 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22779 (fold_build2_stat_loc): Likewise.
22780 (fold_build3_stat_loc): Likewise.
22781 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22782 (fold_build1_loc): Remove macro.
22783 (fold_build2_loc): Likewise.
22784 (fold_build3_loc): Likewise.
22786 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22788 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22789 (gimple_build_debug_bind_source_stat): Likewise.
22790 * gimple.h (gimple_build_debug_bind): Remove macro.
22791 (gimple_build_debug_bind_source): Likewise.
22793 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22795 * bitmap.c (bitmap_alloc): Adjust.
22796 (bitmap_gc_alloc): Likewise.
22797 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22799 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22801 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22802 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22803 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22804 (bitmap_gc_alloc_stat): Likewise.
22805 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22807 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22809 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22810 * rtl.h (shallow_copy_rtx): Remove macro.
22812 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22814 * emit-rtl.c (gen_raw_REG): Adjust.
22815 * gengenrtl.c (gendef): Likewise.
22816 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22817 * rtl.h (rtx_alloc): Remove macro.
22819 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22821 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22822 (build_tree_list_stat): Likewise.
22823 * tree.h (build_tree_list): Remove macro.
22824 (build_tree_list_vec): Likewise.
22826 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22828 * tree.c (make_vector_stat): Remove _stat from name.
22829 (build_vector_stat): Likewise.
22830 * tree.h (make_vector_stat): Remove macro.
22831 (build_vector_stat): Likewise.
22833 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22835 * tree.h (build_var_debug_value): Remove prototype.
22837 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22839 * tree.c (tree_cons_stat): Remove _stat from name.
22840 * tree.h (tree_cons): Remove macro.
22842 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22844 * tree.c (build_vl_exp_stat): Remove _stat from name.
22845 * tree.h (build_vl_exp): Remove macro.
22847 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22849 * tree.c (build_decl_stat): Remove _stat from name.
22850 * tree.h (build_decl): Remove macro.
22852 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22854 * gimple.c (gimple_build_with_ops_stat): Adjust.
22855 (gimple_alloc_stat): Remove _stat from name.
22856 * gimple.h (gimple_alloc): Remove macro.
22858 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22860 * tree.c (make_tree_vec_stat): Remove _stat from name.
22861 (grow_tree_vec_stat): Likewise.
22862 * tree.h (make_tree_vec_stat): Adjust prototype.
22863 (grow_tree_vec_stat): Likewise.
22864 (make_tree_vec): Remove macro.
22865 (grow_tree_vec): Likewise.
22867 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22869 * fold-const.c (fold_build1_stat_loc): Adjust.
22870 (fold_build2_stat_loc): Likewise.
22871 (fold_build3_stat_loc): Likewise.
22872 * tree.c (build0_stat): Remove _stat from name.
22873 (build1_stat): Likewise.
22874 (build2_stat): Likewise.
22875 (build3_stat): Likewise.
22876 (build4_stat): Likewise.
22877 (build5_stat): Likewise.
22878 * tree.h (build1_loc): Remove macro, and rename _stat function
22880 (build2_loc): Likewise.
22881 (build3_loc): Likewise.
22882 (build4_loc): Likewise.
22883 (build5_loc): Likewise.
22885 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22887 * tree.c (make_int_cst_stat): Remove _stat from name.
22888 * tree.h (make_int_cst_stat): Adjust prototype.
22889 (make_int_cst): Remove macro.
22891 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22893 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22894 * tree.h (make_tree_binfo_stat): Adjust prototype.
22895 (make_tree_binfo): Remove.
22897 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22899 * tree.c (copy_node_stat): Rename to copy_node.
22900 (build_distinct_type_copy): Adjust.
22901 * tree.h (copy_node_stat): Adjust prototype.
22902 (copy_node): Remove macro.
22904 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22906 * tree.c (make_node_stat): rename to make_node.
22907 (build_tree_list_stat): Adjust.
22908 (build0_stat): Likewise.
22909 (build2_stat): Likewise.
22910 (build3_stat): Likewise.
22911 (build4_stat): Likewise.
22912 (build5_stat): Likewise.
22913 (build_decl_stat): Likewise.
22914 * tree.h (make_node_stat): Adjust prototype.
22915 (make_node): remove macro.
22917 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22919 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22920 (PPC_FEATURE2_SCV): Likewise.
22921 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22923 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22925 * config/aarch64/aarch64.c
22926 (aarch64_internal_mov_immediate): Add new special pattern.
22927 * config/aarch64/aarch64.md (*movdi_aarch64):
22928 Add reg/32bit const mov case.
22930 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22931 Richard Sandiford <richard.sandiford@linaro.org>
22933 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22934 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22935 Add integer and movi cases.
22936 (movi-split-hf-df-sf split, fp16): New.
22937 (enabled): Added TARGET_FP_F16INST.
22938 * config/aarch64/iterators.md (GPF_HF): New.
22939 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22941 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22943 * config/aarch64/aarch64.c
22944 (aarch64_simd_container_mode): Add prototype.
22945 (aarch64_expand_mov_immediate): Add HI support.
22946 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22947 (aarch64_can_const_movi_rtx_p): New.
22948 (aarch64_preferred_reload_class):
22949 Remove restrictions of using FP registers for certain SIMD operations.
22950 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22951 (aarch64_valid_floating_const): Add integer move validation.
22952 (aarch64_simd_imm_scalar_p): Remove.
22953 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22954 (aarch64_legitimate_constant_p): Expand list of supported cases.
22955 * config/aarch64/aarch64-protos.h
22956 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22957 (aarch64_reinterpret_float_as_int): New.
22958 (aarch64_simd_imm_scalar_p): Remove.
22959 * config/aarch64/constraints.md (Uvi): New.
22960 (Dd): Split into Ds and new Dd.
22961 * config/aarch64/aarch64.md (*movsi_aarch64):
22963 (*movdi_aarch64): Add SIMD mov case.
22965 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22967 * tree-predcom.c: (struct chain): Handle store-store chain in which
22968 stores for elimination only store loop invariant values.
22969 (execute_pred_commoning_chain): Ditto.
22970 (prepare_initializers_chain_store_elim): Ditto.
22971 (prepare_finalizers): Ditto.
22972 (is_inv_store_elimination_chain): New function.
22973 (initialize_root_vars_store_elim_1): New function.
22975 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22977 * tree-predcom.c: Revise general description of the pass.
22978 (enum chain_type): New enum type for store elimination.
22979 (struct chain): New field supporting store elimination.
22980 (struct component): Ditto.
22981 (dump_chain): Dump store-stores chain.
22982 (release_chain): Release resources.
22983 (split_data_refs_to_components): Compute and create component
22984 contains only stores for elimination.
22985 (get_chain_last_ref_at): New function.
22986 (make_invariant_chain): Initialization.
22987 (make_rooted_chain): Specify chain type in parameter and record it.
22988 (add_looparound_copies): Skip for store-stores chain.
22989 (determine_roots_comp): Compute type of chain and pass it to
22991 (initialize_root_vars_store_elim_2): New function.
22992 (finalize_eliminated_stores): New function.
22993 (remove_stmt): Handle store for elimination.
22994 (execute_pred_commoning_chain): Execute predictive commoning on
22995 store-store chains.
22996 (determine_unroll_factor): Skip unroll for store-stores chain.
22997 (prepare_initializers_chain_store_elim): New function.
22998 (prepare_initializers_chain): Hanlde store-store chain.
22999 (prepare_finalizers_chain, prepare_finalizers): New function.
23000 (tree_predictive_commoning_loop): Return integer value indicating
23001 if loop is unrolled or lcssa form is corrupted.
23002 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
23004 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23006 * tree-predcom.c (initialize_root): Delete.
23007 (execute_pred_commoning_chain): Initialize root vars and replace
23008 reference of non-combined chain directly, rather than call above
23011 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23013 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
23014 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
23016 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23018 * tree-predcom.c (struct chain): New field init_seq.
23019 (release_chain): Release init_seq.
23020 (prepare_initializers_chain): Record intialization stmts in above
23022 (insert_init_seqs): New function.
23023 (tree_predictive_commoning_loop): Call insert_init_seqs.
23025 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23027 * tree-predcom.c (determine_roots_comp): Skip trivial components.
23029 2017-07-28 Richard Biener <rguenther@suse.de>
23031 * match.pd: Remove superfluous :c.
23032 * genmatch.c (simplify::id): Add member.
23033 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
23035 (current_id): New global.
23036 (dt_node::parent): Move from ...
23037 (dt_operand::parent): ... here. Add for_id member.
23038 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
23039 (decision_tree::find_node): Relax order requirement when
23040 merging DT_TRUE nodes to ones inbetween the current simplify
23041 and the one we try to merge with. Add diagnostic whenever
23042 we need to enforce pattern order by not merging.
23043 (decision_tree::insert): Set current_id.
23044 (decision_tree::print_node): Dump parent node and for_id.
23045 (parser::last_id): Add member.
23046 (parser::push_simplify): Assign unique id.
23047 (parser::parser): Initialize last_id.
23049 2017-07-28 Martin Liska <mliska@suse.cz>
23052 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
23053 gimple_build_debug_bind.
23055 2017-07-28 Richard Biener <rguenther@suse.de>
23057 PR tree-optimization/81502
23058 * match.pd: Add pattern combining BIT_INSERT_EXPR with
23060 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23062 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23063 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23064 for BIT_FIELD_REF args.
23065 * fold-const.c (make_bit_field_ref): Likewise.
23066 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23068 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23071 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23072 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23073 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23074 Or it into SANITIZER_UNDEFINED.
23075 * ubsan.c: Include gimple-fold.h and varasm.h.
23076 (ubsan_expand_ptr_ifn): New function.
23077 (instrument_pointer_overflow): New function.
23078 (maybe_instrument_pointer_overflow): New function.
23079 (instrument_object_size): Formatting fix.
23080 (pass_ubsan::execute): Call instrument_pointer_overflow
23081 and maybe_instrument_pointer_overflow.
23082 * internal-fn.c (expand_UBSAN_PTR): New function.
23083 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23084 * sanitizer.def (__ubsan_handle_pointer_overflow,
23085 __ubsan_handle_pointer_overflow_abort): New builtins.
23086 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23087 * internal-fn.def (UBSAN_PTR): New internal function.
23088 * opts.c (sanitizer_opts): Add pointer-overflow.
23089 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23090 * fold-const.c (build_range_check): Compute pointer range check in
23091 integral type if pointer arithmetics would be needed. Formatting
23094 2017-07-28 Martin Liska <mliska@suse.cz>
23097 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23098 parameters that are of a variable-length.
23100 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23102 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23104 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23105 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23106 (CRT_CALL_STATIC_FUNCTION): Likewise.
23107 (ASM_DEFAULT_SPEC): New define.
23108 (ASM_SPEC32): Likewise.
23109 (ASM_SPEC64): Likewise.
23110 (ASM_SPEC_COMMON): Likewise.
23111 (ASM_SPEC): Likewise.
23112 (INVALID_64BIT): Likewise.
23113 (LINK_OS_DEFAULT_SPEC): Likewise.
23114 (LINK_OS_SPEC32): Likewise.
23115 (LINK_OS_SPEC64): Likewise.
23116 (POWERPC_LINUX): Likewise.
23117 (PTRDIFF_TYPE): Likewise.
23118 (RESTORE_FP_PREFIX): Likewise.
23119 (RESTORE_FP_SUFFIX): Likewise.
23120 (SAVE_FP_PREFIX): Likewise.
23121 (SAVE_FP_SUFFIX): Likewise.
23122 (SIZE_TYPE): Likewise.
23123 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23124 (TARGET_64BIT): Likewise.
23125 (TARGET_64BIT): Likewise.
23126 (TARGET_AIX): Likewise.
23127 (WCHAR_TYPE_SIZE): Likewise.
23128 (WCHAR_TYPE): Undefine.
23129 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23130 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23131 (CPP_OS_RTEMS_SPEC): Delete.
23132 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23133 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23135 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23137 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23139 PR tree-optimization/81578
23140 * tree-parloops.c (build_new_reduction): Bail out if
23141 reduction_code isn't one of the standard OpenMP reductions.
23142 Move the details printing after that decision.
23144 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23146 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23147 related to reload_in_progress.
23148 (splat_input_operand): Likewise.
23149 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23151 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23153 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23154 (TARGET_INSTANTIATE_DECLS): Likewise.
23155 (legitimate_indexed_address_p): Delete reload_in_progress code.
23156 (rs6000_debug_legitimate_address_p): Likewise.
23157 (rs6000_eliminate_indexed_memrefs): Likewise.
23158 (rs6000_emit_le_vsx_store): Likewise.
23159 (rs6000_emit_move_si_sf_subreg): Likewise.
23160 (rs6000_emit_move): Likewise.
23161 (register_to_reg_type): Likewise.
23162 (rs6000_pre_atomic_barrier): Likewise.
23163 (rs6000_machopic_legitimize_pic_address): Likewise.
23164 (rs6000_allocate_stack_temp): Likewise.
23165 (rs6000_address_for_fpconvert): Likewise.
23166 (rs6000_address_for_altivec): Likewise.
23167 (rs6000_secondary_memory_needed_rtx): Delete function.
23168 (rs6000_check_sdmode): Likewise.
23169 (rs6000_alloc_sdmode_stack_slot): Likewise.
23170 (rs6000_instantiate_decls): Likewise.
23171 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23172 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23173 Delete reload_in_progress.
23174 (*vec_reload_and_plus_<mptrsize>): Likewise.
23175 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23176 (vsx_div_v2di): Likewise.
23177 (vsx_udiv_v2di): Likewise.
23179 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23181 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23182 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23183 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23184 (rs6000_debug_reg_global): Delete print of LRA status.
23185 (rs6000_option_override_internal): Delete dead LRA related code.
23186 (rs6000_lra_p): Delete function.
23187 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23189 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23191 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23192 * config/riscv/rtems.h: New file.
23194 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23195 Sudakshina Das <sudi.das@arm.com>
23197 * config/aarch64/aarch64.md
23198 (define_split for and<mode>3nr_compare): Move
23199 non aarch64_logical_operand to a register.
23200 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23201 register immediate operand to a register.
23202 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23204 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23206 PR middle-end/81564
23207 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23209 2017-07-27 Richard Biener <rguenther@suse.de>
23211 PR tree-optimization/81573
23212 PR tree-optimization/81494
23213 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23214 multi defuse cycle case.
23216 2017-07-27 Richard Biener <rguenther@suse.de>
23218 PR tree-optimization/81571
23219 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23222 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23224 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23225 earlier and only if MASK_FPU is set. Adjust formatting.
23227 2017-07-27 Martin Liska <mliska@suse.cz>
23229 * opt-functions.awk: Add validation of value of Init.
23230 * optc-gen.awk: Pass new argument.
23232 2017-07-27 Martin Liska <mliska@suse.cz>
23234 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23235 Fix wrong condition.
23237 2017-07-27 Martin Liska <mliska@suse.cz>
23239 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23240 BBs and edges seen by autoFDO.
23242 2017-07-27 Richard Biener <rguenther@suse.de>
23244 PR tree-optimization/81502
23245 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23246 with incompatible but same sized type.
23247 (execute_update_addresses_taken): Likewise.
23249 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23251 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23252 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23254 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23257 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23258 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23259 Change s_operand to memory_operand.
23261 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23263 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23264 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23265 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23266 Emit instructions rather than returning an expression. Handle TFmode
23267 and KFmode by casting to TImode.
23268 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23269 (rs6000_emit_le_vsx_store): Likewise.
23270 * config/rs6000/vsx.md (VSX_TI): New iterator.
23271 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23272 (*vsx_le_undo_permute_<mode>): Likewise.
23273 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23274 emit the split sequence.
23275 (*vsx_le_perm_store_<mode>): Likewise.
23277 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23279 PR tree-optimization/81555
23280 PR tree-optimization/81556
23281 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23282 if true, force CHANGED for the recursive invocation.
23283 (reassociate_bb): Remember original length of ops array, pass
23284 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23286 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23287 attributes for noipa attribute. For naked attribute use
23288 lookup_attribute first before lookup_attribute_spec.
23289 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23291 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23292 for functions with noipa attribute.
23293 (cgraph_externally_visible_p): Return true for functions with noipa
23295 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23296 for functions with noipa attribute.
23297 * doc/extend.texi: Document noipa function attribute.
23298 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23299 also for functions with noipa attribute.
23300 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23302 2017-07-26 Andrew Pinski <apinski@cavium.com>
23304 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23305 vec_unalign_load_cost and vec_unalign_store_cost.
23307 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23309 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23310 -mvsx-small-integer option.
23311 (ISA_3_0_MASKS_IEEE): Likewise.
23312 (OTHER_VSX_VECTOR_MASKS): Likewise.
23313 (POWERPC_MASKS): Likewise.
23314 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23315 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23316 code, only testing for DImode being allowed in non-VSX floating
23318 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23319 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23321 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23322 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23323 TARGET_P8_VECTOR test.
23324 (rs6000_secondary_reload_simple_move): Likewise.
23325 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23326 since TARGET_P9_VECTOR was already tested.
23327 (rs6000_opt_masks): Remove -mvsx-small-integer.
23328 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23329 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23331 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23332 test for TARGET_VEXTRACTUB was used, and that uses
23334 (p9 extract splitter): Likewise.
23335 (vsx_extract_<mode>_di_p9): Likewise.
23336 (vsx_extract_<mode>_store_p9): Likewise.
23337 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23338 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23339 the elimination of TARGET_VSX_SMALL_INTEGER.
23340 (vsx_extract_<mode>_p8): Likewise.
23341 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23342 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23343 (vsx_set_<mode>_p9): Likewise.
23344 (vsx_set_v4sf_p9): Likewise.
23345 (vsx_set_v4sf_p9_zero): Likewise.
23346 (vsx_insert_extract_v4sf_p9): Likewise.
23347 (vsx_insert_extract_v4sf_p9_2): Likewise.
23348 * config/rs6000/rs6000.md (sign extend splitter): Change
23349 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23350 (floatsi<mode>2_lfiwax_mem): Likewise.
23351 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23352 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23353 since a test for TARGET_P9_VECTOR was used.
23354 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23355 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23356 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23357 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23358 TARGET_P8_VECTOR test.
23359 (fix_trunc<mode>si2_stfiwx): Likewise.
23360 (fix_trunc<mode>si2_internal): Likewise.
23361 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23362 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23364 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23365 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23366 TARGET_P8_VECTOR test.
23367 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23368 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23369 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23371 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23372 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23373 since a test for TARGET_P9_VECTOR was used.
23374 (splitter for loading small constants): Likewise.
23376 2017-07-26 Andrew Pinski <apinski@cavium.com>
23378 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23381 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23384 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23385 (fp_valid_at): Likewise.
23387 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23389 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23390 (qdf24xx_addrcost_table): Likewise.
23391 (cortexa57_tunings): Update to use generic_branch_cost.
23392 (cortexa72_tunings): Likewise.
23393 (cortexa73_tunings): Likewise.
23394 (qdf24xx_tunings): Likewise.
23396 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23398 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23399 (thunderx2t99_branch_cost): Likewise.
23400 (cortexa35_tunings): Update to use generic_branch_cost.
23401 (cortexa53_tunings): Likewise.
23402 (cortexa57_tunings): Likewise.
23403 (cortexa72_tunings): Likewise.
23404 (cortexa73_tunings): Likewise.
23405 (thunderx2t99_tunings): Likewise.
23407 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23409 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23410 (sparc_option_override): Honour MASK_FSMULD.
23411 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23412 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23413 * config/sparc/sparc.opt (mfsmuld): New option.
23414 * doc/invoke.texi (mfsmuld): Document option.
23416 2017-07-26 Marek Polacek <polacek@redhat.com>
23418 PR middle-end/70992
23419 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23421 2017-07-26 Richard Biener <rguenther@suse.de>
23423 * gimple-match-head.c (do_valueize): Return OP if valueize
23425 (get_def): New helper to get at the def stmt of a SSA name
23426 if valueize allows.
23427 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23428 do_valueize to get at the def stmt.
23429 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23431 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23433 PR middle-end/46932
23434 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23436 2017-07-26 Martin Liska <mliska@suse.cz>
23439 * function.c (expand_function_start): Make expansion of
23440 nonlocal_goto_save_area after parm_birth_insn.
23442 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23444 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23445 from all CPU target flags enable members.
23447 2017-07-26 Richard Biener <rguenther@suse.de>
23449 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23450 (decision_tree::gen): Make 'type' const.
23451 (write_predicate): Likewise.
23453 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23455 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23456 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23457 (rs6000_option_override_internal): Likewise.
23458 (rs6000_expand_vector_set): Likewise.
23459 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23460 (TARGET_UPPER_REGS_SF): Likewise.
23461 (TARGET_UPPER_REGS_DI): Likewise.
23462 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23463 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23464 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23465 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23466 (Splitters for DI constants in Altivec registers): Likewise.
23467 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23468 (vsx_set_v4sf_p9): Likewise.
23469 (vsx_set_v4sf_p9_zero): Likewise.
23470 (vsx_insert_extract_v4sf_p9): Likewise.
23471 (vsx_insert_extract_v4sf_p9_2): Likewise.
23473 2017-07-25 Carl Love <cel@us.ibm.com>
23475 * doc/extend.texi: Update the built-in documentation file for the
23476 existing built-in functions
23477 vector signed char vec_cnttz (vector signed char);
23478 vector unsigned char vec_cnttz (vector unsigned char);
23479 vector signed short vec_cnttz (vector signed short);
23480 vector unsigned short vec_cnttz (vector unsigned short);
23481 vector signed int vec_cnttz (vector signed int);
23482 vector unsigned int vec_cnttz (vector unsigned int);
23483 vector signed long long vec_cnttz (vector signed long long);
23484 vector unsigned long long vec_cnttz (vector unsigned long long);
23486 2017-07-25 Andrew Pinski <apinski@cavium.com>
23488 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23489 accesses where the use is for the first operand of a BIT_INSERT.
23491 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23494 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23495 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23497 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23499 * config/i386/gstabs.h: Delete.
23500 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23502 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23504 * config/i386/i386.c (ix86_decompose_address): Do not check for
23505 register RTX when looking at index_reg or base_reg.
23506 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23508 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23510 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23511 to update EH info here.
23513 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23515 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23517 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23519 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23521 2017-07-25 Torsten Duwe <duwe@suse.de>
23523 * common.opt: Introduce -fpatchable-function-entry
23524 command line option, and its variables function_entry_patch_area_size
23525 and function_entry_patch_area_start.
23526 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23527 including a two-value parser.
23528 * target.def (print_patchable_function_entry): New target hook.
23529 * targhooks.h (default_print_patchable_function_entry): New function.
23530 * targhooks.c (default_print_patchable_function_entry): Likewise.
23531 * toplev.c (process_options): Switch off IPA-RA if
23532 patchable function entries are being generated.
23533 * varasm.c (assemble_start_function): Look at the
23534 patchable-function-entry command line switch and current
23535 function attributes and maybe generate NOP instructions by
23536 calling the print_patchable_function_entry hook.
23537 * doc/extend.texi: Document patchable_function_entry attribute.
23538 * doc/invoke.texi: Document -fpatchable_function_entry
23539 command line option.
23540 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23542 * doc/tm.texi: Re-generate.
23544 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23547 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23548 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23550 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23552 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23554 2017-07-25 Richard Biener <rguenther@suse.de>
23556 PR tree-optimization/81455
23557 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23558 not walk in cycles when looking for guards.
23560 2017-07-25 Richard Biener <rguenther@suse.de>
23562 PR tree-optimization/81529
23563 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23564 when optimizing backedge uses.
23566 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23568 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23570 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23571 to dl_section_ref. On AIX, append an expression to subtract
23572 the size of the section length to dl_section_ref.
23574 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23576 * configure.ac: If any of the config.* scripts fail, exit 1.
23577 * configure: Regenerate.
23579 2017-07-25 Richard Biener <rguenther@suse.de>
23581 PR middle-end/81546
23582 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23583 of immediate uses, be more verbose on errors.
23585 2017-07-25 Richard Biener <rguenther@suse.de>
23587 PR tree-optimization/81510
23588 * tree-vect-loop.c (vect_is_simple_reduction): When the
23589 reduction stmt is not inside the loop bail out.
23591 2017-07-25 Richard Biener <rguenther@suse.de>
23593 PR tree-optimization/81303
23594 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23595 profitability check against LOOP_VINFO_NITERSM1.
23597 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23599 * domwalk.c (cmp_bb_postorder): Simplify.
23600 (sort_bbs_postorder): New function. Use it...
23601 (dom_walker::walk): ...here to optimize common cases.
23603 2017-07-25 Martin Liska <mliska@suse.cz>
23606 * ipa-visibility.c (function_and_variable_visibility): Make the
23607 redirection just on target that supports aliasing.
23608 Fix GNU coding style.
23610 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23613 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23615 * config/arm/rtems.h: Likewise.
23616 * config/bfin/rtems.h: Likewise.
23617 * config/i386/rtemself.h: Likewise.
23618 * config/lm32/rtems.h: Likewise.
23619 * config/m32c/rtems.h: Likewise.
23620 * config/m68k/rtemself.h: Likewise.
23621 * config/microblaze/rtems.h: Likewise.
23622 * config/mips/rtems.h: Likewise.
23623 * config/moxie/rtems.h: Likewise.
23624 * config/nios2/rtems.h: Likewise.
23625 * config/powerpcspe/rtems.h: Likewise.
23626 * config/rs6000/rtems.h: Likewise.
23627 * config/rtems.h: Likewise.
23628 * config/sh/rtems.h: Likewise.
23629 * config/sh/rtemself.h: Likewise.
23630 * config/sparc/rtemself.h: Likewise.
23632 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23635 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23636 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23637 * tree-ssa-structalias.c (alias_get_name): Same.
23639 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23642 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23643 instructions if no du chain is found.
23645 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23647 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23649 2017-07-25 Richard Biener <rguenther@suse.de>
23651 PR middle-end/81505
23652 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23655 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23657 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23658 upper-regs options.
23659 (ISA_2_7_MASKS_SERVER): Likewise.
23660 (ISA_3_0_MASKS_IEEE): Likewise.
23661 (OTHER_P8_VECTOR_MASKS): Likewise.
23662 (OTHER_VSX_VECTOR_MASKS): Likewise.
23663 (POWERPC_MASKS): Likewise.
23664 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23665 duplicate list of options.
23666 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23667 explicit -mupper-regs options.
23668 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23669 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23670 alias for -mupper-regs-df.
23671 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23672 (rs6000_init_hard_regno_mode_ok): Likewise.
23673 (rs6000_option_override_internal): Likewise.
23674 (rs6000_opt_masks): Likewise.
23675 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23676 options in terms of whether -mvsx or -mpower8-vector was used.
23677 (TARGET_UPPER_REGS_DI): Likewise.
23678 (TARGET_UPPER_REGS_SF): Likewise.
23679 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23680 -mupper-regs-* options.
23682 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23684 * passes.c (emergency_dump_function): Print some empty lines and a
23685 header before the RTL dump.
23687 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23689 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23691 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23694 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23695 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23697 2017-07-24 Carl Love <cel@us.ibm.com>
23699 * config/rs6000/rs6000-c.c: Add support for built-in functions
23700 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23701 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23702 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23703 vec_extract_fp_from_shortl): Add defines for the two builtins.
23704 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23705 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23707 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23708 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23709 * doc/extend.texi: Update the built-in documentation file for the
23710 new built-in function.
23712 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23715 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23717 * doc/generic.texi: Likewise.
23718 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23719 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23721 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23723 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23724 (aarch64_mls_elt_merge<mode>): Likewise.
23726 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23728 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23729 having __cxa_atexit.
23731 2017-07-23 Michael Collison <michael.collison@arm.com>
23733 * config/arm/arm.c (arm_option_override): Deprecate
23734 use of -mstructure-size-boundary.
23735 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23736 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23738 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23741 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23742 Reduce cost estimate for direct moves.
23744 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23747 * config/i386/i386.c (ix86_option_override_internal): Disable
23748 BMI, BMI2 and TBM instructions for -m16.
23750 2017-07-21 Carl Love <cel@us.ibm.com>
23752 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23753 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23754 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23755 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23756 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23757 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23758 VMULOSW): New enum "unspec" values.
23759 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23760 altivec_vmulosw): New patterns.
23761 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23762 VMULOSW): Add definitions.
23764 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23766 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23767 (qdf24xx): Likewise.
23768 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23769 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23770 (AARCH64_FL_V8_1): Renumber.
23771 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23772 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23773 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23774 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23775 rdma to feature modifiers list.
23777 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23779 PR middle-end/56727
23780 * ipa-visibility (function_and_variable_visibility): Convert
23781 recursive PLT call to direct call if appropriate.
23783 2017-07-21 Andrew Pinski <apinski@cavium.com>
23785 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23786 operand 1 to see if the types precision matches.
23787 * fold-const.c (operand_equal_p): Likewise.
23789 2017-07-21 Richard Biener <rguenther@suse.de>
23791 PR tree-optimization/81303
23792 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23793 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23794 (vect_peeling_hash_get_lowest_cost): Adjust.
23795 (vect_enhance_data_refs_alignment): Likewise. Use
23796 vect_get_peeling_costs_all_drs to compute the penalty for no
23797 peeling to match up costs.
23799 2017-07-21 Richard Biener <rguenther@suse.de>
23801 PR tree-optimization/81500
23802 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23803 we didn't identify a reduction path.
23805 2017-07-21 Tom de Vries <tom@codesourcery.com>
23806 Cesar Philippidis <cesar@codesourcery.com>
23808 PR gcov-profile/81442
23809 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23812 2017-07-21 Tom de Vries <tom@codesourcery.com>
23815 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23817 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23818 nvptx_override_options_after_change.
23820 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23822 * dwarf2out.c (output_file_names): Avoid double testing for
23823 dwarf_version >= 5.
23825 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23827 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23829 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23831 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23833 (try_crossjump_to_edge): Do not punt on partitioned functions.
23835 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23837 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23838 Put all BBs reachable only via paths crossing cold region to cold
23840 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23842 2016-07-21 Richard Biener <rguenther@suse.de>
23844 PR tree-optimization/81303
23845 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23846 into account prologue and epilogue iterations when raising
23847 min_profitable_iters to sth at least covering one vector iteration.
23849 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23851 * config/arm/arm.c (arm_test_cpu_arch_dat):
23854 2017-07-20 Nathan Sidwell <nathan@acm.org>
23856 Remove TYPE_METHODS.
23857 * tree.h (TYPE_METHODS): Delete.
23858 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23859 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23860 (dbxout_type_methods): Scan TYPE_FIELDS.
23861 (dbxout_type): Don't check TYPE_METHODS here.
23862 * function.c (use_register_for_decl): Always ignore register for
23863 class types when not optimizing.
23864 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23865 * tree.c (free_lang_data_in_type): Stitch out member functions and
23866 templates from TYPE_FIELDS.
23867 (build_distinct_type_copy, verify_type_variant,
23868 verify_type): Member fns are on TYPE_FIELDS.
23869 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23870 * tree-pretty-print.c (dump_generic_node): Likewise.
23872 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23875 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23876 V2TImode and V4TImode.
23877 (ix86_expand_vector_extract): Likewise.
23878 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23879 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23880 (ssescalarmode): Handle V4TImode and V2TImode.
23881 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23882 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23883 (VEXTRACTI128_MODE): New mode iterator.
23884 (splitter for *vec_extractv?ti first element): New.
23885 (VEC_INIT_MODE): New mode iterator.
23886 (vec_init<mode>): Consolidate 3 expanders into one using
23887 VEC_INIT_MODE mode iterator.
23889 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23891 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23892 non_spilled_static_chain_regno_p.
23894 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23896 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23898 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23900 * bb-reorder.c (connect_traces): Allow copying of blocks within
23903 2017-07-20 Richard Biener <rguenther@suse.de>
23905 * gimple.h (gimple_phi_result): Add gphi * overload.
23906 (gimple_phi_result_ptr): Likewise.
23907 (gimple_phi_arg): Likewise. Adjust index assert to only
23908 allow actual argument accesses rather than all slots available
23910 (gimple_phi_arg_def): Add gphi * overload.
23911 * tree-phinodes.c (make_phi_node): Initialize only actual
23913 (resize_phi_node): Clear memory not covered by old node,
23914 do not initialize excess argument slots.
23915 (reserve_phi_args_for_new_edge): Initialize new argument slot
23918 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23920 PR tree-optimization/81388
23922 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23924 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23925 by removing computation of may_be_zero.
23927 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23928 Tom de Vries <tom@codesourcery.com>
23930 PR middle-end/81030
23931 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23932 when gimple level profile disagrees with what RTL expander did.
23934 2017-07-20 Richard Biener <rguenther@suse.de>
23936 PR tree-optimization/61171
23937 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23938 (vect_analyze_stmt): Add slp instance parameter.
23939 (vectorizable_reduction): Likewise.
23940 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23941 (vect_is_simple_reduction): Deal with chains not detected
23942 as SLP reduction chain, specifically not properly associated
23943 chains containing a mix of plus/minus.
23944 (get_reduction_op): Remove.
23945 (get_initial_defs_for_reduction): Simplify, pass in whether
23946 this is a reduction chain, pass in the SLP node for the PHIs.
23947 (vect_create_epilog_for_reduction): Get the SLP instance as
23949 (vectorizable_reduction): Get the SLP instance as arg.
23950 During analysis remember the SLP node with the PHIs in the
23951 instance. Simplify getting at the vectorized reduction PHIs.
23952 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23953 through SLP instance.
23954 (vect_slp_analyze_operations): Likewise.
23955 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23956 (vect_transform_stmt): Likewise.
23958 2017-07-20 Tom de Vries <tom@codesourcery.com>
23960 PR tree-optimization/81489
23961 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23962 read of phi arg location to before loop that modifies phi.
23964 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23966 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23969 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23971 PR middle-end/81331
23972 * except.c (execute): Fix ordering issue.
23974 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23976 PR rtl-optimization/81423
23977 * combine.c (make_compound_operation_int): Don't try to optimize
23978 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23980 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23982 PR rtl-optimization/81423
23983 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23984 with a constant that is -1 in the truncated to mode.
23986 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23988 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23989 (determine_unlikely_bbs): ... here.
23990 * predict.h (propagate_unlikely_bbs_forward): Declare.
23991 * cfgexpand.c (pass_expand::execute): Use it.
23992 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23994 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23995 propagate_unlikely_bbs_forward.
23997 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23999 PR middle-end/81331
24000 * except.c (maybe_add_nop_after_section_switch): New function.
24003 2017-07-19 Tom de Vries <tom@codesourcery.com>
24005 * gimple.h (gimple_phi_set_arg): Make assert more strict.
24007 2017-07-19 Tom de Vries <tom@codesourcery.com>
24009 * gimple.h (gimple_phi_arg): Make assert more strict.
24011 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
24013 * config.gcc (powerpc*-*-*): Add mmintrin.h.
24014 * config/rs6000/mmintrin.h: New file.
24015 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
24017 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24019 PR tree-optimization/81346
24020 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
24022 2017-07-19 Tom de Vries <tom@codesourcery.com>
24024 * config/nvptx/nvptx.md (VECIM): Add V2DI.
24026 2017-07-19 Tom de Vries <tom@codesourcery.com>
24028 * config/nvptx/nvptx-modes.def: Add V2DImode.
24029 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
24030 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
24031 (nvptx_output_mov_insn): Handle lack of mov.b128.
24032 (nvptx_print_operand): Handle 'H' and 'L' codes.
24033 (nvptx_vector_mode_supported): Allow V2DImode.
24034 (nvptx_preferred_simd_mode): New function.
24035 (nvptx_data_alignment): New function.
24036 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
24037 nvptx_preferred_simd_mode.
24038 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
24040 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
24042 2017-07-19 Tom de Vries <tom@codesourcery.com>
24044 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
24045 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
24046 (nvptx_vector_mode_supported): New function. Allow V2SImode.
24047 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
24048 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
24049 (mov<VECIM>_insn): New define_insn.
24050 (define_expand "mov<VECIM>): New define_expand.
24052 2017-07-19 Tom de Vries <tom@codesourcery.com>
24054 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
24056 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24058 PR tree-optimization/81346
24059 * fold-const.h (fold_div_compare, range_check_type): Declare.
24060 * fold-const.c (range_check_type): New function.
24061 (build_range_check): Use range_check_type.
24062 (fold_div_compare): No longer static, rewritten into
24063 a match.pd helper function.
24064 (fold_comparison): Don't call fold_div_compare here.
24065 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24066 as helper function.
24068 2017-07-19 Nathan Sidwell <nathan@acm.org>
24070 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24071 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24072 * tree.c (find_decls_types_r, verify_type): Use
24073 TYPE_{MIN,MAX}_VALUE_RAW.
24074 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24075 (hash_tree): Likewise.
24076 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24078 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24081 2017-07-18 Tom de Vries <tom@codesourcery.com>
24083 PR middle-end/81464
24084 * omp-expand.c (expand_omp_for_static_chunk): Handle
24085 equal-argument loop exit phi.
24087 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24090 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24091 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24092 operand 2 predicate.
24093 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24094 operand 2 predicate.
24095 (ror,rol -> rorx splitters): Use const_int_operand as
24096 operand 2 predicate.
24098 2017-06-18 Richard Biener <rguenther@suse.de>
24100 PR tree-optimization/81410
24101 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24102 the gap in the ! slp_perm SLP case after each group.
24104 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24106 PR middle-end/81463
24107 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24110 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24112 PR middle-end/81462
24113 * predict.c (set_even_probabilities): Cleanup; do not affect
24114 probabilities that are already known.
24115 (combine_predictions_for_bb): Call even when count is set.
24117 2017-07-18 Nathan Sidwell <nathan@acm.org>
24119 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24122 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24125 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24126 optimization for loop niter analysis.
24128 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24131 * config/avr/avr.c (avr_optimize_casesi): Don't use
24132 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24134 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24136 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24137 body_cost_vec from _vect_peel_extended_info.
24138 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24139 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24142 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24144 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24146 2017-07-18 Richard Biener <rguenther@suse.de>
24148 PR tree-optimization/80620
24149 PR tree-optimization/81403
24150 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24151 info when re-using a VN table entry.
24153 2017-07-18 Richard Biener <rguenther@suse.de>
24155 PR tree-optimization/81418
24156 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24157 vectype_in. Verify that with lane-reducing reduction operations
24158 we have a single def-use cycle.
24160 2017-07-17 Carl Love <cel@us.ibm.com>
24162 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24164 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24165 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24166 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24167 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24168 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24169 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24170 VMULOSW): New enum "unspec" values.
24171 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24172 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24173 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24174 altivec_vmulosw): New patterns.
24175 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24176 VMULOSW): Add definitions.
24178 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24180 * config/alpha/alpha.c: Include predict.h.
24182 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24184 * tree-vrp.c (compare_assert_loc): Fix comparison function
24185 to return predictable results.
24187 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24189 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24191 (subdi3): Likewise.
24192 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24193 * doc/invoke.texi (mexpand-adddi): Update text.
24195 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24197 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24198 that also clobbers the CC register. The old expand code is moved
24200 (*arc_clzsi2): ... here.
24201 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24202 the CC register. The old expand code is moved to ...
24203 (arc_ctzsi2): ... here.
24205 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24207 * config/arc/arc.opt (mindexed-loads): Use initial value
24208 TARGET_INDEXED_LOADS_DEFAULT.
24209 (mauto-modify-reg): Use initial value
24210 TARGET_AUTO_MODIFY_REG_DEFAULT.
24211 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24212 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24213 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24214 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24216 2017-07-17 Martin Liska <mliska@suse.cz>
24219 * opts.c (finish_options): Do not allow -fgnu-tm
24220 w/ -fsanitize={kernel-,}address. Say sorry.
24222 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24225 * tree-loop-distribution.c (classify_partition): Only assert on
24226 numer of iterations.
24227 (merge_dep_scc_partitions): Delete prameter. Update function call.
24228 (distribute_loop): Remove code handling loop with unknown niters.
24229 (pass_loop_distribution::execute): Skip loop with unknown niters.
24231 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24234 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24235 function sort_partitions_by_post_order.
24237 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24239 PR tree-optimization/81374
24240 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24241 the max index of basic blocks, rather than number of basic blocks.
24243 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24245 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24247 (arc_legitimate_pic_operand_p): Likewise.
24248 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24250 (arc_needs_pcl_p): Likewise.
24251 (arc_legitimate_pc_offset_p): Likewise.
24252 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24253 function is also used in constrains.md.
24254 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24255 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24256 PLUS. Only return true/false in known cases, otherwise assert.
24257 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24258 is already called in arc_legitimate_constant_p.
24259 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24261 (LEGITIMATE_PIC_OPERAND_P): Use
24262 arc_raw_symbolic_reference_mentioned_p function.
24263 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24268 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24269 Andrew Burgess <andrew.burgess@embecosm.com>
24271 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24272 (arc_return_address_register): New function.
24273 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24274 (arc_handle_fndecl_attribute): Add naked attribute.
24275 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24276 (TARGET_WARN_FUNC_RETURN): Likewise.
24277 (arc_allocate_stack_slots_for_args): New function.
24278 (arc_warn_func_return): Likewise.
24279 (machine_function): Change type fn_type.
24280 (arc_compute_function_type): Consider new naked function type,
24281 change function return type.
24282 (arc_must_save_register): Adapt to handle new
24283 arc_compute_function_type's return type.
24284 (arc_expand_prologue): Likewise.
24285 (arc_expand_epilogue): Likewise.
24286 (arc_return_address_regs): Delete.
24287 (arc_return_address_register): New function.
24288 (arc_epilogue_uses): Use above function.
24289 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24290 (arc_function_type): Change encoding, add naked type.
24291 (ARC_INTERRUPT_P): Change to handle the new encoding.
24292 (ARC_FAST_INTERRUPT_P): Likewise.
24293 (ARC_NORMAL_P): Define.
24294 (ARC_NAKED_P): Likewise.
24295 (arc_compute_function_type): Delete prototype.
24296 * config/arc/arc.md (in_ret_delay_slot): Use
24297 arc_return_address_register function.
24298 (simple_return): Likewise.
24299 (p_return_i): Likewise.
24301 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24303 PR tree-optimization/81428
24304 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24305 can't be built for those types.
24307 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24309 Remove stuff dead since r239246.
24311 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24312 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24313 (avr_inform_devices): Remove dead stuff.
24315 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24317 * config/arm/arm_neon.h: Fix softp typo.
24319 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24321 PR tree-optimization/81365
24322 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24323 aggregate moves onto bb predecessor edges, make sure there are no
24324 loads that could alias the lhs in between the start of bb and the
24327 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24330 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24331 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24332 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24334 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24336 PR tree-optimization/81396
24337 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24338 (init_symbolic_number): Initialize it to 1.
24339 (perform_symbolic_merge): Add n_ops from both operands into the new
24341 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24342 without base_addr as useless if they need more than one operation.
24343 (bswap_replace): Handle !bswap case for NULL base_addr.
24345 2017-07-17 Tom de Vries <tom@codesourcery.com>
24348 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24351 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24353 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24354 conditional builtin define __FIX_LEON3FT_B2BST.
24356 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24358 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24359 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24362 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24364 PR rtl-optimization/81424
24365 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24366 to remove potential trapping from operands if -fnon-call-exceptions.
24368 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24370 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24371 profile_proability for scalling.
24372 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24374 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24376 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24378 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24380 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24381 fixpoint arithmetics.
24383 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24385 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24386 fixpoint arithmetics.
24388 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24390 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24391 fixpoint arithmetics.
24393 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24395 * profile-count.h (profile_probability::from_reg_br_prob_note,
24396 profile_probability::to_reg_br_prob_note): New functions.
24397 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24398 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24399 * predict.c (probability_reliable_p): Update.
24400 (edge_probability_reliable_p): Update.
24401 (br_prob_note_reliable_p): Update.
24402 (invert_br_probabilities): Update.
24403 (add_reg_br_prob_note): New function.
24404 (combine_predictions_for_insn): Update.
24405 * asan.c (asan_clear_shadow): Update.
24406 * cfgbuild.c (compute_outgoing_frequencies): Update.
24407 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24408 (update_br_prob_note): Update.
24409 (rtl_verify_edges): Update.
24410 (purge_dead_edges): Update.
24411 (fixup_reorder_chain): Update.
24412 * emit-rtl.c (try_split): Update.
24413 * ifcvt.c (cond_exec_process_insns): Update.
24414 (cond_exec_process_if_block): Update.
24415 (dead_or_predicable): Update.
24416 * internal-fn.c (expand_addsub_overflow): Update.
24417 (expand_neg_overflow): Update.
24418 (expand_mul_overflow): Update.
24419 * loop-doloop.c (doloop_modify): Update.
24420 * loop-unroll.c (compare_and_jump_seq): Update.
24421 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24422 * predict.h: Update.
24423 * reorg.c (mostly_true_jump): Update.
24425 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24426 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24427 * config/arc/arc.c: (emit_unlikely_jump): Update.
24428 * config/arm/arm.c: (emit_unlikely_jump): Update.
24429 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24430 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24431 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24432 (ix86_print_operand): Update.
24433 (ix86_split_fp_branch): Update.
24434 (predict_jump): Update.
24435 * config/ia64/ia64.c (ia64_print_operand): Update.
24436 * config/mmix/mmix.c (mmix_print_operand): Update.
24437 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24438 (rs6000_expand_split_stack_prologue): Update.
24439 * config/rs6000/rs6000.c: Update.
24440 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24441 (s390_expand_vec_movstr): Update.
24442 (s390_expand_cs_tdsi): Update.
24443 (s390_expand_split_stack_prologue): Update.
24444 * config/sh/sh.c (sh_print_operand): Update.
24445 (expand_cbranchsi4): Update.
24446 (expand_cbranchdi4): Update.
24447 * config/sparc/sparc.c (output_v9branch): Update.
24448 * config/spu/spu.c (get_branch_target): Update.
24449 (ea_load_store_inline): Update.
24450 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24451 * config/tilepro/tilepro.c: Update.
24453 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24455 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24456 (gimplify_save_expr): Likewise.
24458 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24460 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24462 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24464 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24466 * timevar.def (TV_IPA_FNSUMMARY): Define.
24468 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24470 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24471 to back store errata sensitive sequence from being generated.
24472 (sqrtdf2_fix): Likewise.
24474 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24476 * tree-ssa-threadupdate.c (compute_path_counts,
24477 update_joiner_offpath_counts): Use profile_probability.
24479 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24482 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24484 * config/arm/arm-c.c (arm_cpu_builtins): Define
24485 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24487 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24489 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24490 array entries to represent __ieee128 versions of the
24491 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24492 scalar_extract_sig, and scalar_insert_exp built-in functions.
24493 (altivec_resolve_overloaded_builtin): Add special case handling
24494 for the __builtin_scalar_insert_exp function, as represented by
24495 the P9V_BUILTIN_VEC_VSIEDP constant.
24496 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24497 exponent support for __ieee128 argument.
24498 (VSESQP): Add scalar extract signature support for __ieee128
24500 (VSTDCNQP): Add scalar test negative support for __ieee128
24502 (VSIEQP): Add scalar insert exponent support for __int128 argument
24503 with __ieee128 result.
24504 (VSIEQPF): Add scalar insert exponent support for __ieee128
24505 argument with __ieee128 result.
24506 (VSTDCQP): Add scalar test data class support for __ieee128
24508 (VSTDCNQP): Add overload support for scalar test negative with
24509 __ieee128 argument.
24510 (VSTDCQP): Add overload support for scalar test data class
24511 __ieee128 argument.
24512 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24513 UNSPEC_VSX_SXSIGDP.
24514 (UNSPEC_VSX_SIEXPQP): New constant.
24515 (xsxexpqp): New insn for VSX scalar extract exponent quad
24517 (xsxsigqp): New insn for VSX scalar extract significand quad
24519 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24520 precision with floating point argument.
24521 (xststdcqp): New expand for VSX scalar test data class quad
24523 (xststdcnegqp): New expand for VSX scalar test negative quad
24525 (xststdcqp): New insn to match expansions for VSX scalar test data
24526 class quad precision and VSX scalar test negative quad precision.
24527 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24528 special case operand checking to enforce that second operand of
24529 VSX scalar test data class with quad precision argument is a 7-bit
24531 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24532 prototypes and descriptions of __ieee128 versions of
24533 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24534 scalar_test_data_class, and scalar_test_neg built-in functions.
24536 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24538 PR tree-optimization/81162
24539 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24540 replace a negate with an add.
24542 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24544 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24546 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24548 * config/arm/arm-c.c (arm_cpu_builtins): Define
24549 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24551 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24553 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24554 (armv8-r): Set ARM Cortex-R52 as default CPU.
24555 * config/arm/arm-tables.opt: Regenerate.
24556 * config/arm/arm-tune.md: Regenerate.
24557 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24559 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24560 extension for -mcpu=cortex-r52.
24562 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24564 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24565 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24566 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24567 (fp-armv8): Define it as FP_ARMv8 only.
24568 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24569 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24571 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24572 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24573 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24574 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24575 than TARGET_FPU_ARMV8.
24576 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24577 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24578 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24580 * config/arm/neon.md (neon_vrint): Likewise.
24581 (neon_vcvt): Likewise.
24582 (neon_<fmaxmin_op><mode>): Likewise.
24583 (<fmaxmin><mode>3): Likewise.
24584 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24585 * config/arm/predicates.md (arm_cond_move_operator): Check against
24586 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24588 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24590 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24591 to top of function.
24593 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24595 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24596 loop in comment with memset.
24598 2017-07-14 Martin Liska <mliska@suse.cz>
24600 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24601 * dwarf2out.c (is_java): Remove the function.
24602 (output_pubname): Remove usage of the function.
24603 (lower_bound_default): Remove usage of DW_LANG_Java.
24604 (gen_compile_unit_die): Likewise.
24605 * gcc.c: Remove compiler defaults for .java and .zip files.
24606 * gimple-expr.c (remove_suffix): Change as there's no longer
24607 extension than 4-letter one.
24608 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24609 (gimplify_save_expr): Likewise.
24610 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24611 as it's possible even for other languages than Java.
24612 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24613 * lto-opts.c (lto_write_options): Remove reference to Java.
24614 * opts.c (strip_off_ending): Update file extension handling.
24615 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24616 * tree-eh.c (lower_resx): Likewise.
24617 * tree.c (free_lang_data_in_type): Remove dead code.
24618 (find_decls_types_r): Likewise.
24619 (build_common_builtin_nodes): Remove Java from a comment.
24620 (verify_type): Remove dead code.
24621 * varasm.c (assemble_external): Remove Java from a comment.
24623 2017-07-14 Martin Liska <mliska@suse.cz>
24625 * opts.c (finish_options): Add quotes.
24626 (common_handle_option): Likewise.
24628 2017-07-14 Martin Liska <mliska@suse.cz>
24630 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24631 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24632 Remove N_SO_PASCAL.
24633 * dwarf2out.c (lower_bound_default): Do not handle
24635 (gen_compile_unit_die): Likewise.
24636 * gcc.c: Remove default extension binding for GNU Pascal.
24637 * stmt.c: Remove Pascal language from a comment.
24638 * xcoffout.c: Likewise.
24640 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24643 * diagnostic-show-locus.c (fixit_cmp): New function.
24644 (layout::layout): Sort m_fixit_hints.
24645 (column_range::column_range): Assert that the values are valid.
24646 (struct char_span): New struct.
24647 (correction::overwrite): New method.
24648 (struct source_line): New struct.
24649 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24650 calls in terms of classes source_line and char_span, and
24651 correction::overwrite.
24652 (selftest::test_overlapped_fixit_printing_2): New function.
24653 (selftest::diagnostic_show_locus_c_tests): Call it.
24655 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24657 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24658 early if there is no lhs.
24660 2017-07-13 Martin Liska <mliska@suse.cz>
24662 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24663 (gen_reference_type_die): Likewise.
24664 * stor-layout.c: Remove Pascal-related comment.
24666 2017-07-13 Martin Liska <mliska@suse.cz>
24668 * opts.c (finish_options): Add quotes to error messages.
24669 (parse_sanitizer_options): Likewise.
24671 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24673 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24675 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24677 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24679 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24681 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24683 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24685 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24688 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24689 provides the hardware capability bits, define the macro
24690 __BUILTIN_CPU_SUPPORTS__.
24691 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24692 if GLIBC does not provide the hardware capability bits. Add a
24693 gcc_unreachable call if the built-in cpu function is neither
24694 __builtin_cpu_is nor __builtin_cpu_supports.
24695 (rs6000_get_function_versions_dispatcher): Change the warning
24696 that an old GLIBC is used which does not export the capability
24697 bits to be an error.
24698 * doc/extend.texi (target_clones attribute): Document the
24699 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24700 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24701 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24702 the macros defined by GCC if the newer GLIBC is available.
24704 2017-07-12 Jeff Law <law@redhat.com>
24706 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24707 remaining includes slightly.
24708 * config/riscv/riscv-builtins.c: Include profile-count.h.
24710 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24713 * config/avr/avr.c (avr_set_current_function): In diagnostic
24714 messages: Quote keywords and (parts of) identifiers.
24715 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24718 2017-07-12 Carl Love <cel@us.ibm.com>
24720 * config/rs6000/rs6000-c.c: Add support for built-in functions
24721 vector bool char vec_revb (vector bool char);
24722 vector bool short vec_revb (vector short char);
24723 vector bool int vec_revb (vector bool int);
24724 vector bool long long vec_revb (vector bool long long);
24725 * doc/extend.texi: Update the built-in documentation file for the
24726 new built-in functions.
24728 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24730 * config/s390/s390.md: Remove movcc splitter.
24732 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24734 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24735 load/store on condition.
24737 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24740 * config/avr/avr.c (avr_encode_section_info)
24741 [progmem && !TREE_READONLY]: Error if progmem object needs
24744 2017-07-11 Michael Collison <michael.collison@arm.com>
24746 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24749 2017-07-11 Carl Love <cel@us.ibm.com>
24751 * config/rs6000/rs6000-c.c: Add support for builtins
24752 vector unsigned int vec_parity_lsbb (vector signed int);
24753 vector unsigned int vec_parity_lsbb (vector unsigned int);
24754 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24755 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24756 vector unsigned long long vec_parity_lsbb (vector signed long long);
24757 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24758 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24759 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24760 * doc/extend.texi: Update the built-in documentation file for the
24761 new built-in functions.
24763 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24765 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24766 (layout::m_primary_loc): New field.
24767 (layout::layout): Initialize new field. Move location filtering
24768 logic from here to...
24769 (layout::maybe_add_location_range): ...this new method. Add
24770 support for filtering to just the lines already specified by other
24772 (layout::will_show_line_p): New method.
24773 (gcc_rich_location::add_location_if_nearby): New method.
24774 (selftest::test_add_location_if_nearby): New test function.
24775 (selftest::diagnostic_show_locus_c_tests): Call it.
24776 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24779 2017-07-11 Tom de Vries <tom@codesourcery.com>
24781 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24782 (bb_first_real_insn): New function.
24783 (nvptx_single): Add extra initialization of broadcasted condition
24786 2017-07-11 Nathan Sidwell <nathan@acm.org>
24788 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24790 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24792 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24793 Binutils doc as TEXI will mess them up.
24794 * doc/invoke.texi (AVR Options): Same here.
24796 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24798 * config/sparc/sparc.opt (mfix-ut700): New option.
24799 (mfix-gr712rc): Likewise.
24800 (sparc_fix_b2bst): New variable.
24801 * doc/invoke.texi (SPARC options): Document them.
24802 (ARM options): Fix warnings.
24803 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24804 instructions to prevent sequences that can trigger the store-store
24805 errata for certain LEON3FT processors.
24806 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24807 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24808 * config/sparc/sparc.md (fix_b2bst): New attribute.
24809 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24811 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24814 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24816 (*rsqrtsf2_sse): Ditto.
24818 (div<mode>3): Macroize insn from divdf3 and divsf3
24819 using MODEF mode iterator.
24821 2017-07-10 Martin Sebor <msebor@redhat.com>
24823 PR tree-optimization/80397
24824 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24825 instead of testing for equality to INTEGER_TYPE.
24827 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24829 * config.gcc: Remove uclibc from arc target spec.
24831 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24833 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24835 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24838 * lto-wrapper.c (remove_option): New function.
24839 (merge_and_complain): Merge PIC/PIE options more realistically.
24841 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24843 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24847 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24848 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24849 * config.in: Regenerate.
24850 * configure: Regenerate.
24851 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24852 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24853 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24854 (TARGET_GASISR_PROLOGUES): ...target mask.
24855 * common/config/avr/avr-common.c
24856 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24857 Set -mgas-isr-prologues.
24858 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24859 INSERT_PASS_BEFORE for it.
24860 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24861 * config/avr/avr.c (avr_option_override)
24862 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24863 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24864 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24865 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24866 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24868 (make_avr_pass_pre_proep): New function.
24869 (emit_push_sfr) <treg>: Add argument to function and use it
24870 instead of TMP_REG.
24871 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24872 and set machine->gasisr.yes.
24873 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24874 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24875 __gcc_isr.n_pushed to .L__stack_usage.
24876 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24877 (avr_asm_final_postscan_insn): ...this new static function.
24878 * config/avr/avr.h (machine_function)
24879 <is_no_gccisr, use_L__stack_usage>: New fields.
24880 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24881 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24882 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24883 (gasisr, *gasisr): New expander and insn.
24884 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24885 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24886 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24888 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24890 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24893 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24895 Move jump-tables out of .text again.
24898 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24899 (ASM_OUTPUT_ADDR_VEC): New function.
24900 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24901 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24902 INSN_ADDRESSes as asm comment.
24903 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24904 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24905 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24906 * config/avr/avr.md (*tablejump): Adjust comment.
24907 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24908 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24910 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24911 (avr_output_addr_vec): New proto.
24912 (avr_log_t) <insn_addresses>: New field.
24914 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24917 * config/i386/i386.c (ix86_function_arg_advance): Set
24918 outgoing_args_on_stack to true if there are outgoing arguments
24920 (ix86_function_arg): Likewise.
24921 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24922 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24923 * config/i386/i386.h (machine_function): Add
24924 outgoing_args_on_stack.
24926 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24928 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24929 supporting pthreds.
24930 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24932 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24934 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24935 (REAL_H): Remove $(MACHMODE_H).
24936 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24938 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24939 $(MACHMODE_H) and double-int.h.
24940 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24942 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24945 2017-07-07 Andrew Pinski <apinski@cavium.com>
24947 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24948 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24950 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24952 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24953 Add warning if GCC was not configured to link against a GLIBC that
24954 exports the hardware capability bits.
24955 (make_resolver_func): Make resolver function private and not a
24956 COMDAT function. Create the name with clone_function_name instead
24957 of make_unique_name.
24960 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24961 correct operand in doing the split.
24963 2017-07-07 Carl Love <cel@us.ibm.com>
24965 * config/rs6000/rs6000-c: Add support for built-in function
24966 vector unsigned short vec_pack_to_short_fp32 (vector float,
24968 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24969 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24970 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24971 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24972 (convert_4f32_8i16): Add define_expand.
24973 * doc/extend.texi: Update the built-in documentation file for the
24974 new built-in function.
24976 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24978 * config/sparc/m8.md: New file.
24979 * config/sparc/sparc.md: Include m8.md.
24981 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24983 * config/sparc/sparc.opt: New option -mvis4b.
24984 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24985 (sparc_option_override): Handle VIS4B.
24986 (enum sparc_builtins): Define
24987 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24988 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24989 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24990 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24991 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24992 (check_constant_argument): New function.
24993 (sparc_vis_init_builtins): Define builtins
24994 __builtin_vis_dictunpack{8,16,32},
24995 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24996 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24997 __builtin_vis_fpcmpde{8,16,32}shl and
24998 __builtin_vis_fpcmpur{8,16,32}shl.
24999 (sparc_expand_builtin): Check that the constant operands to
25000 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
25001 constant and in range.
25002 * config/sparc/sparc-c.c (sparc_target_macros): Handle
25004 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
25005 (SPARC_IMM5_P): Likewise.
25006 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
25007 (enabled): Handle vis4b.
25008 (UNSPEC_DICTUNPACK): New unspec.
25009 (UNSPEC_FPCMPSHL): Likewise.
25010 (UNSPEC_FPUCMPSHL): Likewise.
25011 (UNSPEC_FPCMPDESHL): Likewise.
25012 (UNSPEC_FPCMPURSHL): Likewise.
25013 (cpu_feature): New CPU feature `vis4b'.
25014 (dictunpack{8,16,32}): New insns.
25015 (FPCSMODE): New mode iterator.
25016 (fpcscond): New code iterator.
25017 (fpcsucond): Likewise.
25018 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
25019 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
25020 (fpcmpde{8,16,32}{si,di}shl): Likewise.
25021 (fpcmpur{8,16,32}{si,di}shl): Likewise.
25022 * config/sparc/constraints.md: Define constraints `q' for unsigned
25023 2-bit integer constants and `t' for unsigned 5-bit integer
25025 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
25027 (imm5_operand_dictunpack16): Likewise.
25028 (imm5_operand_dictunpack32): Likewise.
25029 (imm2_operand): Likewise.
25030 * doc/invoke.texi (SPARC Options): Document -mvis4b.
25031 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
25032 ditunpack* and fpcmp*shl builtins.
25034 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25036 * config.gcc: Handle m8 in --with-{cpu,tune} options.
25037 * config.in: Add HAVE_AS_SPARC6 define.
25038 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
25040 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
25042 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
25043 (CPP_CPU_SPEC): Handle m8.
25044 (ASM_CPU_SPEC): Likewise.
25045 * config/sparc/sparc-opts.h (enum processor_type): Add
25047 * config/sparc/sparc.c (m8_costs): New struct.
25048 (sparc_option_override): Handle TARGET_CPU_m8.
25049 (sparc32_initialize_trampoline): Likewise.
25050 (sparc64_initialize_trampoline): Likewise.
25051 (sparc_issue_rate): Likewise.
25052 (sparc_register_move_cost): Likewise.
25053 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
25054 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
25055 (ASM_CPU64_DEFAULT_SPEC): Likewise.
25056 (CPP_CPU_SPEC): Handle M8.
25057 (ASM_CPU_SPEC): Likewise.
25058 (AS_M8_FLAG): Define.
25059 * config/sparc/sparc.md: Add m8 to the cpu attribute.
25060 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25061 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25063 * configure: Regenerate.
25064 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25067 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25069 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25071 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25072 ("*movdi_insn_sp32"): Do not set v3pipe.
25073 ("*movsi_insn"): Likewise.
25074 ("*movdi_insn_sp64"): Likewise.
25075 ("*movsf_insn"): Likewise.
25076 ("*movdf_insn_sp32"): Likewise.
25077 ("*movdf_insn_sp64"): Likewise.
25078 ("*zero_extendsidi2_insn_sp64"): Likewise.
25079 ("*sign_extendsidi2_insn"): Likewise.
25080 ("*mov<VM32:mode>_insn"): Likewise.
25081 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25082 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25083 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25084 ("<vlop:code><VL:mode>3"): Likewise.
25085 ("*not_<vlop:code><VL:mode>3"): Likewise.
25086 ("*nand<VL:mode>_vis"): Likewise.
25087 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25088 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25089 ("one_cmpl<VL:mode>2"): Likewise.
25090 ("faligndata<VM64:mode>_vis"): Likewise.
25091 ("alignaddrsi_vis"): Likewise.
25092 ("alignaddrdi_vis"): Likweise.
25093 ("alignaddrlsi_vis"): Likewise.
25094 ("alignaddrldi_vis"): Likewise.
25095 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25096 ("bmaskdi_vis"): Likewise.
25097 ("bmasksi_vis"): Likewise.
25098 ("bshuffle<VM64:mode>_vis"): Likewise.
25099 ("cmask8<P:mode>_vis"): Likewise.
25100 ("cmask16<P:mode>_vis"): Likewise.
25101 ("cmask32<P:mode>_vis"): Likewise.
25102 ("pdistn<P:mode>_vis"): Likewise.
25103 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25105 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25107 * config/sparc/sparc.md ("subtype"): New insn attribute.
25108 ("*wrgsr_sp64"): Set insn subtype.
25109 ("*rdgsr_sp64"): Likewise.
25110 ("alignaddrsi_vis"): Likewise.
25111 ("alignaddrdi_vis"): Likewise.
25112 ("alignaddrlsi_vis"): Likewise.
25113 ("alignaddrldi_vis"): Likewise.
25114 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25115 ("fexpand_vis"): Likewise.
25116 ("fpmerge_vis"): Likewise.
25117 ("faligndata<VM64:mode>_vis"): Likewise.
25118 ("bshuffle<VM64:mode>_vis"): Likewise.
25119 ("cmask8<P:mode>_vis"): Likewise.
25120 ("cmask16<P:mode>_vis"): Likewise.
25121 ("cmask32<P:mode>_vis"): Likewise.
25122 ("fchksm16_vis"): Likewise.
25123 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25124 ("fmean16_vis"): Likewise.
25125 ("fp<plusminus_insn>64_vis"): Likewise.
25126 ("<plusminus_insn>v8qi3"): Likewise.
25127 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25128 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25129 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25130 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25131 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25132 ("*movqi_insn"): Likewise.
25133 ("*movhi_insn"): Likewise.
25134 ("*movsi_insn"): Likewise.
25135 ("movsi_pic_gotdata_op"): Likewise.
25136 ("*movdi_insn_sp32"): Likewise.
25137 ("*movdi_insn_sp64"): Likewise.
25138 ("movdi_pic_gotdata_op"): Likewise.
25139 ("*movsf_insn"): Likewise.
25140 ("*movdf_insn_sp32"): Likewise.
25141 ("*movdf_insn_sp64"): Likewise.
25142 ("*zero_extendhisi2_insn"): Likewise.
25143 ("*zero_extendqihi2_insn"): Likewise.
25144 ("*zero_extendqisi2_insn"): Likewise.
25145 ("*zero_extendqidi2_insn"): Likewise.
25146 ("*zero_extendhidi2_insn"): Likewise.
25147 ("*zero_extendsidi2_insn_sp64"): Likewise.
25148 ("ldfsr"): Likewise.
25149 ("prefetch_64"): Likewise.
25150 ("prefetch_32"): Likewise.
25151 ("tie_ld32"): Likewise.
25152 ("tie_ld64"): Likewise.
25153 ("*tldo_ldub_sp32"): Likewise.
25154 ("*tldo_ldub1_sp32"): Likewise.
25155 ("*tldo_ldub2_sp32"): Likewise.
25156 ("*tldo_ldub_sp64"): Likewise.
25157 ("*tldo_ldub1_sp64"): Likewise.
25158 ("*tldo_ldub2_sp64"): Likewise.
25159 ("*tldo_ldub3_sp64"): Likewise.
25160 ("*tldo_lduh_sp32"): Likewise.
25161 ("*tldo_lduh1_sp32"): Likewise.
25162 ("*tldo_lduh_sp64"): Likewise.
25163 ("*tldo_lduh1_sp64"): Likewise.
25164 ("*tldo_lduh2_sp64"): Likewise.
25165 ("*tldo_lduw_sp32"): Likewise.
25166 ("*tldo_lduw_sp64"): Likewise.
25167 ("*tldo_lduw1_sp64"): Likewise.
25168 ("*tldo_ldx_sp64"): Likewise.
25169 ("*mov<VM32:mode>_insn"): Likewise.
25170 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25171 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25173 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25175 * config/sparc/sparc.md ("type"): New insn type viscmp.
25176 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25178 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25179 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25180 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25181 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25183 ("n7_vis_logical_11cycle"): Likewise.
25184 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25185 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25186 * config/sparc/niagara.md ("niag_vis"): Likewise.
25187 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25188 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25190 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25192 * config/sparc/sparc.md: New instruction type `bmask'.
25193 (bmaskdi_vis): Use the `bmask' type.
25194 (bmasksi_vis): Likewise.
25195 * config/sparc/ultra3.md (us3_array): Likewise.
25196 * config/sparc/niagara7.md (n7_array): Likewise.
25197 * config/sparc/niagara4.md (n4_array): Likewise.
25198 * config/sparc/niagara2.md (niag2_vis): Likewise.
25199 (niag3_vis): Likewise.
25200 * config/sparc/niagara.md (niag_vis): Likewise.
25202 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25204 * ipa-comdats.c: Remove optimize check from gate.
25205 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25206 for functions not optimized.
25207 (ipa_fn_summary_read): Skip optimize check.
25208 (ipa_fn_summary_write): Likewise.
25209 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25211 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25213 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25214 (check_callers): Give up on caller which is not optimized.
25215 (inline_small_functions): Likewise.
25216 (ipa_inline): Do not give up when not optimizing.
25217 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25218 away unoptimizes cdtors.
25219 (whole_program_function_and_variable_visibility): Do
25220 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25221 * ipa.c (process_references): Do not check optimize.
25222 (symbol_table::remove_unreachable_nodes): Update optimize check.
25223 (set_writeonly_bit): Update optimize check.
25224 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25225 (pass_ipa_single_use::gate): Remove.
25227 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25229 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25230 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25231 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25232 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25233 permute_load, permute_store, adjust_extract, adjust_splat,
25234 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25235 replace_swap_with_copy, dump_swap_insn_table,
25236 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25237 recombine_lvx_pattern, recombine_stvx_pattern,
25238 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25239 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25240 to file rs6000-p8swap.c.
25241 * config/rs6000/rs6000-p8swap.c: New file.
25242 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25243 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25244 and rs6000*-*-* targets.
25246 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25248 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25250 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25252 * lto-wrapper.c (merge_and_complain): Do not merge
25253 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25254 fsigned_zeros, ftrapping_math, fwrapv.
25255 (append_compiler_options): Do not track these options.
25256 (append_linker_options): Likewie
25258 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25260 * cgraphunit.c (cgraph_node::finalize_function): When
25261 !flag_toplevel_reorde set no_reorder flag.
25262 (varpool_node::finalize_decl): Likewise.
25263 (symbol_table::compile): Drop no toplevel reorder path.
25265 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25267 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25268 edges; zero probability is not better than uninitialized.
25270 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25272 * asan.h (asan_sanitize_allocas_p): Declare.
25273 * asan.c (asan_sanitize_allocas_p): New function.
25274 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25275 (handle_builtin_alloca): Likewise.
25276 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25277 if !asan_sanitize_allocas_p.
25278 * params.def (asan-instrument-allocas): Add new option.
25279 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25280 * opts.c (common_handle_option): Disable allocas sanitization for
25283 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25285 * asan.c: Include gimple-fold.h.
25286 (get_last_alloca_addr): New function.
25287 (handle_builtin_stackrestore): Likewise.
25288 (handle_builtin_alloca): Likewise.
25289 (asan_emit_allocas_unpoison): Likewise.
25290 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25291 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25292 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25293 (instrument_builtin_call): Pass gimple iterator to
25294 get_mem_refs_of_builtin_call.
25295 (last_alloca_addr): New global.
25296 * asan.h (asan_emit_allocas_unpoison): Declare.
25297 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25298 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25299 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25300 if function calls alloca.
25301 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25302 * gimple-fold.h (replace_call_with_value): Declare.
25303 * internal-fn.c: Include asan.h.
25304 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25305 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25307 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25309 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25310 (C_SELFTEST_FLAGS): New.
25311 (CPP_SELFTEST_FLAGS): New.
25312 (SELFTEST_DEPS): New, from deps of s-selftest.
25313 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25314 (CPP_SELFTEST_DEPS): New.
25315 (selftest): Add dependency on s-selftest-c++.
25316 (s-selftest): Rename to...
25317 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25318 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25319 than SELFTEST_FLAGS.
25320 (selftest-gdb): Rename to...
25321 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25323 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25324 (selftest-valgrind): Rename to...
25325 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25327 (selftest-valgrind): Reintroduce as an alias for
25328 selftest-c-valgrind.
25329 (s-selftest-c++): New.
25330 (selftest-c++-gdb): New.
25331 (selftest-c++-valgrind): New.
25333 2017-07-06 Olivier Hainque <hainque@adacore.com>
25335 * gcc.c (process_command): When deciding if undefined variables
25336 should be ignored when processing specs, accept "gcc -v" as well.
25338 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25340 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25341 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25343 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25345 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25346 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25347 * config/arm/arm-tables.opt: Regenerate.
25348 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25350 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25352 2017-07-06 Carl Love <cel@us.ibm.com>
25354 * ChangeLog: Clean up from mid air collision
25356 2017-07-06 Carl Love <cel@us.ibm.com>
25358 * config/rs6000/rs6000-c.c: Add support for built-in functions
25359 vector signed int vec_subc (vector signed int, vector signed int);
25360 vector signed __int128 vec_subc (vector signed __int128,
25361 vector signed __int128);
25362 vector unsigned __int128 vec_subc (vector unsigned __int128,
25363 vector unsigned __int128);
25364 vector signed int vec_sube (vector signed int, vector signed int,
25365 vector signed int);
25366 vector unsigned int vec_sube (vector unsigned int,
25367 vector unsigned int,
25368 vector unsigned int);
25369 vector signed __int128 vec_sube (vector signed __int128,
25370 vector signed __int128,
25371 vector signed__int128);
25372 vector unsigned __int128 vec_sube (vector unsigned __int128,
25373 vector unsigned __int128,
25374 vector unsigned __int128);
25375 vector signed int vec_subec (vector signed int, vector signed int,
25376 vector signed int);
25377 vector unsigned int vec_subec (vector unsigned int,
25378 vector unsigned int,
25379 vector unsigned int);
25380 vector signed __int128 vec_subec (vector signed __int128,
25381 vector signed __int128,
25382 vector signed__int128);
25383 vector unsigned __int128 vec_subec (vector unsigned __int128,
25384 vector unsigned __int128,
25385 vector unsigned __int128);
25386 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25387 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25388 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25389 BU_ALTIVEC_OVERLOAD_X definitions.
25390 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25391 * doc/extend.texi: Update the built-in documentation file for the new
25392 built-in functions.
25394 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25397 * diagnostic-show-locus.c (layout::layout): Use start and finish
25398 spelling location for the start and finish of each range.
25399 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25400 Add unused aspect param.
25401 * input.c (expand_location_1): Add "aspect" param, and use it
25402 to access the correct part of the location.
25403 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25405 (expand_location_to_spelling_point): Likewise.
25406 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25407 param, and pass it to expand_location_1.
25409 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25411 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25412 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25413 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25414 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25415 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25416 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25417 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25418 _mm_maskz_getmant_ss): New intrinsics.
25419 (__builtin_ia32_getexpss128_mask): Changed to ...
25420 __builtin_ia32_getexpss128_round ... this.
25421 (__builtin_ia32_getexpsd128_mask): Changed to ...
25422 __builtin_ia32_getexpsd128_round ... this.
25423 * config/i386/i386-builtin-types.def
25424 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25425 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25426 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25427 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25428 __builtin_ia32_getmantss_mask_round): New builtins.
25429 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25430 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25431 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25432 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25433 * config/i386/sse.md
25434 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25435 avx512f_sgetexp<mode><mask_scalar_name>
25436 <round_saeonly_scalar_name> ... this.
25437 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25438 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25439 vgetexp<ssescalarmodesuffix>
25440 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25441 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25442 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25443 avx512f_vgetmant<mode><mask_scalar_name>
25444 <round_saeonly_scalar_name> ... this.
25445 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25446 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25447 vgetmant<ssescalarmodesuffix>
25448 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25449 %0<mask_scalar_operand4>, %1, %2
25450 <round_saeonly_scalar_mask_op4>, %3} ... this.
25451 * config/i386/subst.md (mask_scalar_operand4,
25452 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25453 round_saeonly_scalar_nimm_predicate): New subst attributes.
25455 2017-07-06 Julia Koval <julia.koval@intel.com>
25457 * config/i386/i386.c (ix86_erase_embedded_rounding):
25458 Remove code for old rounding pattern.
25460 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25462 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25464 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25466 * doc/sourcebuild.texi (Test Directives, Variants of
25467 dg-require-support): Add documentation for dg-require-stack-check.
25469 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25471 * config/i386/subst.md (mask_scalar, round_scalar,
25472 round_saeonly_scalar): New meta-templates.
25473 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25474 round_scalar_mask_operand3, round_scalar_mask_op3,
25475 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25476 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25477 round_saeonly_scalar_constraint,
25478 round_saeonly_scalar_prefix): New subst attribute.
25479 * config/i386/sse.md
25480 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25481 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25482 <round_scalar_name> ... this.
25483 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25484 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25485 <round_scalar_name> ... this.
25486 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25487 <sse>_vm<code><mode>3<mask_scalar_name>
25488 <round_saeonly_scalar_name> ... this.
25489 (v<plusminus_mnemonic><ssescalarmodesuffix>
25490 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25491 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25492 v<plusminus_mnemonic><ssescalarmodesuffix>
25493 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25494 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25495 (v<multdiv_mnemonic><ssescalarmodesuffix>
25496 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25497 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25498 v<multdiv_mnemonic><ssescalarmodesuffix>
25499 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25500 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25501 (v<maxmin_float><ssescalarmodesuffix>
25502 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25503 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25504 v<maxmin_float><ssescalarmodesuffix>
25505 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25506 %0<mask_scalar_operand3>, %1, %<iptr>2
25507 <round_saeonly_scalar_mask_op3>} ... this.
25509 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25511 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25512 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25514 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25515 Alan Hayward <alan.hayward@arm.com>
25516 David Sherwood <david.sherwood@arm.com>
25518 * combine.c (simplify_if_then_else): Remove "enum" before
25520 * compare-elim.c (can_eliminate_compare): Likewise.
25521 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25523 (aarch64_lookup_simd_builtin_type): Likewise.
25524 (aarch64_simd_builtin_type): Likewise.
25525 (aarch64_init_simd_builtin_types): Likewise.
25526 (aarch64_simd_expand_args): Likewise.
25527 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25529 (aarch64_reverse_mask): Likewise.
25530 (aarch64_simd_emit_reg_reg_move): Likewise.
25531 (aarch64_gen_adjusted_ldpstp): Likewise.
25532 (aarch64_ccmp_mode_to_code): Likewise.
25533 (aarch64_operands_ok_for_ldpstp): Likewise.
25534 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25535 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25537 (aarch64_min_divisions_for_recip_mul): Likewise.
25538 (aarch64_reassociation_width): Likewise.
25539 (aarch64_get_condition_code_1): Likewise.
25540 (aarch64_simd_emit_reg_reg_move): Likewise.
25541 (aarch64_simd_attr_length_rglist): Likewise.
25542 (aarch64_reverse_mask): Likewise.
25543 (aarch64_operands_ok_for_ldpstp): Likewise.
25544 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25545 (aarch64_gen_adjusted_ldpstp): Likewise.
25546 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25548 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25549 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25550 (arm_lookup_simd_builtin_type): Likewise.
25551 (arm_simd_builtin_type): Likewise.
25552 (arm_init_simd_builtin_types): Likewise.
25553 (arm_expand_builtin_args): Likewise.
25554 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25555 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25556 (ft32_setup_incoming_varargs): Likewise.
25557 (ft32_function_arg): Likewise.
25558 (ft32_function_arg_advance): Likewise.
25559 (ft32_pass_by_reference): Likewise.
25560 (ft32_arg_partial_bytes): Likewise.
25561 (ft32_valid_pointer_mode): Likewise.
25562 (ft32_addr_space_pointer_mode): Likewise.
25563 (ft32_addr_space_legitimate_address_p): Likewise.
25564 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25566 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25567 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25568 (iamcu_alignment): Likewise.
25569 (canonicalize_vector_int_perm): Likewise.
25570 (ix86_noce_conversion_profitable_p): Likewise.
25571 (ix86_mpx_bound_mode): Likewise.
25572 (ix86_operands_ok_for_move_multiple): Likewise.
25573 * config/microblaze/microblaze-protos.h
25574 (microblaze_expand_conditional_branch_reg): Likewise.
25575 * config/microblaze/microblaze.c
25576 (microblaze_expand_conditional_branch_reg): Likewise.
25577 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25579 (rs6000_reassociation_width): Likewise.
25580 (rs6000_invalid_binary_op): Likewise.
25581 (fusion_p9_p): Likewise.
25582 (emit_fusion_p9_load): Likewise.
25583 (emit_fusion_p9_store): Likewise.
25584 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25586 (riscv_hard_regno_mode_ok_p): Likewise.
25587 (riscv_address_insns): Likewise.
25588 (riscv_split_symbol): Likewise.
25589 (riscv_legitimize_move): Likewise.
25590 (riscv_function_value): Likewise.
25591 (riscv_hard_regno_nregs): Likewise.
25592 (riscv_expand_builtin): Likewise.
25593 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25594 (riscv_build_integer): Likewise.
25595 (riscv_split_integer): Likewise.
25596 (riscv_legitimate_constant_p): Likewise.
25597 (riscv_cannot_force_const_mem): Likewise.
25598 (riscv_regno_mode_ok_for_base_p): Likewise.
25599 (riscv_valid_base_register_p): Likewise.
25600 (riscv_valid_offset_p): Likewise.
25601 (riscv_valid_lo_sum_p): Likewise.
25602 (riscv_classify_address): Likewise.
25603 (riscv_legitimate_address_p): Likewise.
25604 (riscv_address_insns): Likewise.
25605 (riscv_load_store_insns): Likewise.
25606 (riscv_force_binary): Likewise.
25607 (riscv_split_symbol): Likewise.
25608 (riscv_force_address): Likewise.
25609 (riscv_legitimize_address): Likewise.
25610 (riscv_move_integer): Likewise.
25611 (riscv_legitimize_const_move): Likewise.
25612 (riscv_legitimize_move): Likewise.
25613 (riscv_address_cost): Likewise.
25614 (riscv_subword): Likewise.
25615 (riscv_output_move): Likewise.
25616 (riscv_canonicalize_int_order_test): Likewise.
25617 (riscv_emit_int_order_test): Likewise.
25618 (riscv_function_arg_boundary): Likewise.
25619 (riscv_pass_mode_in_fpr_p): Likewise.
25620 (riscv_pass_fpr_single): Likewise.
25621 (riscv_pass_fpr_pair): Likewise.
25622 (riscv_get_arg_info): Likewise.
25623 (riscv_function_arg): Likewise.
25624 (riscv_function_arg_advance): Likewise.
25625 (riscv_arg_partial_bytes): Likewise.
25626 (riscv_function_value): Likewise.
25627 (riscv_pass_by_reference): Likewise.
25628 (riscv_setup_incoming_varargs): Likewise.
25629 (riscv_print_operand): Likewise.
25630 (riscv_elf_select_rtx_section): Likewise.
25631 (riscv_save_restore_reg): Likewise.
25632 (riscv_for_each_saved_reg): Likewise.
25633 (riscv_register_move_cost): Likewise.
25634 (riscv_hard_regno_mode_ok_p): Likewise.
25635 (riscv_hard_regno_nregs): Likewise.
25636 (riscv_class_max_nregs): Likewise.
25637 (riscv_memory_move_cost): Likewise.
25638 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25639 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25640 (rl78_addr_space_address_mode): Likewise.
25641 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25643 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25644 (rs6000_reassociation_width): Likewise.
25645 (rs6000_invalid_binary_op): Likewise.
25646 (fusion_p9_p): Likewise.
25647 (emit_fusion_p9_load): Likewise.
25648 (emit_fusion_p9_store): Likewise.
25649 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25650 (ok_for_simple_move_operands): Likewise.
25651 (ok_for_simple_move_strict_operands): Likewise.
25652 (ok_for_simple_arith_logic_operands): Likewise.
25653 (visium_legitimize_reload_address): Likewise.
25654 (visium_select_cc_mode): Likewise.
25655 (output_cbranch): Likewise.
25656 (visium_split_double_move): Likewise.
25657 (visium_expand_copysign): Likewise.
25658 (visium_expand_int_cstore): Likewise.
25659 (visium_expand_fp_cstore): Likewise.
25660 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25661 (visium_function_arg): Likewise.
25662 (visium_function_arg_advance): Likewise.
25663 (visium_libcall_value): Likewise.
25664 (visium_setup_incoming_varargs): Likewise.
25665 (visium_legitimate_constant_p): Likewise.
25666 (visium_legitimate_address_p): Likewise.
25667 (visium_legitimize_address): Likewise.
25668 (visium_secondary_reload): Likewise.
25669 (visium_register_move_cost): Likewise.
25670 (visium_memory_move_cost): Likewise.
25671 (prepare_move_operands): Likewise.
25672 (ok_for_simple_move_operands): Likewise.
25673 (ok_for_simple_move_strict_operands): Likewise.
25674 (ok_for_simple_arith_logic_operands): Likewise.
25675 (visium_function_value_1): Likewise.
25676 (rtx_ok_for_offset_p): Likewise.
25677 (visium_legitimize_reload_address): Likewise.
25678 (visium_split_double_move): Likewise.
25679 (visium_expand_copysign): Likewise.
25680 (visium_expand_int_cstore): Likewise.
25681 (visium_expand_fp_cstore): Likewise.
25682 (visium_split_cstore): Likewise.
25683 (visium_select_cc_mode): Likewise.
25684 (visium_split_cbranch): Likewise.
25685 (output_cbranch): Likewise.
25686 (visium_print_operand_address): Likewise.
25687 * expmed.c (flip_storage_order): Likewise.
25688 * expmed.h (emit_cstore): Likewise.
25689 (flip_storage_order): Likewise.
25690 * genrecog.c (validate_pattern): Likewise.
25691 * hsa-gen.c (gen_hsa_addr): Likewise.
25692 * internal-fn.c (expand_arith_overflow): Likewise.
25693 * ira-color.c (allocno_copy_cost_saving): Likewise.
25694 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25695 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25696 (process_invariant_for_inheritance): Likewise.
25697 * lra-eliminations.c (move_plus_up): Likewise.
25698 * omp-low.c (lower_oacc_reductions): Likewise.
25699 * simplify-rtx.c (simplify_subreg): Likewise.
25700 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25701 (TARGET_CHKP_BOUND_MODE): Likewise..
25702 * targhooks.c (default_chkp_bound_mode): Likewise.
25703 (default_setup_incoming_vararg_bounds): Likewise.
25704 * targhooks.h (default_chkp_bound_mode): Likewise.
25705 (default_setup_incoming_vararg_bounds): Likewise.
25706 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25707 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25708 (have_whole_vector_shift): Likewise.
25709 * tree-vect-stmts.c (vectorizable_load): Likewise.
25710 * doc/tm.texi: Regenerate.
25712 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25714 Graceful degrade if Binutils PR21472 is not available.
25717 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25718 .rodata in flash test fails.
25719 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25720 * confgure: Regenerate.
25721 * config.in: Regenerate.
25722 * config/avr/avr.c (avr_asm_named_section)
25723 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25724 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25725 (avr_asm_init_sections): Same.
25727 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25729 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25730 (fma<VH:mode>4_intrinsic): Likewise.
25731 (*fmsub<VCVTF:mode>4): Likewise.
25732 (*fmsub<VH:mode>4_intrinsic): Likewise.
25734 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25737 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25738 Don't depend on "optimize > 0".
25739 (out_movhi_r_mr, out_movqi_mr_r): Same.
25740 (out_movhi_mr_r, out_movqi_r_mr): Same.
25741 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25742 io_address_operand on "optimize > 0".
25744 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25746 * tree-loop-distribution.c: Add general explanantion on the pass.
25747 (generate_loops_for_partition): Mark distributed loop.
25748 (pg_add_dependence_edges): New parameter. Handle alias data
25749 dependence specially and record it in the parameter if asked.
25750 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25751 (init_partition_graph_vertices, add_partition_graph_edge): New.
25752 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25753 (free_partition_graph_vdata, build_partition_graph): New.
25754 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25755 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25756 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25757 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25758 (version_for_distribution_p, finalize_partitions): New.
25759 (distribute_loop): Handle alias data dependence specially. Factor
25760 out loop fusion code as functions and call these functions.
25762 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25764 * tree-loop-distribution.c (classify_partition): New parameter and
25765 better handle reduction statement.
25766 (rdg_build_partitions): Revise comment.
25767 (distribute_loop): Compute statements in all partitions and pass it
25768 to classify_partition.
25770 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25772 * tree-loop-distribution.c (enum partition_type): New.
25773 (struct partition): New field type.
25774 (partition_merge_into): Add parameter. Update partition type.
25775 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25776 (build_rdg_partition_for_vertex): Compute partition type.
25777 (rdg_build_partitions): Dump partition type.
25778 (distribute_loop): Update calls to partition_merge_into.
25780 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25782 * tree-loop-distribution.c (struct ddr_hasher): New.
25783 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25785 (classify_partition): Call get_data_dependence.
25786 (pg_add_dependence_edges): Ditto.
25787 (distribute_loop): Release data dependence hash table.
25789 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25791 * tree-loop-distribution.c (ref_base_address): Delete.
25792 (similar_memory_accesses): Rename ...
25793 (share_memory_accesses): ... to this. Check if partitions access
25794 the same memory reference.
25795 (distribute_loop): Call share_memory_accesses.
25797 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25799 * tree-loop-distribution.c (struct partition): New field recording
25800 its data reference.
25801 (partition_alloc, partition_free): Init and release data refs.
25802 (partition_merge_into): Merge data refs.
25803 (build_rdg_partition_for_vertex): Collect data refs for partition.
25804 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25806 (distribute_loop): Remve data refs from vertice data of partition
25809 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25811 * tree-loop-distribution.c (params.h): Include header file.
25812 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25813 (datarefs_vec): New global var.
25814 (create_rdg_vertices): Use datarefs_vec directly.
25815 (free_rdg): Don't free data references.
25816 (build_rdg): Update use. Don't free data references.
25817 (distribute_loop): Compute global variable for data references.
25818 Bail out if there are too many data references.
25820 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25822 * tree-loop-distribution.c (loop_nest): New global var.
25823 (build_rdg): Use loop directly, rather than loop nest.
25824 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25826 (distribute_loop): Compute global variable loop nest. Update use.
25828 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25830 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25831 (partition_merge_into): New parameter. Dump reason for fusion.
25832 (distribute_loop): Update use of partition_merge_into.
25834 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25836 * tree-loop-distribution.c (bb_top_order_index): New.
25837 (bb_top_order_index_size, bb_top_order_cmp): New.
25838 (stmts_from_loop): Use topological order.
25839 (pass_loop_distribution::execute): Compute and release topological
25840 order for basic blocks.
25842 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25844 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25847 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25849 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25850 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25851 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25852 * internal-fn.def (LOOP_DIST_ALIAS): New.
25853 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25854 (fold_loop_internal_call): ... this.
25855 (vect_loop_dist_alias_call): New function.
25856 (set_uid_loop_bbs): Call fold_loop_internal_call.
25857 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25860 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25863 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25864 Require dead FLAGS_REG at the beginning of a peephole.
25866 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25869 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25870 arguments in the call to __builtin_ia32_sbb_u32.
25871 (_subborrow_u64): Swap _X and _Y arguments in the call to
25872 __builtin_ia32_sbb_u64.
25874 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25877 * tree-vrp.c (compare_assert_loc): Turn into a function template
25878 with stable template parameter. Only test if a->e is NULL,
25879 !a->e == !b->e has been verified already. Use e == NULL or
25880 e != NULL instead of e or ! e tests. If stable is true, don't use
25881 iterative_hash_expr, on the other side allow a or b or both NULL
25882 and sort the NULLs last.
25883 (process_assert_insertions): Sort using compare_assert_loc<false>
25884 instead of compare_assert_loc, later sort using
25885 compare_assert_loc<true> before calling process_assert_insertions_for
25886 in a loop. Use break instead of continue once seen NULL pointer.
25888 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25890 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25891 Cortex-R7 and Cortex-R8 processors.
25893 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25895 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25896 uninitialized while src is not.
25898 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25900 * common/config/arm/arm-common.c: Adjust include path for
25902 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25903 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25904 (arm-cpu-data.h): Likewise.
25905 (arm-cpu-cdata.h): Likewise.
25906 * config/arm/arm-cpu.h: Delete.
25907 * config/arm/arm-cpu-cdata.h: Delete.
25908 * config/arm/arm-cpu-data.h: Delete.
25910 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25912 * config/arm/arm-cpus.in (cortex-a55): New.
25913 (cortex-a75): Likewise.
25914 (cortex-a75.cortex-a55): Likewise.
25915 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25917 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25918 * config/arm/arm-cpu-cdata.h: Regenerate.
25919 * config/arm/arm-cpu-data.h: Regenerate.
25920 * config/arm/arm-cpu.h: Regenerate.
25921 * config/arm/arm-tables.opt: Regenerate.
25922 * config/arm/arm-tune.md: Regenerate.
25924 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25926 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25928 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25930 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25933 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25935 PR tree-optimization/81292
25936 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25937 full_string_p, also call adjust_related_strinfos if the adjustment
25938 is simple, otherwise invalidate related strinfos.
25940 2017-07-04 Martin Liska <mliska@suse.cz>
25943 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25944 newly created variable as DECL_IGNORED_P.
25946 2017-07-04 Martin Liska <mliska@suse.cz>
25949 * ipa-inline.c (inline_small_functions):
25950 Use xstrdup_for_dump.
25952 2017-07-04 Tom de Vries <tom@codesourcery.com>
25954 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25956 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25959 * config/darwin.c (darwin_function_switched_text_sections):
25962 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25964 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25966 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25968 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25970 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25972 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25973 min_profitable_iters, and th as inclusive lower bounds.
25974 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25975 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25976 for min_profitable_iters and min_profitable_estimate.
25977 (vect_transform_loop): Treat th as an inclusive lower bound.
25978 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25980 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25983 * config/darwin.c (darwin_function_switched_text_sections):
25984 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25985 in two pieces, and suppress the use of buf.
25987 2017-07-03 Nathan Sidwell <nathan@acm.org>
25989 * hash-table.h (hash_table_mod1): Fix indentation.
25991 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25993 PR middle-end/81290
25994 * predict.c (force_edge_cold): Be more careful about propagation
25996 * profile-count.h (profile_probability::guessed,
25997 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25999 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
26001 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
26003 * doc/invoke.texi (rcpc architecture extension): Document it.
26005 2017-07-03 Richard Biener <rguenther@suse.de>
26007 PR tree-optimization/60510
26008 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
26009 the scalar reduction PHI and use it.
26010 (vectorizable_reduction): Properly guard the single_defuse_cycle
26011 path for non-SLP reduction chains where we cannot use it.
26012 Rework reduc_def/index and vector type deduction. Rework
26013 vector operand gathering during reduction op code-gen.
26014 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
26015 chains dissolve the chain and leave it to non-SLP reduction
26018 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26020 * tree-data-ref.h (dr_alignment): Declare.
26021 * tree-data-ref.c (dr_alignment): New function.
26022 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
26023 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
26025 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
26027 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26029 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
26030 and base_misalignment fields.
26031 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
26032 * tree-data-ref.c: Include builtins.h.
26033 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
26034 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
26035 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
26036 * tree-vect-data-refs.c: Include tree-cfg.h.
26037 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
26038 fields instead of calculating an alignment here.
26039 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
26040 innermost_loop_behavior fields.
26042 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26044 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
26046 (DR_STEP_ALIGNMENT): New macro.
26047 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
26048 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
26049 (create_data_ref): Print it.
26050 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
26051 to tell whether the step preserves vector (mis)alignment.
26052 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26053 Move the check for an integer step and generalise to all INTEGER_CST.
26054 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
26055 Print the outer step alignment.
26057 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26059 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
26060 with offset_alignment.
26061 (DR_ALIGNED_TO): Delete.
26062 (DR_OFFSET_ALIGNMENT): New macro.
26063 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26064 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26065 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26066 (create_data_ref): Likewise.
26067 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26068 (vect_analyze_data_refs): Likewise.
26069 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26070 creating dummy innermost behavior.
26072 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26074 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26075 with a "innermost_loop_behavior *" and refeence tree.
26076 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26077 (create_data_ref): Update call accordingly.
26078 * tree-predcom.c (find_looparound_phi): Likewise.
26080 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26082 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26083 fields with dr_wrt_vec_loop.
26084 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26085 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26086 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26087 (vect_dr_behavior): New function.
26088 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26089 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26090 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26091 track whether the step preserves the misalignment.
26092 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26093 Use vect_dr_behavior.
26094 (vect_setup_realignment): Update call accordingly.
26095 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26096 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26097 call to vect_create_addr_base_for_vector_ref.
26098 (vect_create_cond_for_align_checks): Likewise.
26099 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26100 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26101 (vect_recog_mask_conversion_pattern): Likewise.
26102 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26103 (new_stmt_vec_info): Remove redundant zeroing.
26105 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26107 * common/config/arm/arm-common.c (arm_be8_option): New function.
26108 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26109 (ISA_ARMv6): Add isa_bit_be8.
26110 * config/arm/arm.h (arm_be8_option): Add prototype.
26111 (BE8_SPEC_FUNCTION): New define.
26112 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26113 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26114 (mlittle-endian): Similarly.
26115 (mbe8, mbe32): New options.
26116 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26117 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26119 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26121 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26123 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26125 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26126 (cleanup_tree_cfg_bb): Use it.
26127 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26129 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26131 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26134 * loop-doloop.c (add_test): Update profile.
26136 2017-07-03 Martin Liska <mliska@suse.cz>
26139 * sanopt.c (rewrite_usage_of_param): New function.
26140 (sanitize_rewrite_addressable_params): Likewise.
26141 (pass_sanopt::execute): Call rewrite_usage_of_param.
26143 2017-07-03 Richard Biener <rguenther@suse.de>
26145 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26146 back to using VIEW_CONVERT_EXPR.
26148 2017-07-03 Martin Liska <mliska@suse.cz>
26151 * doc/extend.texi: Document when a resolver function is
26152 generated for target_clones.
26154 2017-07-03 Martin Liska <mliska@suse.cz>
26156 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26157 and shadow memory of auto variables which are subject of
26158 use-after-scope sanitization.
26159 (asan_expand_mark_ifn): Add do set only when is_poison.
26161 2016-07-03 Richard Biener <rguenther@suse.de>
26163 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26165 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26166 (vectorizable_reduction): Perform reduction PHI creation when
26167 visiting a reduction PHI and adjust and simplify code generation
26168 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26169 (vect_transform_loop): Visit reduction PHIs.
26170 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26171 defs into the SLP tree.
26172 (vect_build_slp_tree): Reduction defs terminate the recursion.
26173 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26175 (vect_get_vec_defs_for_stmt_copy): Export.
26176 (vect_get_vec_defs): Likewise.
26177 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26179 (vect_get_vec_defs_for_stmt_copy): Declare.
26180 (vect_get_vec_defs): Likewise.
26182 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26184 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26185 parameter with a "loop" parameter and use it instead of the
26186 loop containing DR_STMT. Don't check simple_iv when doing
26187 BB analysis. Describe the two analysis modes in the comment.
26189 2017-07-03 Tom de Vries <tom@codesourcery.com>
26191 PR tree-optimization/69468
26192 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26193 (find_same_succ_bb): Handle ignore_edge_flags.
26195 2017-07-03 Tom de Vries <tom@codesourcery.com>
26197 PR tree-optimization/81192
26198 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26200 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26202 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26204 2017-07-03 Tom de Vries <tom@codesourcery.com>
26206 PR tree-optimization/81192
26207 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26208 BB_SAME_SUCC (bb) == NULL.
26210 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26212 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26215 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26217 * dumpfile.c: Include profile-count.h
26218 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26220 (insert_cond_bb): Update profile.
26221 * tree-cfg.h (insert_cond_bb): Update prototype.
26222 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26223 * tree-dump.c: Do not include tree-cfg.
26225 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26227 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26229 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26231 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26234 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26236 * tree-complex.c (expand_complex_div_wide): update profile.
26238 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26239 Alan Hayward <alan.hayward@arm.com>
26240 David Sherwood <david.sherwood@arm.com>
26242 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26243 (CORETYPES_H): New define.
26244 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26245 (insn-modes-inline.h, s-modes-inline-h): New rules.
26246 (generated_files): Add insn-modes-inline.h.
26247 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26248 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26249 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26250 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26251 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26252 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26253 (build/gencodes.o, build/genconditions.o): Likewise.
26254 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26255 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26256 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26257 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26258 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26259 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26260 * coretypes.h: Include everything up to real.h for generators.
26261 Include insn-modes.h first. Include wide-int-print.h after
26262 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26263 * machmode.h: Don't include insn-modes.h here.
26264 * function-tests.c: Remove includes of signop.h, machmode.h,
26265 double-int.h and wide-int.h.
26267 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26269 * optc-save-gen.awk: Likewise.
26270 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26271 * godump.c: Remove include of wide-int-print.h.
26272 * pretty-print.h: Likewise.
26273 * wide-int-print.cc: Likewise.
26274 * wide-int.cc: Likewise.
26275 * hash-map-tests.c: Remove include of signop.h.
26276 * hash-set-tests.c: Likewise.
26277 * rtl-tests.c: Likewise.
26278 * mkconfig.sh: Remove include of machmode.h.
26279 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26281 (emit_insn_modes_inline_h): ...this new function. Emit the code
26282 into an insn-modes-inline.h header file, adding appropriate
26283 include guards and end comments.
26284 (emit_insn_modes_c_header): Remove include of machmode.h.
26285 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26287 (main): Handle -i flag and call emit_insn_modes_inline_h when
26290 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26292 * tree-ssa-strlen.c (strinfo): Rename the length field to
26293 nonzero_chars. Add a full_string_p field.
26294 (compare_nonzero_chars, zero_length_string_p): New functions.
26295 (get_addr_stridx): Add an offset_out parameter.
26296 Use compare_nonzero_chars.
26297 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26298 (new_strinfo): Update after above changes to strinfo.
26299 (set_endptr_and_length): Set full_string_p.
26300 (get_string_length): Update after above changes to strinfo.
26301 (unshare_strinfo): Update call to new_strinfo.
26302 (maybe_invalidate): Likewise.
26303 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26304 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26305 as a uhwi instead of an shwi. Update after above changes to
26306 strinfo and new_strinfo.
26307 (zero_length_string): Assert that chainsi contains full strings.
26308 Use zero_length_string_p. Update call to new_strinfo.
26309 (adjust_related_strinfos): Update after above changes to strinfo.
26310 Copy full_string_p from origsi.
26311 (adjust_last_stmt): Use zero_length_string_p.
26312 (handle_builtin_strlen): Update after above changes to strinfo and
26313 new_strinfo. Install the lhs as the string length if the previous
26314 entry didn't describe a full string.
26315 (handle_builtin_strchr): Update after above changes to strinfo
26317 (handle_builtin_strcpy): Likewise.
26318 (handle_builtin_strcat): Likewise.
26319 (handle_builtin_malloc): Likewise.
26320 (handle_pointer_plus): Likewise.
26321 (handle_builtin_memcpy): Likewise. Track nonzero characters
26322 that aren't necessarily followed by a nul terminator.
26323 (handle_char_store): Likewise.
26325 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26327 PR tree-optimization/80769
26328 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26329 for malloc and calloc. Document the new invariant that all related
26330 strinfos have delayed lengths or none do.
26331 (verify_related_strinfos): Move earlier in file.
26332 (set_endptr_and_length): New function, split out from...
26333 (get_string_length): ...here. Also set the lengths of related
26335 (zero_length_string): Assert that chainsi has known (rather than
26337 (adjust_related_strinfos): Likewise.
26339 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26341 PR tree-optimization/81136
26342 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26343 assert that two references with the same misalignment have the same
26344 compile-time misalignment if those compile-time misalignments
26347 2017-07-01 Andi Kleen <ak@linux.intel.com>
26349 * print-tree.c (print_node): Print all attributes.
26351 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26353 * cfg.c (scale_bbs_frequencies): New function.
26354 * cfg.h (scale_bbs_frequencies): Declare it.
26355 * cfgloopanal.c (single_likely_exit): Cleanup.
26356 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26358 (scale_loop_profile): Likewise.
26359 (loop_version): Likewise.
26360 (create_empty_loop_on_edge): Update.
26361 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26362 scale_loop_frequencies, scale_loop_profile, loopify,
26363 loop_version): Update prototypes.
26364 * modulo-sched.c (sms_schedule): Update.
26365 * predict.c (unlikely_executed_edge_p): Also check probability.
26366 (probably_never_executed_edge_p): Fix typo.
26367 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26368 * tree-parloops.c (gen_parallel_loop): Update.
26369 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26370 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26371 * tree-ssa-loop-split.c (split_loop): Update.
26372 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26373 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26374 (vect_loop_versioning): Update.
26375 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26377 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26379 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26381 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26383 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26384 to keep profile consistent.
26386 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26388 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26389 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26390 * profile-count.h (max_safe_multiplier): Make unsigned.
26391 (profile_count::guessed_zero): New.
26393 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26395 * bb-reorder.c (fix_up_crossing_landing_pad,
26396 fix_crossing_conditional_branches): Use make_single_succ_edge
26397 to keep profile consistent.
26399 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26401 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26404 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26407 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26408 the right scopes, make sure cond_jump isn't preserved between multiple
26409 iterations. Search for fallthru edge whenever there are 3+ edges and
26410 use find_fallthru_edge for it.
26412 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26414 Patch by Alexander Monakov <amonakov@ispras.ru>
26415 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26416 probabilities consistently.
26418 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26420 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26421 emit_cmp_and_jump_insns.
26423 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26426 * tree-inline.c (expand_call_inline): Combine profile statuses.
26428 2017-06-30 Andrew Pinski <apinski@cavium.com>
26430 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26431 fold_stmt returned true.
26433 2017-06-30 Nathan Sidwell <nathan@acm.org>
26435 * ggc.h (empty_string): Delete.
26436 * cfgexpand.c (expand_asm_stmt): Use plain "".
26437 * optabs.c (expand_asm_memory_barrier): Likewise.
26438 * stringpool.c (empty_string): Delete.
26439 (digit_vector, digit_string): Delete.
26440 (ggc_alloc_string): Use plain "", don't optimize single digit
26441 strings. Use ggc_alloc_atomic.
26443 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26445 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26446 comparison set and one other set, use the cost of the non-comparison
26449 2017-06-30 Nathan Sidwell <nathan@acm.org>
26451 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26454 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26456 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26457 loops. Remove now unneeded calls to gimple_switch_set_label() that
26458 just set removed labels to NULL_TREE.
26460 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26462 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26463 (set_range_info): ...here. Only call set_range_info_raw if domain
26465 (set_nonzero_bits): Call set_range_info_raw.
26466 * tree-ssanames.h (set_range_info_raw): New.
26468 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26471 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26472 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26473 of nonimmediate_operand and <store_mask_constraint> instead of m
26474 for the input operand. For V8FI iterator, always split if input
26475 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26476 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26477 <store_mask_predicate> instead of register_operand and
26478 <store_mask_constraint> instead of v for the input operand. Make
26479 sure both operands aren't MEMs for if not <mask_applied>.
26481 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26483 * lto-wrapper.c (copy_file) Close both file descriptors before
26486 2017-06-30 Martin Liska <mliska@suse.cz>
26489 * multiple_target.c (create_dispatcher_calls): Make ifunc
26490 also for function that don't have calls or are not referenced.
26492 2017-06-30 Richard Biener <rguenther@suse.de>
26494 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26495 analyze the first scalar stmt. Move vector type computation
26496 for the BB case here from ...
26497 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26498 live operation processing in the SLP case properly.
26500 2017-06-30 Richard Biener <rguenther@suse.de>
26502 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26504 2017-06-30 Martin Liska <mliska@suse.cz>
26507 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26508 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26510 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26512 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26515 2017-06-30 Richard Biener <rguenther@suse.de>
26517 PR tree-optimization/81249
26518 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26519 condition reduction result to original scalar type.
26521 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26523 * profile-count.h (enum profile_quality): Fix typos and whitespace
26526 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26528 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26529 type for branch probabilities.
26531 2017-06-29 Julian Brown <julian@codesourcery.com>
26532 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26534 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26535 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26536 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26537 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26539 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26541 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26542 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26543 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26544 CC usage from generic code to here.
26545 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26546 CC usage into the target macros.
26548 2017-06-29 Maya Rashish <coypu@sdf.org>
26550 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26553 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26555 * arm/arm-builtins.c: Include profile-count.h
26556 * except.c (sjlj_emit_function_enter): Use
26557 profile_probability::unlikely.
26559 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26561 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26562 and tocrel_offset be pointer args rather than implicitly using
26564 (legitimate_constant_pool_address_p, rs6000_emit_move,
26565 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26566 tocrel_offset and use in toc_relative_expr_p call.
26567 (print_operand, print_operand_address): Use static tocrel_base_oac
26568 and tocrel_offset_oac.
26569 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26572 2017-06-29 Maya Rashish <coypu@sdf.org>
26574 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26576 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26578 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26579 objects, take into account only the alignment of 'op0' and 'mode1' if
26582 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26584 * ccmp.c (ccmp_tree_comparison_p): New function.
26585 (ccmp_candidate_p): Update to use above function.
26586 (get_compare_parts): New function.
26587 (expand_ccmp_next): Update to use new functions.
26588 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26590 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26591 take mode as argument.
26592 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26593 * expr.c (expand_expr_real_1): Pass mode as argument.
26595 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26597 * combine.c (combine_instructions): Print insns to dump_file, together
26600 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26602 * asan.c (asan_emit_stack_protection): Update.
26603 (create_cond_insert_point): Update.
26604 * auto-profile.c (afdo_propagate_circuit): Update.
26605 * basic-block.h (struct edge_def): Turn probability to
26606 profile_probability.
26607 (EDGE_FREQUENCY): Update.
26608 * bb-reorder.c (find_traces_1_round): Update.
26609 (better_edge_p): Update.
26610 (sanitize_hot_paths): Update.
26611 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26612 (make_single_succ_edge): Update.
26613 (check_bb_profile): Update.
26614 (dump_edge_info): Update.
26615 (update_bb_profile_for_threading): Update.
26616 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26617 probabilitycount to 0.
26618 * cfgbuild.c (compute_outgoing_frequencies): Update.
26619 * cfgcleanup.c (try_forward_edges): Update.
26620 (outgoing_edges_match): Update.
26621 (try_crossjump_to_edge): Update.
26622 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26623 (expand_gimple_tailcall): Update.
26624 (construct_init_block): Use make_single_succ_edge.
26625 (construct_exit_block): Use make_single_succ_edge.
26626 * cfghooks.c (verify_flow_info): Update.
26627 (redirect_edge_succ_nodup): Update.
26628 (split_edge): Update.
26629 (account_profile_record): Update.
26630 * cfgloopanal.c (single_likely_exit): Update.
26631 * cfgloopmanip.c (scale_loop_profile): Update.
26632 (set_zero_probability): Remove.
26633 (duplicate_loop_to_header_edge): Update.
26634 * cfgloopmanip.h (loop_version): Update prototype.
26635 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26636 (force_nonfallthru_and_redirect): Update.
26637 (update_br_prob_note): Update.
26638 (rtl_verify_edges): Update.
26639 (purge_dead_edges): Update.
26640 (rtl_lv_add_condition_to_bb): Update.
26641 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26642 * cgraphunit.c (init_lowered_empty_function): Update.
26643 (cgraph_node::expand_thunk): Update.
26644 * cilk-common.c: Include profile-count.h
26645 * dojump.c (inv): Remove.
26646 (jumpifnot): Update.
26647 (jumpifnot_1): Update.
26648 (do_jump_1): Update.
26650 (do_jump_by_parts_greater_rtx): Update.
26651 (do_compare_rtx_and_jump): Update.
26652 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26653 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26654 * dwarf2cfi.c: Include profile-count.h
26655 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26656 (sjlj_emit_dispatch_table): Likewise.
26657 * explow.c: Include profile-count.h
26658 * expmed.c (emit_store_flag_force): Update.
26659 (do_cmp_and_jump): Update.
26660 * expr.c (compare_by_pieces_d::generate): Update.
26661 (compare_by_pieces_d::finish_mode): Update.
26662 (emit_block_move_via_loop): Update.
26663 (store_expr_with_bounds): Update.
26664 (store_constructor): Update.
26665 (expand_expr_real_2): Update.
26666 (expand_expr_real_1): Update.
26667 * expr.h (try_casesi, try_tablejump): Update prototypes.
26668 * gimple-pretty-print.c (dump_probability): Update.
26669 (dump_profile): New.
26670 (dump_gimple_label): Update.
26671 (dump_gimple_bb_header): Update.
26672 * graph.c (draw_cfg_node_succ_edges): Update.
26673 * hsa-gen.c (convert_switch_statements): Update.
26674 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26675 (find_if_case_1): Update.
26676 (find_if_case_2): Update.
26677 * internal-fn.c (expand_arith_overflow_result_store): Update.
26678 (expand_addsub_overflow): Update.
26679 (expand_neg_overflow): Update.
26680 (expand_mul_overflow): Update.
26681 (expand_vector_ubsan_overflow): Update.
26682 * ipa-cp.c (good_cloning_opportunity_p): Update.
26683 * ipa-split.c (split_function): Use make_single_succ_edge.
26684 * ipa-utils.c (ipa_merge_profiles): Update.
26685 * loop-doloop.c (add_test): Update.
26686 (doloop_modify): Update.
26687 * loop-unroll.c (compare_and_jump_seq): Update.
26688 (unroll_loop_runtime_iterations): Update.
26689 * lra-constraints.c (lra_inheritance): Update.
26690 * lto-streamer-in.c (input_cfg): Update.
26691 * lto-streamer-out.c (output_cfg): Update.
26692 * mcf.c (adjust_cfg_counts): Update.
26693 * modulo-sched.c (sms_schedule): Update.
26694 * omp-expand.c (expand_omp_for_init_counts): Update.
26695 (extract_omp_for_update_vars): Update.
26696 (expand_omp_ordered_sink): Update.
26697 (expand_omp_for_ordered_loops): Update.
26698 (expand_omp_for_generic): Update.
26699 (expand_omp_for_static_nochunk): Update.
26700 (expand_omp_for_static_chunk): Update.
26701 (expand_cilk_for): Update.
26702 (expand_omp_simd): Update.
26703 (expand_omp_taskloop_for_outer): Update.
26704 (expand_omp_taskloop_for_inner): Update.
26705 * omp-simd-clone.c (simd_clone_adjust): Update.
26706 * optabs.c (expand_doubleword_shift): Update.
26707 (expand_abs): Update.
26708 (emit_cmp_and_jump_insn_1): Update.
26709 (expand_compare_and_swap_loop): Update.
26710 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26711 * predict.c (predictable_edge_p): Update.
26712 (edge_probability_reliable_p): Update.
26713 (set_even_probabilities): Update.
26714 (combine_predictions_for_insn): Update.
26715 (combine_predictions_for_bb): Update.
26716 (propagate_freq): Update.
26717 (estimate_bb_frequencies): Update.
26718 (force_edge_cold): Update.
26719 * profile-count.c (profile_count::dump): Add missing space into dump.
26720 (profile_count::debug): Add newline.
26721 (profile_count::differs_from_p): Explicitly convert to unsigned.
26722 (profile_count::stream_in): Update.
26723 (profile_probability::dump): New member function.
26724 (profile_probability::debug): New member function.
26725 (profile_probability::differs_from_p): New member function.
26726 (profile_probability::differs_lot_from_p): New member function.
26727 (profile_probability::stream_in): New member function.
26728 (profile_probability::stream_out): New member function.
26729 * profile-count.h (profile_count_quality): Rename to ...
26730 (profile_quality): ... this one.
26731 (profile_probability): New.
26732 (profile_count): Update.
26733 * profile.c (compute_branch_probabilities): Update.
26734 * recog.c (peep2_attempt): Update.
26735 * sched-ebb.c (schedule_ebbs): Update.
26736 * sched-rgn.c (find_single_block_region): Update.
26737 (compute_dom_prob_ps): Update.
26738 (schedule_region): Update.
26739 * sel-sched-ir.c (compute_succs_info): Update.
26740 * stmt.c (struct case_node): Update.
26741 (do_jump_if_equal): Update.
26742 (get_outgoing_edge_probs): Update.
26743 (conditional_probability): Update.
26744 (emit_case_dispatch_table): Update.
26745 (expand_case): Update.
26746 (expand_sjlj_dispatch_table): Update.
26747 (emit_case_nodes): Update.
26748 * targhooks.c: Update.
26749 * tracer.c (better_p): Update.
26750 (find_best_successor): Update.
26751 * trans-mem.c (expand_transaction): Update.
26752 * tree-call-cdce.c: Update.
26753 * tree-cfg.c (gimple_split_edge): Upate.
26754 (move_sese_region_to_fn): Upate.
26755 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26756 * tree-eh.c (lower_resx): Upate.
26757 (cleanup_empty_eh_move_lp): Upate.
26758 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26759 * tree-inline.c (copy_edges_for_bb): Update.
26760 (copy_cfg_body): Update.
26761 * tree-parloops.c (gen_parallel_loop): Update.
26762 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26763 (gimple_gen_time_profiler): Update.
26764 * tree-ssa-dce.c (remove_dead_stmt): Update.
26765 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26766 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26767 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26768 (unloop_loops): Update.
26769 (try_peel_loop): Update.
26770 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26771 * tree-ssa-loop-split.c (connect_loops): Update.
26772 (split_loop): Update.
26773 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26774 (hoist_guard): Update.
26775 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26776 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26777 (value_replacement): Update.
26778 * tree-ssa-reassoc.c (branch_fixup): Update.
26779 * tree-ssa-tail-merge.c (replace_block_by): Update.
26780 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26781 (create_edge_and_update_destination_phis): Update.
26782 (compute_path_counts): Update.
26783 (recompute_probabilities): Update.
26784 (update_joiner_offpath_counts): Update.
26785 (freqs_to_counts_path): Update.
26786 (duplicate_thread_path): Update.
26787 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26788 (struct switch_conv_info): Update.
26789 (gen_inbound_check): Update.
26790 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26791 (vect_do_peeling): Update.
26792 (vect_loop_versioning): Update.
26793 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26794 (optimize_mask_stores): Update.
26795 * ubsan.c (ubsan_expand_null_ifn): Update.
26796 * value-prof.c (gimple_divmod_fixed_value): Update.
26797 (gimple_divmod_fixed_value_transform): Update.
26798 (gimple_mod_pow2): Update.
26799 (gimple_mod_pow2_value_transform): Update.
26800 (gimple_mod_subtract): Update.
26801 (gimple_mod_subtract_transform): Update.
26802 (gimple_ic): Update.
26803 (gimple_stringop_fixed_value): Update.
26804 (gimple_stringops_transform): Update.
26805 * value-prof.h: Update.
26807 2017-06-29 Carl Love <cel@us.ibm.com>
26809 * config/rs6000/rs6000-c.c: Add support for built-in functions
26810 vector signed int vec_signed (vector float);
26811 vector signed long long vec_signed (vector double);
26812 vector signed int vec_signed2 (vector double, vector double);
26813 vector signed int vec_signede (vector double);
26814 vector signed int vec_signedo (vector double);
26815 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26816 instruction generator.
26817 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26818 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26819 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26821 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26822 vunsignede_v2df): Add define_expands.
26823 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26824 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26825 VEC_UNSIGNEDO): Add definitions.
26826 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26827 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26828 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26829 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26830 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26831 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26832 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26833 vec_unsignede, vec_unsignedo): Add builtin defines.
26834 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26836 * doc/extend.texi: Update the built-in documentation file for the
26837 new built-in functions.
26839 2017-06-29 Richard Biener <rguenther@suse.de>
26841 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26842 reduction chains to LOOP_VINFO_REDUCTIONS.
26843 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26844 SLP reductions after processing reduction chains.
26846 2017-06-29 Nathan Sidwell <nathan@acm.org>
26848 * builtins.c (fold_builtin_FUNCTION): Use
26849 lang_hooks.decl_printable_name.
26851 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26853 PR middle-end/81194
26854 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26855 with only one label.
26856 * stmt.c (expand_case): Assert NCASES is greater than one.
26858 2017-06-29 Richard Biener <rguenther@suse.de>
26860 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26862 (group_case_labels): Likewise.
26863 (find_taken_edge): Push sanity checking on val to workers...
26864 (find_taken_edge_cond_expr): ... here
26865 (find_taken_edge_switch_expr): ... and here, handle cases
26866 with just a default label.
26867 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26868 (group_case_labels): Likewise.
26869 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26870 group_case_labels does anything cleanup the CFG again.
26872 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26874 PR tree-optimization/81196
26875 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26876 exit condition comparing two IVs.
26878 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26880 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26881 profile to the dummy entry at the end of the list of architectures.
26882 * config/arm/arm-cpu-cdata.h: Regenerated.
26884 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26885 Michael Collison <michael.collison@arm.com>
26888 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26890 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26891 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26892 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26893 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26894 of shift when the shift amount is masked with constant equal to
26895 the size of the mode.
26896 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26899 2017-06-29 Martin Liska <mliska@suse.cz>
26901 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26903 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26906 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26909 (edge_growth_cache_entry::edge_growth_cache_entry): New
26911 (reset_edge_growth_cache): Update to use constructor.
26913 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26915 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26916 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26917 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26919 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26921 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26922 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26924 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26926 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26927 (*-linux-uclibc*): Add t-uclibc tmake_file.
26928 * config/t-musl: New.
26929 * config/t-uclibc: New.
26931 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26933 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26935 (gen_comm_data): Emit architectural setting of arch_prof.
26936 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26938 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26939 (armv8-m.base, armv8-m.main): Likewise.
26940 * arm-protos.h (arm_build_target): Add profile field.
26941 (arch_option): Likewise.
26942 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26944 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26945 arm_active_target.profile.
26947 2017-06-28 Richard Biener <rguenther@suse.de>
26949 PR middle-end/81227
26950 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26951 TYPE_OVERFLOW_WRAPS.
26952 * match.pd (negate_expr_p): Likewise.
26953 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26954 fold_build2, not fold_binary.
26956 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26958 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26959 Convert memory address to Pmode.
26960 (aarch64_print_operand): Assert MEM operands are always Pmode.
26962 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26965 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26966 Remove redundant if.
26967 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26968 * config/arm/aarch-common-protos.h
26969 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26970 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26972 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26975 * multiple_target.c (create_dispatcher_calls): Set the default
26976 clone to be static, not public.
26978 2017-06-28 Richard Biener <rguenther@suse.de>
26980 * tree-vect-loop.c (vectorizable_reduction): Move special
26981 cond reduction IV var creation ...
26982 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26983 parameter. Use STMT_VINFO_VECTYPE.
26984 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26987 2017-06-28 Martin Liska <mliska@suse.cz>
26990 * ipa-visibility.c (non_local_p): Handle visibility.
26992 2017-06-28 Martin Liska <mliska@suse.cz>
26995 * common.opt: Add IntegerRange to various options.
26996 * opt-functions.awk (integer_range_info): New function.
26997 * optc-gen.awk: Add integer_range_info to cl_options struct.
26998 * opts-common.c (decode_cmdline_option): Handle
26999 CL_ERR_INT_RANGE_ARG.
27000 (cmdline_handle_error): Likewise.
27001 * opts.c (print_filtered_help): Show valid interval in
27002 when --help is provided.
27003 * opts.h (struct cl_option): Add range_min and range_max fields.
27004 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
27006 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
27008 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
27009 (x * C EQ/NE y * C): New transformation.
27011 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
27013 * genmultilib (combination_space): Accept '+' in option names.
27015 2017-06-28 Martin Liska <mliska@suse.cz>
27018 * asan.c (instrument_derefs): Bail out inner references
27019 that are hard register variables.
27021 2017-06-28 Jakub Jelinek <jakub@redhat.com>
27024 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
27025 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
27027 2017-06-28 Richard Biener <rguenther@suse.de>
27029 * tree-vectorizer.h (vect_get_vec_defs): Remove.
27030 (vect_get_slp_defs): Adjust.
27031 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
27033 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
27035 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
27036 get_initial_defs_for_reduction instead of vect_get_vec_defs.
27037 (vectorizable_reduction): Adjust.
27038 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
27040 (vect_get_slp_defs): Likewise.
27041 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
27042 (vectorizable_bswap): Adjust.
27043 (vectorizable_call): Likewise.
27044 (vectorizable_conversion): Likewise.
27045 (vectorizable_assignment): Likewise.
27046 (vectorizable_shift): Likewise.
27047 (vectorizable_operation): Likewise.
27048 (vectorizable_store): Likewise.
27049 (vectorizable_condition): Likewise.
27050 (vectorizable_comparison): Likewise.
27052 2017-06-28 Michael Collison <michael.collison@arm.com>
27055 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
27057 (arm_gen_movmemqi): Removed unused variable 'i'.
27058 Convert 'for' loop into 'while' loop.
27059 (arm_expand_prologue): Remove last unnecessary set of insn.
27060 (thumb_pop): Remove unused variable 'pushed_words'.
27061 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27063 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27065 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27066 * config/s390/s390-protos.h: Add prototype of
27067 s390_rel_address_ok_p.
27068 * config/s390/s390.c (s390_got_symbol): New function.
27069 (s390_rel_address_ok_p): New function.
27070 (legitimize_pic_address): Use s390_rel_address_ok_p.
27071 (s390_load_got): Use s390_got_symbol.
27072 (s390_option_override): Issue error if
27073 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27074 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27076 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27078 2017-06-27 Andrew Pinski <apinski@cavium.com>
27080 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27081 (X * copysign (1.0, X)): New pattern.
27082 (X * copysign (1.0, -X)): New pattern.
27083 (copysign (-1.0, CST)): New pattern.
27085 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27087 * genmultilib (combination_space): Remove variable.
27088 Validate reuse rules against regular expression for any sequence
27089 of multilib options in any order.
27091 2017-06-27 Michael Collison <michael.collison@arm.com>
27093 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27094 call aarch64_split_simd_combine.
27095 * (aarch64_combine_internal<mode>): Delete pattern.
27096 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27097 Allow register and subreg operands.
27099 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27101 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27102 specific need, just fallback on defaults.
27103 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27105 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27106 Olivier Hainque <hainque@adacore.com>
27108 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27110 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27111 targets. Pick a default if no particular attempt applied.
27112 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27115 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27117 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27118 (x86_64-wrs-vxworks7): Likewise.
27120 2017-06-27 Marek Polacek <polacek@redhat.com>
27123 * ubsan.c (instrument_null): Check get_base_address's result for null.
27125 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27127 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27129 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27131 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27132 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27133 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27134 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27135 New function types.
27136 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27137 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27138 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27139 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27140 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27141 BUILT_IN_FEUPDATEENV): New builtins.
27142 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27143 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27144 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27145 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27147 (builtin_structptr_types): Adjust size.
27148 * tree.c (builtin_structptr_types): Add four entries.
27150 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27151 Olivier Hainque <hainque@adacore.com>
27153 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27154 (TLS_SYM): New local macro, forcing reference to __tls__ on
27155 link command lines for VxWorks 7 RTPs, triggering initialization
27157 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27158 OS features TLS support, true for RTPs on VxWorks 7.
27159 * config/vxworks.c (vxworks_override_options): Setup emutls
27162 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27164 * predict.c (test_prediction_value_range): Use -1U instead of -1
27165 to avoid narrowing conversion warning.
27166 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27167 to avoid narrowing conversion warning.
27168 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27170 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27172 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27174 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27175 64bit configurations.
27176 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27177 (SIZE_TYPE): Likewise.
27178 * config/vxworks.c (vxworks_emutls_var_fields): Use
27179 long_unsigned_type_node instead of unsigned_type_node as the offset
27180 field type, which is "pointer" mode in emutls.c.
27182 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27185 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27187 PR middle-end/81207
27188 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27189 gimple_vuse copying separately from gimple_vdef copying.
27191 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27193 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27196 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27197 Olivier Hainque <hainque@adacore.com>
27199 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27200 for all vxworks7 targets.
27201 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27202 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27203 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27204 variations for VX6/VX7 and 32/64bits later on in ...
27205 (VXWORKS_LIB_SPEC): Leverage new macros.
27206 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27207 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27209 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27211 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27212 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27214 2017-06-26 Carl Love <cel@us.ibm.com>
27216 * config/rs6000/rs6000-c.c: Add support for built-in functions
27217 vector bool char vec_reve (vector bool char);
27218 vector signed char vec_reve (vector signed char);
27219 vector unsigned char vec_reve (vector unsigned char);
27220 vector bool int vec_reve (vector bool int);
27221 vector signed int vec_reve (vector signed int);
27222 vector unsigned int vec_reve (vector unsigned int);
27223 vector bool long long vec_reve (vector bool long long);
27224 vector signed long long vec_reve (vector signed long long);
27225 vector unsigned long long vec_reve (vector unsigned long long);
27226 vector bool short vec_reve (vector bool short);
27227 vector signed short vec_reve (vector signed short);
27228 vector double vec_reve (vector double);
27229 vector float vec_reve (vector float);
27230 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27231 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27232 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27233 (altivec_vreve): New pattern.
27234 * config/rs6000/altivec.h (vec_reve): New define.
27235 * doc/extend.texi (vec_rev): Update the built-in documentation file
27236 for the new built-in functions.
27238 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27240 PR tree-optimization/71815
27241 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27243 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27245 (slsr_process_phi): Likewise.
27246 (replace_uncond_cands_and_profitable_phis): Don't replace a
27247 multiply candidate with a stride of 1 (copy or cast).
27248 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27250 (lowest_cost_path): Likewise.
27251 (total_savings): Likewise.
27253 2017-06-26 Richard Biener <rguenther@suse.de>
27256 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27257 Use def_builtin_pure for all gather builtins.
27259 2017-06-26 Richard Biener <rguenther@suse.de>
27261 PR tree-optimization/81203
27262 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27263 non-dominating BBs.
27265 2017-06-26 Marek Polacek <polacek@redhat.com>
27268 * doc/invoke.texi: Document -Wmultistatement-macros.
27270 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27272 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27273 arm_neon_ok_no_float_abi effective target.
27275 2017-06-26 Richard Biener <rguenther@suse.de>
27277 PR tree-optimization/80928
27278 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27279 (copy_bbs): Set BB_DUPLICATED flag early.
27280 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27282 (execute_on_shrinking_pred): Likewise.
27283 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27284 BB_DUPLICATED blocks.
27285 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27286 iterate over all PHIs considering removal of *gsi.
27288 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27290 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27293 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27295 * config/rs6000/rs6000-string.c: (expand_block_clear,
27296 do_load_for_compare, select_block_compare_mode,
27297 compute_current_alignment, expand_block_compare,
27298 expand_strncmp_align_check, expand_strn_compare,
27299 expand_block_move, rs6000_output_load_multiple)
27300 Move functions related to string/block move/compare
27301 to a separate file.
27302 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27303 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27304 for this function which is now used in two files.
27305 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27306 * config.gcc: Add rs6000-string.o to extra_objs for
27307 targets powerpc*-*-* and rs6000*-*-*.
27309 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27312 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27313 32-bit, since indexed is not valid for DImode.
27314 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27315 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27316 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27317 (define_peephole2 for Altivec d-form store): Likewise.
27320 * multiple_target.c (create_dispatcher_calls): Only create the
27321 dispatcher call if the function is the default clone of a
27322 versioned function.
27324 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27326 PR middle-end/80902
27327 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27328 a call, force the call to not be a tail call.
27330 2017-06-23 Jeff Law <law@redhat.com>
27332 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27335 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27337 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27338 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27339 handling for early expansion of vector shifts (sl,sr,sra,rl).
27340 (builtin_function_type): Add vector shift right instructions
27341 to the unsigned argument list.
27343 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27345 rtl-optimizatoin/79286
27346 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27347 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27348 trap. PIC register plus a const unspec without offset can never trap.
27350 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27352 * tree.h (builtin_structptr_type): New type.
27353 (builtin_structptr_types): Declare new array.
27354 * tree.c (builtin_structptr_types): New array.
27355 (free_lang_data, build_common_tree_nodes): Use it.
27357 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27360 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27363 2017-06-22 Matt Turner <mattst88@gmail.com>
27365 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27366 Lake models to skylake case. Assume skylake for unknown
27367 models with clflushopt.
27369 2017-06-22 Jeff Law <law@redhat.com>
27371 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27372 frame sizes that do not satisfy aarch64_uimm12_shift.
27374 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27376 * profile-count.h (apply_probability,
27377 apply_scale, probability_in): Fix checks for zero.
27379 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27381 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27382 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27384 2016-06-22 Richard Biener <rguenther@suse.de>
27386 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27387 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27388 REDUC_MAX_EXPR support.
27389 (vectorizable_reduction): Likewise.
27390 (vect_create_epilog_for_reduction): Likewise.
27392 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27394 * match.pd (A / (1 << B) -> A >> B): New.
27395 * generic-match-head.c: Include optabs-tree.h.
27396 * gimple-match-head.c: Likewise.
27397 * optabs-tree.h (target_supports_op_p): New.
27398 * optabs-tree.c (target_supports_op_p): New.
27400 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27402 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27403 $gcc_cv_ld --help output.
27404 (gcc_cv_ld_demangle): Likewise.
27405 (gcc_cv_ld_eh_frame_hdr): Likewise.
27406 (gcc_cv_ld_pie): Likewise.
27407 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27408 (gcc_cv_ld_buildid): Likewise.
27409 (gcc_cv_ld_sysroot): Likewise.
27410 (ld_bndplt_support): Likewise.
27411 (ld_pushpopstate_support): Likewise.
27412 * configure: Regenerate.
27413 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27415 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27418 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27419 operands indexes to avoid gap between operands and match_dups.
27421 2017-06-21 Andrew Pinski <apinski@cavium.com>
27423 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27424 Increment Arith_shift and Arith_shift_reg by 1.
27425 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27427 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27428 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27429 (aarch64_strip_extend): Add new argument and test for it.
27430 (aarch64_cheap_mult_shift_p): New function.
27431 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27432 add a cost if it is true.
27433 Update calls to aarch64_strip_extend.
27434 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27436 2017-06-21 Andrew Pinski <apinski@cavium.com>
27438 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27440 (thunderxt88): Likewise.
27441 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27442 (thunderx_prefetch_tune): New variable.
27443 (thunderx2t99_prefetch_tune): Update for the correct values.
27444 (thunderxt88_tunings): New variable.
27445 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27446 generic_prefetch_tune.
27447 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27449 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27451 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27452 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27453 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27454 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27455 (aarch64_atomic_cas<mode>, GPI): Likewise.
27457 2017-06-21 Martin Liska <mliska@suse.cz>
27459 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27460 statements on cold and hot labels.
27461 * predict.c (tree_estimate_probability_bb): Remove the
27462 prediction from this place.
27464 2017-06-21 Martin Liska <mliska@suse.cz>
27466 PR tree-optimization/79489
27467 * gimplify.c (maybe_add_early_return_predict_stmt): New
27469 (gimplify_return_expr): Call the function.
27470 * predict.c (tree_estimate_probability_bb): Remove handling
27472 * predict.def: Update comment about early return predictor.
27473 * gimple-predict.h (is_gimple_predict): New function.
27474 * predict.def: Change default value of early return to 66.
27475 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27477 * passes.def: Put pass_strip_predict_hints to the beginning of
27480 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27482 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27483 FUNCTION_DECL declarations.
27484 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27486 (dwaf2out_decl): Likewise.
27487 * godump.c (go_early_global_decl): Skip call to the real debug hook
27488 for FUNCTION_DECL declarations.
27489 * passes.c (rest_of_decl_compilation): Skip call to the
27490 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27491 -fdump-go-spec is passed.
27493 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27495 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27496 Reorder for compactness.
27497 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27498 (def_builtin_pure, def_builtin_pure2): New functions.
27499 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27501 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27503 * match.pd (nop_convert): New predicate.
27504 ((A +- CST1) +- CST2): Allow some NOP conversions.
27506 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27509 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27510 with ctors/dtors if GOVD_SHARED is set.
27512 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27514 * config/aarch64/aarch64.md (movti_aarch64):
27515 Emit mov rather than orr.
27516 (movtf_aarch64): Likewise.
27517 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27518 Emit mov rather than orr.
27520 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27522 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27523 Swap alternatives, make integer dup more expensive.
27525 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27527 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27528 Return true for non-tls symbols.
27530 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27532 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27533 (cortex-a75): Likewise.
27534 (cortex-a75.cortex-a55): Likewise.
27535 * config/aarch64/aarch64-tune.md: Regenerate.
27536 * doc/invoke.texi (-mtune): Document new values for -mtune.
27538 2017-06-21 Tom de Vries <tom@codesourcery.com>
27540 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27541 stack_size feature.
27542 (Effective-Target Keywords, Other attributes): Suggest using
27543 dg-add-options stack_size feature to get stack limit in stack_size
27544 effective target documentation.
27546 2017-06-21 Julian Brown <julian@codesourcery.com>
27547 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27549 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27550 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27551 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27553 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27554 attribute type list for neon_multiply.
27555 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27556 attribute type list for neon_multiply.
27557 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27558 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27559 attribute type list for neon_multiply.
27560 * config/arm/types.md (crypto_pmull): Add.
27561 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27562 attribute type list.
27564 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27566 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27569 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27571 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27572 to make sure not to dereference a NULL cost_classes_ptr pointer.
27574 2017-06-20 Carl Love <cel@us.ibm.com>
27576 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27577 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27578 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27579 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27580 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27581 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27582 VMULOSW): New enum "unspec" values.
27583 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27584 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27585 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27586 altivec_vmulosw): New patterns.
27587 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27588 VMULOSW): Add definitions.
27590 2017-06-20 Julia Koval <julia.koval@intel.com>
27592 * config/i386/i386.c: Fix rounding expand for new pattern.
27593 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27595 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27597 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27598 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27600 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27602 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27605 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27607 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27608 architecture, then by implementer ID.
27609 * config/aarch64/aarch64-tune.md: Regenerate.
27611 2017-06-20 Richard Biener <rguenther@suse.de>
27613 PR middle-end/81097
27614 * fold-const.c (split_tree): Fold to type before negating.
27616 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27618 * diagnostic-show-locus.c
27619 (selftest::test_fixit_deletion_affecting_newline): New function.
27620 (selftest::diagnostic_show_locus_c_tests): Call it.
27622 2017-06-20 Andreas Schwab <schwab@suse.de>
27625 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27628 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27630 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27631 __ARM_FEATURE_COPROC according to support.
27633 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27635 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27636 Rewritten to avoid overflow for > 32-bit pointers.
27639 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27640 by removing enum keyword.
27641 (ubsan_type_descriptor): Likewise. Formatting fix.
27644 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27645 splitter): Require TARGET_SSE2 in the condition.
27647 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27650 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27651 for doing vector set of SFmode on ISA 3.0.
27652 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27653 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27655 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27656 SFmode value into a V4SF variable that was extracted from another
27657 V4SF variable without converting the element to double precision
27658 and back to single precision vector format.
27659 (vsx_insert_extract_v4sf_p9_2): Likewise.
27661 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27663 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27664 in UWHI to avoid undefined overflow.
27667 * ubsan.h (enum ubsan_encode_value_phase): New.
27668 (ubsan_encode_value): Change second argument to
27669 enum ubsan_encode_value_phase with default value of
27670 UBSAN_ENCODE_VALUE_GENERIC.
27671 * ubsan.c (ubsan_encode_value): Change second argument to
27672 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27673 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27674 create_tmp_var_raw instead of create_tmp_var and use a
27676 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27677 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27678 ubsan_encode_value callers.
27681 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27682 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27683 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27685 2017-06-19 Richard Biener <rguenther@suse.de>
27687 PR middle-end/81118
27688 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27689 estimates if we changed anything.
27691 2017-06-19 Richard Biener <rguenther@suse.de>
27693 PR tree-optimization/80887
27694 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27695 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27696 simplified lookups, then reset mprts_hook.
27697 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27699 (try_to_simplify): Likewise.
27701 2017-06-19 Martin Liska <mliska@suse.cz>
27704 * gimplify.c (gimplify_switch_expr):
27705 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27707 2017-06-19 Martin Liska <mliska@suse.cz>
27709 * doc/install.texi: Document that PGO runs in 4 stages.
27711 2017-06-19 Martin Liska <mliska@suse.cz>
27714 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27715 to dispatcher function name.
27716 * multiple_target.c (replace_function_decl): New function.
27717 (create_dispatcher_calls): Redirect both edges and references.
27719 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27721 * profile-count.c (profile_count::dump): Dump quality.
27722 (profile_count::differs_from_p): Update for unsigned val.
27723 * profile-count.h (profile_count_quality): New enum.
27724 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27726 2017-06-19 Richard Biener <rguenther@suse.de>
27728 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27729 struct function as arg.
27730 (estimate_numbers_of_iterations): Export overload with loop arg.
27731 (free_numbers_of_iterations_estimates_loop): Use an overload of
27732 free_numbers_of_iterations_estimates instead.
27733 * tree-cfg.c (remove_bb): Adjust.
27734 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27735 * tree-parloops.c (gen_parallel_loop): Likewise.
27736 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27738 (tree_unroll_loops_completely): Likewise.
27739 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27740 Use an overload instead and export.
27741 (estimated_loop_iterations): Adjust.
27742 (max_loop_iterations): Likewise.
27743 (likely_max_loop_iterations): Likewise.
27744 (estimate_numbers_of_iterations): Take struct function as arg
27746 (loop_exits_before_overflow): Adjust.
27747 (free_numbers_of_iterations_estimates_loop): Use an overload.
27748 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27749 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27751 2017-06-19 Richard Biener <rguenther@suse.de>
27754 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27755 RANGE_EXPR conservatively.
27757 2017-06-16 Carl Love <cel@us.ibm.com>
27759 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27760 definitions for vec_float, vec_float2, vec_floato,
27761 vec_floate built-ins.
27762 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27763 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27765 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27766 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27767 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27768 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27769 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27770 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27771 vec_floato): Add builtin defines.
27772 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27773 Update the built-in documentation file for the new built-in
27776 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27778 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27779 (mthumb): Mark as the negative of -marm.
27781 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27783 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27785 (ARM Options, -mtune): Document that this accepts the same
27786 extension options as -mcpu.
27787 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27789 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27791 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27792 permitted extensions.
27794 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27796 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27797 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27798 (armv8-m.main): Add option +nodsp.
27799 * config/arm/arm-cpu-cdata.h: Regenerated.
27801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27803 * config/arm/t-fuchsia: New file.
27804 * config.gcc (arm*-*-fuchsia*): Use it.
27806 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27808 * config/arm/t-symbian: Rewrite for new option infrastructure.
27810 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27812 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27813 (MULTILIB_REQUIRED): Likewise.
27815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27817 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27818 (MULTILIB_RESUE): Likewise.
27819 (MULTILIB_MATCHES): Likewise.
27820 (MULTLIB_REQUIRED): Likewise.
27822 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27824 * config/arm/t-rtems: Rewrite for new option framework.
27826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27828 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27829 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27830 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27831 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27832 * config/arm/t-multilib: ... here.
27833 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27834 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27835 armv7-a and armv8*-a when A-profile libraries have not been built.
27836 * config/arm/t-rmprofile: Rewrite.
27838 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27840 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27841 with a backslash. Remove the backslash after substituting unescaped
27843 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27845 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27847 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27848 the driver to pass through the default setting of -mfloat-abi.
27849 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27851 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27852 (all_feat_combs): New rule.
27853 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27855 * config/arm/t-aprofile: Rewrite.
27857 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27859 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27860 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27861 fpu is not specified by the user/command-line.
27862 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27863 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27864 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27865 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27866 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27867 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27869 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27871 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27872 * config/arm/t-arm-elf: Rewritten.
27874 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27876 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27877 have some floating-point instructions.
27878 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27879 (TARGET_MAYBE_HARD_FLOAT): New macro.
27880 * config/arm/arm-builtins.c (arm_init_builtins): Use
27881 TARGET_MAYBE_HARD_FLOAT.
27882 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27884 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27886 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27887 (configargs.h): Include it.
27888 (arm_print_hint_for_fpu_option): New function.
27889 (arm_parse_fpu_option): New function.
27890 (candidate_extension): New class.
27891 (arm_canon_for_multilib): New function.
27892 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27893 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27894 (ARCH_CANONICAL_SPECS): New macro.
27895 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27899 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27900 are set after handling multilib fragments. Set target_cpu_default2
27903 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27905 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27907 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27908 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27910 (check_arch): Likewise.
27911 * config/arm/arm.c (arm_configure_build_target): Handle
27912 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27913 options in the default.
27915 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27917 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27918 when an option is an alias of another.
27919 * config/arm/parsecpu.awk (optalias): New parser token.
27920 (gen_comm_data): Mark non-alias options as such. Emit entries
27921 for extension aliases.
27922 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27923 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27924 (armv6kz, armv6zk, armv6t2): Likewise.
27925 (armv7): Make vfpv3-d16 an alias.
27926 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27928 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27929 Sort in canonical order.
27930 (armv8-a): Sort in canonical order.
27931 (armv8.1-a, armv8.2-a): Likewise.
27932 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27934 (cortex-a9): Sort in canonical order.
27935 * config/arm/arm.c (selftests.h): Include it.
27936 (arm_test_cpu_arch_data): New function.
27937 (arm_run_self_tests): New function.
27938 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27939 (targetm): Move declaration to the end of the file.
27940 * arm-cpu-cdata.h: Regenerated.
27942 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27944 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27945 call to target_mode_check describing the type of option passed.
27946 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27947 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27948 arm_parse_cpu_option_name to match parameters against list of
27950 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27951 arm_arch_core_flags data structure.
27952 * config/arm/arm-cpu_cdata.h: Regenerated.
27954 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27956 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27958 (arm_print_hint_for_cpu_option): Likewise.
27959 (arm_print_hint_for_arch_option): Likewise.
27960 (arm_parse_cpu_option_name): Likewise.
27961 (arm_parse_arch_option_name): Likewise.
27962 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27963 of entries in the all_fpus list.
27964 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27965 (arm_parse_cpu_option_name): Declare.
27966 (arm_parse_arch_option_name): Declare.
27967 (arm_parse_option_features): Declare.
27968 (arm_intialize_isa): Declare.
27969 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27971 (gen_comm_data): ... here. Make definitions non-static.
27972 * config/arm/arm-cpu-data.h: Regenerated.
27973 * config/arm/arm-cpu-cdata.h: Regenerated.
27975 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27977 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27978 (cpu_arch_extension): New structure.
27979 (cpu_arch_option, arch_option, cpu_option): New structures.
27980 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27981 architecture types.
27982 (gen_data): Generate new format data tables.
27983 * config/arm/arm.c (cpu_tune): New structure.
27984 (cpu_option, processors): Delete.
27985 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27986 (arm_print_hint_for_cpu_option): ... this and ...
27987 (arm_print_hint_for_arch_option): ... this.
27988 (arm_parse_arch_cpu_name): Delete. Replace with ...
27989 (arm_parse_cpu_option_name): ... this and ...
27990 (arm_parse_arch_option_name): ... this.
27991 (arm_unrecognized_feature): Change type of target parameter to
27993 (arm_parse_arch_cpu_features): Delete. Replace with ...
27994 (arm_parse_option_features): ... this.
27995 (arm_configure_build_target): Rework to use new configuration data
27997 (arm_print_tune_info): Rework for new configuration data tables.
27998 * config/arm/arm-cpu-data.h: Regenerated.
27999 * config/arm/arm-cpu.h: Regenerated.
28001 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28003 * Makefile.in (OBJS): Move sbitmap.o from here ...
28004 (OBJS-libcommon): ... to here.
28006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28008 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
28009 (ISA_ALL_CRYPTO): New macro.
28010 (ISA_ALL_SIMD): New macro
28011 (ISA_ALL_FP): New macro.
28012 * config/arm/arm.c (fpu_bitlist): Update initializer.
28013 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
28015 (arm9e): Add fpu. Add option for nofp
28016 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
28017 (arm926ej-s, arm1026ej-s): Likewise.
28018 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
28019 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
28020 neon-fp16, neon-vfpv4, nofp and nosimd.
28021 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
28022 (cortex-a8): Add fpu. Add option for nofp.
28023 (cortex-a9): Add fpu. Add options for nosimd and nofp.
28024 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
28025 (cortex-r4f): Add fpu.
28026 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
28027 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
28029 (cortex-r8): Likewise.
28030 (cortex-m4): Add fpu. Add option for nofp.
28031 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
28032 (cortex-a17.cortex-a7): Likewise.
28033 (cortex-a32): Add fpu. Add options for crypto and nofp.
28034 (cortex-a35, cortex-a53): Likewise.
28035 (cortex-a57): Add fpu. Add option for crypto.
28036 (cortex-a72, cortex-a73): Likewise.
28037 (exynos-m1): Likewise.
28038 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
28039 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
28040 (cortex-m33): Add fpu. Add option for nofp.
28041 * config/arm/arm-cpu-cdata.h: Regenerated
28042 * config/arm/arm-cpu-data.h: Regenerated.
28044 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28046 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
28047 (armv5te, armv5tej): Likewise.
28048 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
28049 (armv7): Add options fp and vfpv3-d16.
28050 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
28051 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
28053 (armv7ve): Likewise.
28054 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
28055 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
28056 (armv8-a): Add nocrypto option.
28057 (armv8.1-a, armv8.2-a): Likewise.
28058 (armv8-m.main): add options fp, fp.dp and nofp.
28060 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28062 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28064 (armv8-a+crc): Delete.
28065 (armv8.1-a): Add options simd, crypto and nofp.
28066 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28067 (armv8.2-a+fp16): Delete.
28068 (armv8-m.main): Add option dsp.
28069 (armv8-m.main+dsp): Delete.
28070 (cortex-a8): Add fpu. Add nofp option.
28071 (cortex-a9): Add fpu. Add nofp and nosimd options.
28072 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28073 link to main cpu and architecture data structures.
28074 (gen_comm_data): Only put isa attributes from the main architecture
28076 (option): New statement for architecture and CPU entries.
28077 * arm.c (struct cpu_option): New structure.
28078 (struct processors): Add entry for options.
28079 (arm_unrecognized_feature): New function.
28080 (arm_parse_arch_cpu_name): Ignore any characters after the first
28082 (arm_parse_arch_cpu_feature): New function.
28083 (arm_configure_build_target): Separate out any CPU and architecture
28084 features and parse separately. Don't error out if -mfpu=auto is
28085 used with only an architecture string.
28086 (arm_print_asm_arch_directives): New function.
28087 (arm_file_start): Call it.
28088 * config/arm/arm-cpu-cdata.h: Regenerated.
28089 * config/arm/arm-cpu-data.h: Likewise.
28090 * config/arm/arm-tables.opt: Likewise.
28092 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28094 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28095 assembler when it is not -mfpu=auto.
28097 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28099 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28100 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28101 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28102 (ASM_CPU_SPEC): Rewrite.
28103 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28104 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28105 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28106 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28107 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28108 copied string is NUL-terminated. Also strip any characters prefixed
28110 (arm_rewrite_selected_arch): New function.
28111 (arm_rewrite_march): New function.
28113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28115 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28116 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28117 (march, mcpu, mtune): Convert to string-based options.
28118 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28119 (arm_parse_arch_cpu_name): New function.
28120 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28121 identify selected architecture or CPU.
28122 (arm_option_save): New function.
28123 (TARGET_OPTION_SAVE): Redefine.
28124 (arm_option_restore): Restore string options.
28125 (arm_option_print): Print string options.
28127 2017-06-16 Martin Sebor <msebor@redhat.com>
28129 PR tree-optimization/80933
28130 PR tree-optimization/80934
28131 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28132 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28133 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28134 (gimple_fold_builtin): Call them.
28136 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28138 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28139 as unlikely; update profile.
28141 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28143 * predict.c (force_edge_cold): Handle declaring edges impossible
28146 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28148 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28150 (try_unroll_loop_completely): Fix reporting.
28152 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28154 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28156 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28159 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28160 if given a non-constant argument for an intrinsic which requires a
28163 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28165 * profile.c (compare_freqs): New function.
28166 (branch_prob): Sort edge list.
28167 (find_spanning_tree): Assume that the list is priority sorted.
28169 2017-06-16 Richard Biener <rguenther@suse.de>
28171 PR tree-optimization/81090
28172 * passes.def (pass_record_bounds): Remove.
28173 * tree-pass.h (make_pass_record_bounds): Likewise.
28174 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28175 make_pass_record_bounds): Likewise.
28176 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28177 not free niter estimates at the beginning but at the end.
28178 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28180 2017-06-16 Richard Biener <rguenther@suse.de>
28182 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28183 initializer to workaround ICE in host GCC 4.8.
28185 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28187 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28189 (clone_inlined_nodes): Update.
28191 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28193 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28194 prefetch settings, and enable prefetching by default at -O3.
28196 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28198 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28199 Set flag_prefetch_loop_arrays according to tuning data.
28201 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28203 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28204 New tune structure.
28205 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28206 [Unrelated to main purpose of the patch] Place the pointer field last
28207 to enable type checking errors when tune structure are wrongly merged.
28208 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28209 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28210 (thunderx2t99_prefetch_tune): New tune constants.
28211 (tune_params *_tunings): Update all tunings (no functional change).
28212 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28213 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28214 from tunings structures.
28216 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28219 * ubsan.c (instrument_null): Add T argument, use it instead
28220 of computing it based on IS_LHS.
28221 (instrument_object_size): Likewise.
28222 (pass_ubsan::execute): Adjust instrument_null and
28223 instrument_object_size callers to pass gimple_get_lhs or
28224 gimple_assign_rhs1 result to it. Use instrument_null instead of
28225 calling get_base_address and instrument_mem_ref. Handle
28226 aggregate call arguments for object-size sanitization.
28228 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28230 PR tree-optimization/81089
28231 * tree-vrp.c (is_masked_range_test): Validate operands of
28234 2017-06-15 Martin Sebor <msebor@redhat.com>
28237 * dumpfile.c (dump_register): Avoid calling memset to initialize
28238 a class with a default ctor.
28239 * gcc.c (struct compiler): Remove const qualification.
28240 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28241 * hash-table.h: Ditto.
28242 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28244 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28245 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28247 * params.h (struct param_info): Make struct members non-const.
28248 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28249 with default initialization.
28250 * vec.h (vec_copy_construct, vec_default_construct): New helper
28252 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28253 with vec_copy_construct.
28254 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28255 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28256 * doc/invoke.texi (-Wclass-memaccess): Document.
28258 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28260 * emit-rtl.h (is_leaf): Update comment about local
28261 register allocator.
28263 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28266 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28267 for a variable to have a section before checking if the section has a
28269 Set section to.persistent if persistent attribute is set.
28270 Warn if .persistent attribute is used on an automatic variable.
28272 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28274 PR rtl-optimization/80474
28275 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28277 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28279 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28282 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28284 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28285 (casesi): Emit bounds checking as RTL.
28286 (casesi_internal_mips16_<mode>): Remove bounds checking.
28288 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28290 * config/xtensa/xtensa.c (xtensa_option_override): Append
28291 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28292 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28293 xtensa_doloop_hooks): Define unconditionally.
28294 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28296 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28297 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28298 for it in xtensa_option_override.
28299 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28300 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28302 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28304 * doc/cppopts.texi: Document '-' special value to -MF.
28306 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28308 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28309 (cortex_a53_fconst): Likewise.
28310 (cortex_a53_fpmul): Likewise.
28311 (cortex_a53_f_load_64): Likewise.
28312 (cortex_a53_f_load_many): Likewise.
28313 (cortex_a53_advsimd_alu): Likewise.
28314 (cortex_a53_advsimd_alu_q): Likewise.
28315 (cortex_a53_advsimd_mul): Likewise.
28316 (cortex_a53_advsimd_mul_q): Likewise.
28317 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28318 Add missing fmul, r2f_cvt and fconst cases.
28320 2017-06-14 Richard Biener <rguenther@suse.de>
28322 PR middle-end/81088
28323 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28325 (fold_binary_loc): When associating do not treat pre-existing
28326 TREE_OVERFLOW on literal constants as a reason to allow
28327 TREE_OVERFLOW on associated literal constants.
28329 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28331 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28332 (MASK_FEATURES): New macro.
28333 * config/sparc/sparc.c (sparc_option_override): Remove the special
28334 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28336 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28338 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28339 a division of 0 if non-call exceptions are enabled.
28341 2017-06-14 Andrew Pinski <apinski@cavium.com>
28342 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28345 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28346 Improve vector initialization code gen for only variable case.
28348 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28350 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28352 2017-06-14 Richard Biener <rguenther@suse.de>
28354 PR tree-optimization/81083
28355 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28358 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28360 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28361 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28362 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28363 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28364 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28365 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28367 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28369 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28370 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28372 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28374 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28376 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28378 * config/rs6000/t-rtems: Don't handle SPE.
28380 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28382 * config/rs6000/t-linux: Don't handle SPE.
28384 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28386 * config/rs6000/eabispe.h: Delete file.
28388 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28390 * config/rs6000/t-spe: Delete file.
28392 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28394 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28395 (rs6000_legitimate_offset_address_p): Return false for anything in
28396 V2SImode or V2SFmode.
28398 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28400 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28401 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28402 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28404 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28405 (rs6000_legitimate_offset_address_p): Ditto.
28406 (rs6000_emit_move): Ditto.
28407 (rs6000_init_builtins): Remove V4HI_type_node.
28409 2017-06-13 Martin Liska <mliska@suse.cz>
28412 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28413 (gate_asan): Likewise.
28414 * asan.h (asan_no_sanitize_address_p): Remove the function.
28415 (sanitize_flags_p): New function.
28416 * builtins.def: Fix coding style.
28417 * common.opt: Use renamed enum value.
28418 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28419 * doc/extend.texi: Document no_sanitize attribute.
28420 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28421 to SANITIZE_UNDEFINED_NONDEFAULT.
28422 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28423 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28424 Use sanitize_flags_p.
28425 * gimplify.c (gimplify_function_tree): Likewise.
28426 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28427 * opts.c (parse_no_sanitize_attribute): New function.
28428 (common_handle_option): Use renamed enum value.
28429 * opts.h (parse_no_sanitize_attribute): Declare.
28430 * tree.c (sanitize_flags_p): New function.
28431 * tree.h: Declared here.
28432 * tsan.c: Use sanitize_flags_p.
28433 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28434 (instrument_mem_ref): Likewise.
28435 (instrument_bool_enum_load): Likewise.
28436 (do_ubsan_in_current_function): Remove the function.
28437 (pass_ubsan::execute): Use sanitize_flags_p.
28438 * ubsan.h: Remove do_ubsan_in_current_function
28439 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28440 (dump_function_to_file): Use it here.
28442 2017-06-13 Martin Jambor <mjambor@suse.cz>
28444 PR tree-optimization/80803
28445 PR tree-optimization/81063
28446 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28447 (propagate_subaccesses_across_link): Enqueue subtree whenever
28448 necessary instead of relying on the caller.
28450 2017-06-13 Martin Jambor <mjambor@suse.cz>
28452 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28453 that have a first_link.
28454 (sort_and_splice_var_accesses): Do not check first_link before
28456 (subtree_mark_written_and_enqueue): Likewise.
28457 (propagate_all_subaccesses): Likewise and do not stop at first
28458 parent with a first_link.
28460 2017-06-13 Martin Jambor <mjambor@suse.cz>
28462 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28465 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28467 * match.pd: New pattern.
28469 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28471 * tree-vrp.c (is_masked_range_test): New function.
28472 (register_edge_assert_for): Determine ranges for
28475 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28477 PR tree-optimization/67328
28478 * fold-const.c (maskable_range_p): New function.
28479 (build_range_check): Generate bittests if possible.
28481 2017-06-13 Martin Liska <mliska@suse.cz>
28483 * gimple-pretty-print.c (dump_probability): Add new argument.
28484 (dump_edge_probability): Dump both probability and count.
28485 (dump_gimple_label): Likewise.
28486 (dump_gimple_bb_header): Likewise.
28488 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28491 * config/avr/avr-devices.c: Fix indentation.
28492 * config/avr/gen-avr-mmcu-specs.c: Dito.
28494 2017-06-13 Richard Biener <rguenther@suse.de>
28496 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28497 instead get vector type from stmt_info.
28498 (vectorizable_reduction): Adjust. Remove dead code.
28500 2017-06-13 Richard Biener <rguenther@suse.de>
28502 PR middle-end/81065
28503 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28504 case of C * (x * C2 + C3).
28505 (fold_addr_of_array_ref_difference): Properly fold index difference.
28507 2017-06-12 David S. Miller <davem@davemloft.net>
28510 * config/sparc/sparc.md (return expander): Emit frame blockage if
28511 function uses alloca.
28513 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28515 * combine.c (make_field_assignment): Check len rather than the mode
28516 precision when calling force_to_mode.
28518 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28520 Support multilibs and devices that see flash in RAM address range.
28523 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28524 (avr_mcu_t) <flash_pm_offset>: New field.
28525 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28526 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28527 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28528 (AVR_TINY_PM_OFFSET): Remove macro.
28529 * config/avr/avr.opt (-mshort-calls): New option.
28530 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28531 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28532 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28533 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28534 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28535 instead of avr_arch->have_jmp_call.
28536 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28537 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28538 avr_arch->flash_pm_offset to define.
28539 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28540 new field flash_pm_offset. Add entry for avrxmega3.
28541 (avr_texinfo): Add entry for avrxmega3.
28542 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28543 attiny212, attiny214,
28544 attiny412, attiny414, attiny416, attiny417,
28545 attiny814, attiny816, attiny817,
28546 attiny1614, attiny1616, attiny1617,
28547 attiny3214, attiny3216, attiny3217.
28548 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28549 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28550 (avr_print_operand_address) [AVR_TINY]: Same.
28551 (avr_asm_init_sections) <readonly_data_section>: Only patch
28552 callback if avr_arch->flash_pm_offset = 0.
28553 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28554 for rodata if avr_arch->flash_pm_offset != 0.
28555 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28556 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28557 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28558 (m_options): Append opt_rcall.
28559 (m_dirnames): Append dir_rcall.
28560 * config/avr/t-multilib: Regenerate.
28562 * configure.ac [target=avr]: Check whether avrxmega3 default
28563 linker description file works as needed.
28564 * configure: Regenerate.
28565 * doc/avr-mmcu.texi: Regenerate.
28566 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28567 <__AVR_ARCH__>: Document avrxmega3 and 103.
28568 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28569 <__AVR_SHORT_CALLS__>: Document it.
28570 <__AVR_PM_BASE_ADDRESS__>: Document it.
28571 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28572 (AVR Variable Attributes) <progmem>: Document this is
28573 not needed for avrxmega3.
28574 (AVR Named Address Spaces) <__flash>: Dito.
28576 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28578 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28580 2017-06-12 Doug Rupp <rupp@adacore.com>
28582 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28583 Append vxworks-stdint.h to the tm_file list.
28584 * config/vxworks-stdint.h: New file.
28586 2017-06-12 Martin Liska <mliska@suse.cz>
28588 PR tree-optimization/81041
28589 * tree-profile.c (gimple_gen_ic_func_profiler):
28590 Create an extra BB in profile-generate
28591 (gimple_gen_time_profiler): Likewise.
28593 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28595 PR tree-optimization/81003
28596 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28597 (update_range_test): Use it instead of force_gimple_operand_gsi.
28599 2017-06-12 Richard Biener <rguenther@suse.de>
28601 PR tree-optimization/81053
28602 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28603 with backedge value not defined in loop. Simplify def stmt
28606 2017-06-11 Tom de Vries <tom@codesourcery.com>
28609 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28611 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28612 nvptx_cannot_force_const_mem.
28614 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28616 * opts.c (finish_options): Move test for flag_split_stack after
28617 it has been initialized.
28619 2017-06-11 Jason Merrill <jason@redhat.com>
28621 * tree.h (id_equal): New.
28622 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28623 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28624 instead of strcmp of IDENTIFIER_POINTER.
28626 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28628 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28629 (mark_all_inlined_calls_cdtor): Fix formating.
28630 (inline_transform): Rescale profile before inlining.
28632 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28634 * cgraph.h (cgraph_edge::clone): Update prototype.
28635 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28636 (cgraph_node::create_clone): Update.
28637 (cgraph_node::create_version_clone): Update.
28638 * tree-inline.c (copy_bb): Update.
28639 (expand_call_inline): Update.
28641 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28643 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28644 factored out from ...
28645 (rs6000_emit_prologue): ... here.
28647 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28649 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28650 factored out from ...
28651 (rs6000_emit_prologue): ... here.
28653 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28655 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28657 (handle_missing_profiles): Fix computation of tp_first_run.
28658 (counts_to_freqs): Do not touch freqs when count is 0.
28660 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28662 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28665 2017-06-10 Tom de Vries <tom@codesourcery.com>
28667 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28668 attributes): Document signal effective target.
28670 2017-06-10 Tom de Vries <tom@codesourcery.com>
28672 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28673 Document effective target stack_size.
28675 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28677 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28678 to the edit_context if they can be auto-applied.
28680 2017-06-9 Ian Lance Taylor <iant@golang.org>
28682 * opts.c (finish_options): If -fsplit-stack, disable implicit
28683 -forder-blocks-and-partition.
28684 * doc/invoke.texi (Optimize Options): Document that when using
28685 -fsplit-stack -forder-blocks-and-partition is not implicitly
28688 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28690 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28691 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28692 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28693 * builtins.def (abort, trap, unreachable): Declare cold.
28694 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28695 * tree-core.h (ECF_COLD): New.
28696 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28697 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28699 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28701 * predict.c (unlikely_executed_stmt_p): Cleanup.
28703 2017-06-09 Richard Biener <rguenther@suse.de>
28705 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28706 model if the ref is always written to.
28708 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28710 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28712 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28714 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28717 2017-06-09 Tom de Vries <tom@codesourcery.com>
28720 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28721 "target cannot support label values" when encountering LABEL_REF.
28723 2017-06-09 Martin Liska <mliska@suse.cz>
28725 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28726 (gimple_gen_ic_func_profiler): Emit direct comparison
28727 of __gcov_indirect_call_callee with NULL.
28728 (gimple_gen_time_profiler): Change probability from
28729 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28731 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28733 * profile.c (edge_gcov_counts): Turn to pointer.
28734 (compute_branch_probabilities, compute_branch_probabilities): Update.
28735 (branch_prob): Do not clear edge_gcov_count.
28736 * profile.h (edge_gcov_counts): Turn to pointer.
28737 (edge_gcov_count): Update.
28739 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28741 * gimple.h (gimple_check_failed): Mark cold.
28743 2017-06-09 Richard Biener <rguenther@suse.de>
28745 PR tree-optimization/66623
28746 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28747 refactor check_reduction into two parts, properly computing
28748 whether we have to check reduction validity for outer loop
28751 2017-06-09 Richard Biener <rguenther@suse.de>
28753 PR tree-optimization/79483
28754 * graphite-scop-detection.c (order): New global.
28755 (get_order): Compute bb to order mapping that satisfies code
28756 generation constraints.
28757 (cmp_pbbs): New helper.
28758 (build_scops): Start domwalk at entry block, sort generated
28761 2017-06-09 Richard Biener <rguenther@suse.de>
28763 PR middle-end/81007
28764 * ipa-polymorphic-call.c
28765 (ipa_polymorphic_call_context::restrict_to_inner_class):
28766 Skip FIELD_DECLs with error_mark_node type.
28767 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28770 2017-06-09 Martin Liska <mliska@suse.cz>
28772 * predict.c (struct branch_predictor): New struct.
28773 (test_prediction_value_range): New test.
28774 (predict_c_tests): New function.
28775 * selftest-run-tests.c (selftest::run_tests): Run the function.
28776 * selftest.h: Declare new tests.
28778 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28781 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28782 gen_add3_insn did not fail.
28783 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28784 r0, construct that number in a temporary reg and add that reg to r0.
28785 If asked to put the result in r0 as well, fail.
28787 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28789 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28790 for early expansion of vec_eqv.
28792 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28794 PR middle-end/81005
28795 * ubsan.c (instrument_null): Avoid pointless code temporary.
28796 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28798 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28802 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28805 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28808 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28810 * predict.c (unlikely_executed_edge_p): Move ahead.
28811 (probably_never_executed_edge_p): Use it.
28813 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28815 PR middle-end/79988
28816 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28817 gimple_call_builtin_p call.
28819 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28821 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28822 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28823 rtl_check_failed_type2, rtl_check_failed_code1,
28824 rtl_check_failed_code2, rtl_check_failed_code_mode,
28825 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28826 rtvec_check_failed_bounds, rtl_check_failed_flag,
28827 _fatal_insn_not_found, _fatal_insn): Likewise.
28828 * tree.h (tree_contains_struct_check_failed,
28829 tree_check_failed, tree_not_check_failed,
28830 tree_class_check_failed, tree_range_check_failed,
28831 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28832 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28833 tree_operand_check_failed, omp_clause_check_failed,
28834 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28837 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28839 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28840 flag_branch_probabilities.
28841 * ipa-inline.c (edge_badness): Likewise.
28842 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28843 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28844 * predict.c (maybe_hot_frequency_p): Likewise.
28845 (probably_never_executed): Likewise.
28846 * sched-ebb.c (schedule_ebbs): Likewise.
28847 * sched-rgn.c (find_single_block_region): Likewise.
28848 * tracer.c (tail_duplicate): Likewise.
28850 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28852 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28853 longer requires x_flag_profile_use.
28855 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28857 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28858 instead of flag_reorder_blocks_and_partition.
28859 * dbxout.c (dbxout_function_end): Likewise.
28860 * dwarf2out.c (gen_subprogram_die): Likewise.
28861 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28862 * hw-doloop.c (reorg_loops): Likewise.
28863 * varasm.c (assemble_start_function,
28864 assemble_end_function): Likewise.
28865 (decide_function_section): Do not check for
28866 flag_reorder_blocks_and_partition.
28868 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28870 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28872 (chkp_get_hard_register_fake_addr_expr): Ditto.
28873 (chkp_build_addr_expr): Add check for hard reg case.
28874 (chkp_parse_array_and_component_ref): Ditto.
28875 (chkp_find_bounds_1): Ditto.
28876 (chkp_process_stmt): Don't generate bounds store for
28879 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28881 * predict.c (maybe_hot_bb_p): Do not check profile status.
28882 (maybe_hot_edge_p): Likewise.
28883 (probably_never_executed): Check for zero counts even if profile
28885 (unlikely_executed_edge_p): New function.
28886 (unlikely_executed_stmt_p): New function.
28887 (unlikely_executed_bb_p): New function.
28888 (set_even_probabilities): Use unlikely predicates.
28889 (combine_predictions_for_bb): Likewise.
28890 (predict_paths_for_bb): Likewise.
28891 (predict_paths_leading_to_edge): Likewise.
28892 (determine_unlikely_bbs): New function.
28893 (estimate_bb_frequencies): Use it.
28894 (compute_function_frequency): Use zero counts even if profile is
28896 * profile-count.h: Fix typo.
28898 2017-08-08 Julia Koval <julia.koval@intel.com>
28900 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28901 _mm512_mask_cvtsepi16_storeu_epi8,
28902 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28903 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28904 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28905 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28906 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28907 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28908 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28909 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28910 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28911 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28912 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28913 __builtin_ia32_pmovuswb256mem_mask,
28914 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28915 __builtin_ia32_pmovwb512mem_mask): New builtins.
28917 2017-08-08 Julia Koval <julia.koval@intel.com>
28919 PR target/73350,80862
28920 * config/i386/subst.md (round): Fix round pattern.
28921 * config/i386/i386.c (ix86_erase_embedded_rounding):
28922 Fix erasing rounding for the fixed pattern.
28924 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28926 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28928 2017-06-08 Martin Liska <mliska@suse.cz>
28930 PR gcov-profile/80911
28931 * gcov.c (block_info::block_info): New constructor.
28933 2017-06-07 Carl Love <cel@us.ibm.com>
28935 * config/rs6000/rs6000-c: The return type of the following
28936 built-in functions was implemented as int not long long. Fix sign
28937 of return value for the unsigned version of vec_mulo and vec_mule.
28938 vector unsigned long long vec_bperm (vector unsigned long long,
28939 vector unsigned char)
28940 vector signed long long vec_mule (vector signed int,
28942 vector unsigned long long vec_mule (vector unsigned int,
28943 vector unsigned int)
28944 vector signed long long vec_mulo (vector signed int,
28946 vector unsigned long long vec_mulo (vector unsigned int,
28947 vector unsigned int)
28948 * doc/extend.texi: Fix the documentation for the built-in
28951 2017-06-07 Carl Love <cel@us.ibm.com>
28954 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28957 2017-06-07 Carl Love <cel@us.ibm.com>
28959 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28960 support, Generate doublehv for signed int/float for BE case only.
28962 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28964 * doc/invoke.texi (mcx16): Rewrite.
28966 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28968 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28969 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28970 *mov<mode>_softfloat, and an anonymous splitter): Use
28971 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28973 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28975 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28977 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28978 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28979 (rs6000_debug_reg_global): Adjust.
28980 (rs6000_init_hard_regno_mode_ok): Adjust.
28981 (rs6000_dbx_register_number): Adjust.
28982 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28983 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28984 Remove SPE_ACC and SPEFSCR.
28985 (REG_ALLOC_ORDER): Ditto.
28986 (FRAME_POINTER_REGNUM): Change to 111.
28987 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28988 (REG_CLASS_NAMES): Ditto.
28989 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28990 (REGISTER_NAMES): Ditto.
28991 (ADDITIONAL_REG_NAMES): Ditto.
28992 (rs6000_reg_names): Ditto.
28993 * config/rs6000/rs6000.md: Renumber some register number
28996 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28998 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
29000 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29001 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
29003 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
29004 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29005 Delete the SPE high registers.
29006 (REG_ALLOC_ORDER): Ditto.
29007 (enum reg_class): Remove SPE_HIGH_REGS.
29008 (REG_CLASS_NAMES): Ditto.
29009 (REG_CLASS_CONTENTS): Delete the SPE high registers.
29010 (REGISTER_NAMES): Ditto.
29011 (rs6000_reg_names): Ditto.
29012 * doc/tm.texi.in: Remove SPE as example.
29013 * doc/tm.texi: Regenerate.
29015 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29017 * config/rs6000/8540.md (ppc8540_brinc): Delete.
29018 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
29019 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
29020 * config/rs6000/rs6000.md (type): Remove "brinc".
29022 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29024 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
29025 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
29026 * config/rs6000/linuxspe.h: Delete file.
29027 * config/rs6000/rs6000.md: Don't include spe.md.
29028 * config/rs6000/spe.h: Delete file.
29029 * config/rs6000/spe.md: Delete file.
29030 * config/rs6000/t-rs6000: Remove spe.md.
29032 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29034 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
29035 (reg_or_none500mem_operand): Delete.
29036 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
29037 instead of reg_or_none500mem_operand.
29039 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29041 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29042 handling of SPE flags.
29043 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
29045 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29047 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
29049 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
29050 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
29051 paired_divv2sf3): Similar.
29052 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
29053 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
29054 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
29056 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
29057 Rename the paired_* instruction patterns.
29058 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
29060 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29061 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29062 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29063 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29064 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29065 PAIRED_VECTOR_MODE.
29066 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29067 (spe_func_has_64bit_regs_p): Delete.
29068 (spe_expand_predicate_builtin): Delete.
29069 (spe_expand_evsel_builtin): Delete.
29070 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29071 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29072 (invalid_e500_subreg): Delete.
29073 (rs6000_legitimize_address): Always force_reg op2 as well, for
29074 paired single memory accesses.
29075 (rs6000_member_type_forces_blk): Delete.
29076 (rs6000_spe_function_arg): Delete.
29077 (rs6000_expand_unop_builtin): Delete SPE handling.
29078 (rs6000_expand_binop_builtin): Ditto.
29079 (spe_expand_stv_builtin): Delete.
29080 (bdesc_2arg_spe): Delete.
29081 (spe_expand_builtin): Delete.
29082 (spe_expand_predicate_builtin): Delete.
29083 (spe_expand_evsel_builtin): Delete.
29084 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29085 (spe_init_builtins): Delete.
29086 (spe_func_has_64bit_regs_p): Delete.
29087 (savres_routine_name): Delete "info" parameter. Adjust callers.
29088 (rs6000_emit_stack_reset): Ditto.
29089 (rs6000_dwarf_register_span): Delete.
29090 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29091 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29092 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29094 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29096 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29097 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29098 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29099 mulv2sf3, divv2sf3): Delete expanders.
29101 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29103 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29105 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29107 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29108 * config/rs6000/rs6000.c: Ditto.
29110 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29112 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29113 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29114 comparison_operator.
29116 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29118 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29119 * config/rs6000/rs6000.opt: Ditto.
29120 * config/rs6000/t-rtems: Ditto.
29122 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29124 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29125 TARGET_E500_SINGLE by 0, simplify.
29126 * config/rs6000/rs6000.c: Ditto.
29127 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29128 (spe_build_register_parallel): Delete.
29129 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29130 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29131 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29132 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29133 (E500_CONVERT): Delete.
29134 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29136 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29138 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29139 * config/rs6000/dfp.md: Ditto.
29140 (negdd2, *negdd2_fpr): Merge.
29141 (absdd2, *absdd2_fpr): Merge.
29142 (negtd2, *negtd2_fpr): Merge.
29143 (abstd2, *abstd2_fpr): Merge.
29144 * config/rs6000/e500.h: Delete file.
29145 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29146 TARGET_FPRS by 1 and simplify.
29147 * config/rs6000/rs6000-c.c: Ditto.
29148 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29149 TARGET_DF_SPE by 0.
29150 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29152 * config/rs6000/rs6000.md: Ditto.
29153 (floatdidf2, *floatdidf2_fpr): Merge.
29154 (move_from_CR_gt_bit): Delete.
29155 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29156 (E500_CR_IOR_COMPARE): Delete.
29157 (All patterns that require !TARGET_FPRS): Delete.
29158 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29160 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29162 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29164 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29166 * graphds.c (add_edge): Intitialize edge's attached data.
29167 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29168 pointer parameter. Call pointed function on each edge during
29169 graph traversing. Skip traversing the edge when the function
29171 (graphds_dfs, graphds_scc): Ditto.
29172 (for_each_edge): New parameter. Pass the new parameter to callback
29174 * graphds.h (skip_edge_callback): New function pointer type.
29175 (graphds_dfs, graphds_scc): New function pointer parameter.
29176 (graphds_edge_callback, for_each_edge): New parameter.
29178 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29180 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29181 out code checking if runtime alias check is possible to below ...
29182 Call the new function.
29183 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29184 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29186 2017-06-07 Marek Polacek <polacek@redhat.com>
29189 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29190 TYPE_OVERFLOW_WRAPS check.
29192 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29194 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29195 if versioning is required.
29196 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29197 peeling with the check for versioning.
29199 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29201 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29202 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29203 Set true to new parameter if new ssa variable is defined.
29204 (vect_gen_vector_loop_niters): Refactor. Set range information
29205 for the new vector loop bound variable.
29206 (vect_do_peeling): Ditto.
29208 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29210 * tree-affine.c (ssa.h): Include header file.
29211 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29212 has wrapping overflow behavior.
29214 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29216 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29218 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29220 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29221 (tree_to_aff_combination): ... here.
29223 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29225 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29226 reg_pressure model function.
29227 (ivopts_global_cost_for_size): Delete.
29228 (determine_set_costs, iv_ca_recount_cost): Call new model function
29229 ivopts_estimate_reg_pressure.
29231 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29233 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29234 expensive than udiv. Remove floating point cases from mod.
29236 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29238 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29239 Increase idiv cost.
29241 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29243 * config/aarch64/aarch64.md
29244 (copysignsf3): Fix mask generation.
29246 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29248 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29250 (class dump_manager): Add register_dumps method.
29251 * dumpfile.c: Include langhooks.h.
29252 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29253 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29254 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29255 (dump_manager::dump_register): Start with 512 entries instead of 32.
29256 (dump_manager::register_dumps): New method.
29257 * toplev.c (general_init): Instead of invoking register_dumps
29258 langhook, invoke register_dumps method on the dump manager.
29259 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29262 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29264 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29265 Say that # requires an associated define_split to exist, and that
29266 the define_split must be suitable for use after register allocation.
29268 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29270 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29271 (compute_outgoing_frequencies): Also initialize zero counts.
29272 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29273 around loops; preserve more of profile when nothing changes.
29275 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29277 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29279 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29280 * config/arm/arm-cpu-cdata.h: Regenerate.
29281 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29282 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29283 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29284 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29285 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29287 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29289 * config/arm/t-rmprofile: Likewise.
29290 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29292 2017-06-06 David S. Miller <davem@davemloft.net>
29295 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29296 blockage if function uses alloca.
29298 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29300 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29301 New "uid" fields to hold pretty-print IDs of group and ref.
29302 Memory references are now identified as <group_id>:<ref_id>
29303 instead of using [random] addresses.
29304 (dump_mem_details): Simplify, no functional change.
29305 (dump_mem_ref): Simplify and make output more concise.
29306 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29307 (find_or_create_group): Initialize group uid.
29308 (record_ref): Initialize ref uid. Improve debug output.
29309 (prune_group_by_reuse, should_issue_prefetch_p,)
29310 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29311 (mark_nontemporal_store, determine_loop_nest_reuse):
29312 Improve debug output.
29314 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29316 * dbgcnt.def (prefetch): New debug counter.
29317 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29318 (schedule_prefetches): Stop issueing prefetches if debug counter
29321 2017-06-06 Tom de Vries <tom@codesourcery.com>
29323 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29324 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29326 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29328 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29329 Use aarch64_reg_or_zero predicate for operand 4.
29330 (aarch64_compare_and_swap<mode> define_insn_and_split):
29331 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29332 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29334 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29336 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29337 (arm_compute_save_core_reg_mask): This.
29338 (thumb1_compute_save_reg_mask): Rename into ...
29339 (thumb1_compute_save_core_reg_mask): This.
29340 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29341 (arm_compute_frame_layout): Likewise.
29343 2017-06-06 Richard Biener <rguenther@suse.de>
29345 PR tree-optimization/80974
29346 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29347 keep or clear leaders SSA info.
29349 2017-06-06 Tom de Vries <tom@codesourcery.com>
29351 * config/nvptx/nvptx.c (split_mode_p): New function.
29352 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29354 2017-06-06 Tom de Vries <tom@codesourcery.com>
29356 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29358 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29361 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29364 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29366 * shrink-wrap.c (handle_simple_exit): Update profile.
29367 (try_shrink_wrapping): Upate profile.
29369 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29371 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29372 (tree_guess_outgoing_edge_probabilities): New.
29373 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29374 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29376 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29378 * ipa-split.c (split_function): Initialize return bb profile.
29380 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29382 * profile.c (compute_branch_probabilities): Also initialize
29383 EXIT_BLOCK profile.
29385 2017-06-06 Richard Biener <rguenther@suse.de>
29387 PR tree-optimization/80928
29388 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29389 (vect_analyze_loop_operations): Properly guard analysis for
29391 (vect_transform_loop): Likewise.
29392 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29393 (vect_model_induction_cost): Do not cost for pure SLP.
29394 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29395 of induction in inner loop vectorization.
29396 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29397 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29398 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29400 (vect_analyze_slp_cost_1): Cost induction.
29401 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29402 (vect_get_slp_vect_defs): Likewise.
29403 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29404 (vect_transform_stmt): Handle SLP reductions.
29405 * tree-vectorizer.h (vectorizable_induction): Adjust.
29407 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29409 * config/rs6000/rs6000.c (make_resolver_func): Update
29410 init_lowered_empty_function call.
29412 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29414 * doc/invoke.texi: Document the -fprofile-abs-path option.
29415 * common.opt (fprofile-abs-path): New option.
29416 * gcov-io.h (gcov_write_filename): Declare.
29417 * gcov-io.c (gcov_write_filename): New function.
29418 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29419 * profile.c (output_location): Likewise.
29421 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29423 * shring-wrap.c: Revert accidental commit.
29425 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29427 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29429 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29431 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29433 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29434 profile in callgraph edge.
29435 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29436 (apply_scale): Likewise.
29437 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29438 Also scale profile when inlining function with zero profile.
29439 (initialize_cfun): Update exit block profile even when it is zero.
29440 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29441 when profile is read.
29443 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29445 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29446 (CLONE_*): New constants to define the processors we can generate
29447 code for with the target_clone attribute.
29448 (rs6000_clone_map): New array to identify which clone processors
29449 the current program is running on.
29450 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29451 target_clone attribute.
29452 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29453 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29454 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29455 (cpu_expand_builtin): Add support for target_clone attribute.
29456 (rs6000_valid_attribute_p): Allow "default" attribute.
29457 (get_decl_name): New debug function to simplify printing the
29458 current function name in debugging statements.
29459 (rs6000_clone_priority): New functions to support the target_clone
29460 attribute, and be able to generate code to switch between ISA 2.05
29461 through ISA 3.0 (power6 through power9).
29462 (rs6000_compare_version_priority): Likewise.
29463 (rs6000_get_function_versions_dispatcher): Likewise.
29464 (make_resolver_func): Likewise.
29465 (add_condition_to_bb): Likewise.
29466 (dispatch_function_versions): Likewise.
29467 (rs6000_generate_version_dispatcher_body): Likewise.
29468 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29469 (fusion_gpr_load_p): Fix a spacing issue.
29470 * doc/extend.texi (Common Function Attributes): Document that the
29471 PowerPC supports the target_clone attribute.
29473 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29475 * config/arm/arm.h: explain F symbol found in description of ARM
29476 register allocation in its legend.
29478 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29480 * config/mips/frame-header-opt.c: Include profile-count.h.
29481 * config/riscv/riscv.c: Include profile-count.h
29483 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29485 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29487 (sm_set_flag_if_changed): Add bbs field.
29488 (execute_sm_if_changed_flag_set): Pass BBS.
29489 (execute_sm): Update.
29491 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29493 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29496 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29498 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29499 (peephole2): New peephole2 to emit the above.
29500 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29502 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29504 * config/aarch64/aarch64.c (define_peephole2 above
29505 *sub_<shift>_<mode>): New peephole.
29507 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29509 * config/i386/i386.c (make_resolver_func): Update.
29510 * Makefile.in: Add profile-count.h and profile-count.o
29511 * auto-profile.c (afdo_indirect_call): Update to new API.
29512 (afdo_set_bb_count): Update.
29513 (afdo_propagate_edge): Update.
29514 (afdo_propagate_circuit): Update.
29515 (afdo_calculate_branch_prob): Update.
29516 (afdo_annotate_cfg): Update.
29517 * basic-block.h: Include profile-count.h
29518 (struct edge_def): Turn count to profile_count.
29519 (struct basic_block_def): Likewie.
29520 (REG_BR_PROB_BASE): Move to profile-count.h
29521 (RDIV): Move to profile-count.h
29522 * bb-reorder.c (max_entry_count): Turn to profile_count.
29523 (find_traces): Update.
29524 (rotate_loop):Update.
29525 (connect_traces):Update.
29526 (sanitize_hot_paths):Update.
29527 * bt-load.c (migrate_btr_defs): Update.
29528 * cfg.c (RDIV): Remove.
29529 (init_flow): Use alloc_block.
29530 (alloc_block): Uninitialize count.
29531 (unchecked_make_edge): Uninitialize count.
29532 (check_bb_profile): Update.
29533 (dump_edge_info): Update.
29534 (dump_bb_info): Update.
29535 (update_bb_profile_for_threading): Update.
29536 (scale_bbs_frequencies_int): Update.
29537 (scale_bbs_frequencies_gcov_type): Update.
29538 (scale_bbs_frequencies_profile_count): New.
29539 * cfg.h (update_bb_profile_for_threading): Update.
29540 (scale_bbs_frequencies_profile_count): Declare.
29541 * cfgbuild.c (compute_outgoing_frequencies): Update.
29542 (find_many_sub_basic_blocks): Update.
29543 * cfgcleanup.c (try_forward_edges): Update.
29544 (try_crossjump_to_edge): Update.
29545 * cfgexpand.c (expand_gimple_tailcall): Update.
29546 (construct_exit_block): Update.
29547 * cfghooks.c (verify_flow_info): Update.
29548 (dump_bb_for_graph): Update.
29549 (split_edge): Update.
29550 (make_forwarder_block): Update.
29551 (duplicate_block): Update.
29552 (account_profile_record): Update.
29553 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29554 (get_estimated_loop_iterations): Update.
29555 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29556 (single_likely_exit): Update.
29557 * cfgloopmanip.c (scale_loop_profile): Update.
29559 (set_zero_probability): Update.
29560 (lv_adjust_loop_entry_edge): Update.
29561 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29562 (purge_dead_edges): Update.
29563 (rtl_account_profile_record): Update.
29564 * cgraph.c (cgraph_node::create): Uninitialize count.
29565 (symbol_table::create_edge): Uninitialize count.
29566 (cgraph_update_edges_for_call_stmt_node): Update.
29567 (cgraph_edge::dump_edge_flags): Update.
29568 (cgraph_node::dump): Update.
29569 (cgraph_edge::maybe_hot_p): Update.
29570 * cgraph.h: Include profile-count.h
29571 (create_clone), create_edge, create_indirect_edge): Update.
29572 (cgraph_node): Turn count to profile_count.
29573 (cgraph_edge0: Likewise.
29574 (make_speculative, clone): Update.
29575 (create_edge): Update.
29576 (init_lowered_empty_function): Update.
29577 * cgraphclones.c (cgraph_edge::clone): Update.
29578 (duplicate_thunk_for_node): Update.
29579 (cgraph_node::create_clone): Update.
29580 * cgraphunit.c (cgraph_node::analyze): Update.
29581 (cgraph_node::expand_thunk): Update.
29582 * final.c (dump_basic_block_info): Update.
29583 * gimple-streamer-in.c (input_bb): Update.
29584 * gimple-streamer-out.c (output_bb): Update.
29585 * graphite.c (print_global_statistics): Update.
29586 (print_graphite_scop_statistics): Update.
29587 * hsa-brig.c: Include basic-block.h.
29588 * hsa-dump.c: Include basic-block.h.
29589 * hsa-gen.c (T sum_slice): Update.
29590 (convert_switch_statements):Update.
29591 * hsa-regalloc.c: Include basic-block.h.
29592 * ipa-chkp.c (chkp_produce_thunks): Update.
29593 * ipa-cp.c (struct caller_statistics): Update.
29594 (init_caller_stats): Update.
29595 (gather_caller_stats): Update.
29596 (ipcp_cloning_candidate_p): Update.
29597 (good_cloning_opportunity_p): Update.
29598 (get_info_about_necessary_edges): Update.
29599 (dump_profile_updates): Update.
29600 (update_profiling_info): Update.
29601 (update_specialized_profile): Update.
29602 (perhaps_add_new_callers): Update.
29603 (decide_about_value): Update.
29604 (ipa_cp_c_finalize): Update.
29605 * ipa-devirt.c (struct odr_type_warn_count): Update.
29606 (struct decl_warn_count): Update.
29607 (struct final_warning_record): Update.
29608 (possible_polymorphic_call_targets): Update.
29609 (ipa_devirt): Update.
29610 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29611 * ipa-icf.c (sem_function::merge): Update.
29612 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29613 * ipa-inline.c (compute_uninlined_call_time): Update.
29614 (compute_inlined_call_time): Update.
29615 (want_inline_small_function_p): Update.
29616 (want_inline_self_recursive_call_p): Update.
29617 (edge_badness): Update.
29618 (lookup_recursive_calls): Update.
29619 (recursive_inlining): Update.
29620 (inline_small_functions): Update.
29621 (dump_overall_stats): Update.
29622 (dump_inline_stats): Update.
29623 * ipa-profile.c (ipa_profile_generate_summary): Update.
29624 (ipa_propagate_frequency): Update.
29625 (ipa_profile): Update.
29626 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29627 * ipa-utils.c (ipa_merge_profiles): Update.
29628 * loop-doloop.c (doloop_modify): Update.
29629 * loop-unroll.c (report_unroll): Update.
29630 (unroll_loop_runtime_iterations): Update.
29631 * lto-cgraph.c (lto_output_edge): Update.
29632 (lto_output_node): Update.
29633 (input_node): Update.
29634 (input_edge): Update.
29635 (merge_profile_summaries): Update.
29636 * lto-streamer-in.c (input_cfg): Update.
29637 * lto-streamer-out.c (output_cfg): Update.
29638 * mcf.c (create_fixup_graph): Update.
29639 (adjust_cfg_counts): Update.
29640 (sum_edge_counts): Update.
29641 * modulo-sched.c (sms_schedule): Update.
29642 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29643 * predict.c (maybe_hot_count_p): Update.
29644 (probably_never_executed): Update.
29645 (dump_prediction): Update.
29646 (combine_predictions_for_bb): Update.
29647 (propagate_freq): Update.
29648 (handle_missing_profiles): Update.
29649 (counts_to_freqs): Update.
29650 (rebuild_frequencies): Update.
29651 (force_edge_cold): Update.
29652 * predict.h: Include profile-count.h
29653 (maybe_hot_count_p, counts_to_freqs): UPdate.
29654 * print-rtl-function.c: Do not include cfg.h
29655 * print-rtl.c: Include basic-block.h
29656 * profile-count.c: New file.
29657 * profile-count.h: New file.
29658 * profile.c (is_edge_inconsistent): Update.
29659 (correct_negative_edge_counts): Update.
29660 (is_inconsistent): Update.
29661 (set_bb_counts): Update.
29662 (read_profile_edge_counts): Update.
29663 (compute_frequency_overlap): Update.
29664 (compute_branch_probabilities): Update; Initialize and deinitialize
29666 (branch_prob): Update.
29667 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29668 (edge_gcov_count): New.
29669 (bb_gcov_count): New.
29670 * shrink-wrap.c (try_shrink_wrapping): Update.
29671 * tracer.c (better_p): Update.
29672 * trans-mem.c (expand_transaction): Update.
29673 (ipa_tm_insert_irr_call): Update.
29674 (ipa_tm_insert_gettmclone_call): Update.
29675 * tree-call-cdce.c: Update.
29676 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29677 (gimple_duplicate_sese_tail): Update.
29678 (gimple_account_profile_record): Update.
29679 (execute_fixup_cfg): Update.
29680 * tree-inline.c (copy_bb): Update.
29681 (copy_edges_for_bb): Update.
29682 (initialize_cfun): Update.
29683 (freqs_to_counts): Update.
29684 (copy_cfg_body): Update.
29685 (expand_call_inline): Update.
29686 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29687 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29688 (try_unroll_loop_completely): Update.
29689 (try_peel_loop): Update.
29690 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29691 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29692 * tree-ssa-loop-split.c (connect_loops): Update.
29693 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29694 * tree-ssa-reassoc.c (branch_fixup): Update.
29695 * tree-ssa-tail-merge.c (replace_block_by): Update.
29696 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29697 (compute_path_counts): Update.
29698 (update_profile): Update.
29699 (recompute_probabilities): Update.
29700 (update_joiner_offpath_counts): Update.
29701 (estimated_freqs_path): Update.
29702 (freqs_to_counts_path): Update.
29703 (clear_counts_path): Update.
29704 (ssa_fix_duplicate_block_edges): Update.
29705 (duplicate_thread_path): Update.
29706 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29707 (struct switch_conv_info): Update.
29708 * tree-tailcall.c (decrease_profile): Update.
29709 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29710 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29711 * value-prof.c (check_counter): Update.
29712 (gimple_divmod_fixed_value): Update.
29713 (gimple_mod_pow2): Update.
29714 (gimple_mod_subtract): Update.
29715 (gimple_ic_transform): Update.
29716 (gimple_stringop_fixed_value): Update.
29717 * value-prof.h (gimple_ic): Update.
29719 2017-06-02 Carl Love <cel@us.ibm.com>
29721 * config/rs6000/rs6000-c: Add support for built-in functions
29722 vector double vec_doublee (vector signed int);
29723 vector double vec_doublee (vector unsigned int);
29724 vector double vec_doublee (vector float);
29725 vector double vec_doubleh (vector signed int);
29726 vector double vec_doubleh (vector unsigned int);
29727 vector double vec_doubleh (vector float);
29728 vector double vec_doublel (vector signed int);
29729 vector double vec_doublel (vector unsigned int);
29730 vector double vec_doublel (vector float);
29731 vector double vec_doubleo (vector signed int);
29732 vector double vec_doubleo (vector unsigned int);
29733 vector double vec_doubleo (vector float);.
29734 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29735 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29737 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29738 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29739 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29741 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29742 vec_doublel, vec_doubleh.
29743 * doc/extend.texi: Update the built-in documentation file for the
29744 new built-in functions.
29746 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29749 * ipa-inline-analysis.c (free_growth_caches): Set
29750 edge_removal_hook_holder to NULL after removing it.
29752 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29754 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29755 comparision with zero.
29757 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29758 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29759 for early expansion of vec_min and vec_max builtins.
29760 (builtin_function_type): Add min/max unsigned variants to those
29761 identified as having unsigned arguments.
29763 2017-06-02 Olivier Hainque <hainque@adacore.com>
29765 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29767 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29769 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29770 Use VALL_F16 iterator rather than VALL.
29772 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29774 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29775 Emit CBNZ inside loop when doing a strong exchange and comparing
29776 against zero. Generate the CC flags after the loop.
29778 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29780 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29781 (dl_section_ref): New.
29782 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29783 On AIX, append an expression to subtract the size of the
29784 section length to dl_section_ref.
29786 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29788 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29789 for early expansion of vector absolute builtins.
29791 2017-06-02 Richard Biener <rguenther@suse.de>
29793 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29794 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29796 2017-06-02 Richard Biener <rguenther@suse.de>
29798 PR tree-optimization/80948
29799 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29800 stmt order as well.
29802 2017-06-02 Richard Biener <rguenther@suse.de>
29804 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29806 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29807 for inductions as relevant.
29809 2017-06-02 Richard Biener <rguenther@suse.de>
29811 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29812 (vectorizable_induction): ... this. Remove dead code.
29814 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29816 * builtins. (expand_builtin_alloca): Remove second parameter and
29817 infer its value from the first parameter instead.
29818 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29820 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29822 PR rtl-optimization/80903
29823 * loop-doloop.c (add_test): Unshare sequence.
29825 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29827 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29829 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29831 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29833 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29834 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29835 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29836 (xlogue_layout::get_stub_rtx): Make static.
29837 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29838 (xlogue_layout::compute_stub_managed_regs): Rename to...
29839 (xlogue_layout::count_stub_managed_regs): ...this.
29840 (xlogue_layout::is_stub_managed_reg): New function.
29841 (xlogue_layout::m_stub_names): Rename to...
29842 (xlogue_layout::s_stub_names): ...this, make static.
29843 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29844 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29845 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29846 xlogue_layout::s_stub_names): Instantiate statics.
29847 (stub_managed_regs): Remove.
29848 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29849 (disable_call_ms2sysv_xlogues): Rename to...
29850 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29851 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29853 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29854 change after reload_completed.
29855 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29857 (ix86_expand_prologue): Likewise.
29858 (ix86_expand_epilogue): Likewise.
29859 (ix86_expand_split_stack_prologue): Likewise.
29860 (ix86_compute_frame_layout): Remove frame parameter ...
29861 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29862 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29864 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29865 (ix86_frame): Move from here ...
29866 * config/i386/i386.h (ix86_frame): ... to here.
29867 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29868 complete ix86_frame data structure instead. Remove some_ld_name.
29870 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29872 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29873 symbols that hold a DECL_VALUE_EXPR.
29875 2017-06-01 Martin Jambor <mjambor@suse.cz>
29877 PR tree-optimization/80898
29878 * tree-sra.c (process_subtree_disqualification): Removed.
29879 (disqualify_candidate): Do not acll
29880 process_subtree_disqualification.
29881 (subtree_mark_written_and_enqueue): New function.
29882 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29883 RHS has been disqualified and re-queue LHS if necessary. Apart
29884 from that, ignore disqualified RHS.
29886 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29888 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29889 address fetch for z10 or later.
29891 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29893 * config/arc/arc.md (tst_movb): Add guard when splitting.
29895 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29897 * config/arc/arc.c (arc_can_eliminate): Test against
29898 arc_frame_pointer_needed.
29900 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29902 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29903 to prevent store reordering.
29904 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29905 (type): Add block type.
29906 (stack_tie): Define special instruction to be used in
29909 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29911 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29912 constraint. It is not valid for the pattern.
29913 (noncommutative_binary_comparison): Likewise.
29915 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29917 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29920 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29922 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29923 be used by the reg-alloc.
29925 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29927 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29928 reg-alloc when having mul64 or mul32x16 instructions.
29929 (mulsidi3): Likewise.
29930 (umulsidi3): Likewise.
29931 (mulsi32x16): New pattern.
29932 (mulsi64): Likewise.
29933 (mulsidi64): Likewise.
29934 (umulsidi64): Likewise.
29935 (MUL32x16_REG): Define.
29936 (mul64_600): Use MUL32x16_REG.
29937 (mac64_600): Likewise.
29938 (umul64_600): Likewise.
29939 (umac64_600): Likewise.
29941 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29943 * config/arc/arc.md (mulsi3_700): Make it commutative.
29945 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29947 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29949 (*sign_extendsidi2_insn): Likewise for movstosw.
29951 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29953 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29954 the type of the input discriminant value. Convert the
29955 discriminant value of signedness vary.
29957 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29959 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29960 Add to -Wall section.
29962 2017-06-01 Richard Biener <rguenther@suse.de>
29964 PR middle-end/66313
29965 * fold-const.c (fold_plusminus_mult_expr): If the factored
29966 factor may be zero use a wrapping type for the inner operation.
29967 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29968 and handle moved defs.
29969 (process_assignment): Properly guard the unary op case. Return a
29970 tri-state indicating that moving the stmt before the call may allow
29971 to continue. Pass through to_move.
29972 (find_tail_calls): Handle moving unrelated defs before
29975 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29978 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29979 splitter result in the canonical way.
29981 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29983 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29984 also for 32bit target. Update insn attributes.
29985 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29987 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29989 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29990 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29991 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29992 (_mm_maskz_min_ss): New intrinsics.
29994 2017-05-31 Martin Liska <mliska@suse.cz>
29996 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29997 Change comment style to one we normally use.
29998 (vectorizable_reduction): Likewise.
29999 (vectorizable_induction): Likewise.
30000 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
30001 (vectorizable_call): Likewise.
30002 (vectorizable_simd_clone_call): Likewise.
30003 (vectorizable_conversion): Likewise.
30004 (vectorizable_assignment): Likewise.
30005 (vectorizable_shift): Likewise.
30006 (vectorizable_operation): Likewise.
30007 (vectorizable_store): Likewise.
30008 (vectorizable_load): Likewise.
30009 * tree-vectorizer.h: Likewise.
30011 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
30013 * passes.c (emergency_dump_function): New.
30014 * tree-pass.h (emergency_dump_function): Declare.
30015 * plugin.c (plugins_internal_error_function): Remove.
30016 * plugin.h (plugins_internal_error_function): Remove declaration.
30017 * toplev.c (internal_error_function): New static function. Use it...
30018 (general_init): ...here.
30020 2017-05-31 Graham Markall <graham.markall@embecosm.com>
30022 * config/arc/arc.c (arc_print_operand): Handle constant operands.
30023 (arc_rtx_costs): Add costs for new patterns.
30024 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
30025 * config/arc/predicates.md: Add _1_2_3_operand predicate.
30027 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30029 * tree-ssa-strlen.c (get_next_strinfo): New function.
30030 (get_stridx_plus_constant): Use it.
30031 (zero_length_string): Likewise.
30032 (adjust_related_strinfos): Likewise.
30033 (adjust_last_stmt): Likewise.
30035 2017-05-31 Richard Biener <rguenther@suse.de>
30038 * config/i386/i386.c (ix86_expand_builtin): Remove assert
30039 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
30041 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30043 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
30044 loop_vinfo argument and use of dependence distance vectors.
30045 Check instead whether the two references differ only in their
30046 initial value and assume that they have the same alignment if the
30047 difference is a multiple of the vector alignment.
30048 (vect_analyze_data_refs_alignment): Update call accordingly.
30050 2017-05-31 Martin Liska <mliska@suse.cz>
30053 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
30055 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30057 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
30058 (create_intersect_range_checks): Move from ...
30059 * tree-data-ref.c (create_intersect_range_checks_index)
30060 (create_intersect_range_checks): ... to here.
30061 (create_runtime_alias_checks): New function factored from ...
30062 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30063 here. Call above function.
30064 * tree-data-ref.h (create_runtime_alias_checks): New function.
30066 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30068 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30069 segment length for dr_b and compute it in wide_int.
30071 2017-05-31 Richard Biener <rguenther@suse.de>
30073 PR tree-optimization/80906
30074 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30075 and pass through iv_map.
30076 (copy_bb_and_scalar_dependences): Adjust.
30077 (translate_pending_phi_nodes): Likewise.
30078 (copy_loop_close_phi_args): Handle code-generating IVs instead
30081 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30083 * diagnostic-color.c (color_dict): Add "type-diff".
30084 (parse_gcc_colors): Update comment.
30085 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30086 -fdiagnostics-show-template-tree and -fno-elide-type.
30087 (GCC_COLORS): Add type-diff to example.
30089 (-fdiagnostics-show-template-tree): New.
30090 (-fno-elide-type): New.
30091 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30092 the pp_format_decoder callback. Call any m_format_postprocessor's
30094 (pretty_printer::pretty_printer): Initialize
30095 m_format_postprocessor.
30096 (pretty_printer::~pretty_printer): Delete any
30097 m_format_postprocessor.
30098 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30099 (class format_postprocessor): New class.
30100 (struct pretty_printer::format_decoder): Document the new parameters.
30101 (struct pretty_printer::m_format_postprocessor): New field.
30102 * tree-diagnostic.c (default_tree_printer): Update for new
30103 bool and const char ** params.
30104 * tree-diagnostic.h (default_tree_printer): Likewise.
30106 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30108 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30109 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30111 rs6000_gen_cell_microcode code.
30112 (rs6000_final_prescan_insn): Delete.
30113 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30114 "warn-cell-microcode" entries.
30115 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30116 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30117 throughout. Change cc_reg_not_micro_cr0_operand to
30118 cc_reg_not_cr0_operand throughout.
30119 (*extendhi<mode>2_noload): Delete.
30120 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30121 (mwarn-cell-microcode): Delete.
30122 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30123 -mgen-cell-microcode and -mwarn-cell-microcode.
30125 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30128 * config/i386/constraints.md (Yd): New constraint.
30130 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30131 and (?Yd, r) alternatives. Update insn attributes.
30132 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30133 and (?*Yd, r) alternatives. Update insn attributes.
30134 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30136 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30138 * gimplify.c (gimplify_modify_expr): Don't create a
30139 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30142 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30144 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30146 2017-05-30 Richard Biener <rguenther@suse.de>
30148 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30149 and reduc_def fields.
30150 (STMT_VINFO_REDUC_TYPE): New define.
30151 (STMT_VINFO_REDUC_DEF): Likewise.
30152 (vect_force_simple_reduction): Adjust prototype.
30153 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30154 (vect_is_simple_reduction): Remove check_reduction argument.
30155 (vect_force_simple_reduction): Adjust and set
30156 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30157 (vectorizable_reduction): Do not re-do reduction analysis
30158 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30159 * tree-parloops.c (gather_scalar_reductions): Adjust.
30161 2017-05-30 Richard Biener <rguenther@suse.de>
30163 PR middle-end/80901
30164 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30167 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30169 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30170 Introduce unknown_misalignment parameter and remove vf.
30171 (vect_peeling_hash_get_lowest_cost):
30172 Pass unknown_misalignment parameter.
30173 (vect_enhance_data_refs_alignment):
30174 Fix unsupportable data ref treatment.
30176 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30178 * tree-vect-data-refs.c (vect_get_data_access_cost):
30179 Workaround for SLP handling.
30180 (vect_enhance_data_refs_alignment):
30181 Compute costs for doing no peeling at all, compare to the best
30182 peeling costs so far and avoid peeling if cheaper.
30184 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30186 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30187 Return peeling info and set costs to zero for unlimited cost
30189 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30190 with unknown misalignment. Compute and costs for unknown
30191 misalignment, compare them to the costs for known misalignment
30192 and choose the cheapest for peeling.
30194 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30196 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30197 (vect_get_peeling_costs_all_drs): Create function.
30198 (vect_peeling_hash_get_lowest_cost):
30199 Use vect_get_peeling_costs_all_drs.
30200 (vect_peeling_supportable): Create function.
30201 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30203 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30205 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30206 DR_HAS_NEGATIVE_STEP.
30207 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30208 (vect_enhance_data_refs_alignment): Use.
30209 (vect_duplicate_ssa_name_ptr_info): Use.
30210 * tree-vectorizer.h (dr_misalignment): Use.
30211 (known_alignment_for_access_p): Use.
30213 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30216 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30218 (has_section_name): New function.
30220 2017-05-30 Martin Liska <mliska@suse.cz>
30223 * auto-profile.c (get_function_decl_from_block): Fix
30226 2017-05-30 Richard Biener <rguenther@suse.de>
30228 PR middle-end/80876
30229 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30231 2017-05-30 Martin Liska <mliska@suse.cz>
30233 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30234 * dumpfile.h (struct dump_file_info): Remove ctors.
30236 2017-05-30 Martin Liska <mliska@suse.cz>
30238 * predict.def: Fix GNU coding style.
30240 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30242 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30243 Mark 'to' argument with ATTRIBUTE_UNUSED.
30245 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30247 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30248 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30249 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30252 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30254 * doc/install.texi (Options specification): Restore entry of
30255 --enable-sjlj-exceptions.
30257 2017-05-27 Michael Eager <eager@eagercon.com>
30260 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30262 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30264 * config/microblaze/microblaze.h
30265 (FIXED_REGISTERS): Update in macro.
30266 (CALL_USED_REGISTERS): Update in macro.
30268 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30270 * doc/install.texi: Add links to macOS binary distributions.
30272 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30276 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30278 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30280 2017-05-26 Martin Liska <mliska@suse.cz>
30282 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30284 2017-05-26 Martin Liska <mliska@suse.cz>
30286 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30287 always leading ';; '.
30288 (dump_bb_info): Likewise.
30289 (brief_dump_cfg): Likewise.
30290 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30291 * dumpfile.c: Remove usage of TDF_VERBOSE.
30292 * dumpfile.h (enum dump_kind): Likewise.
30293 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30294 * print-tree.c (debug_verbose): Remove.
30295 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30296 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30297 * tree-diagnostic.c (default_tree_printer): Replace
30298 TDF_DIAGNOSTIC with TDF_SLIM.
30300 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30302 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30303 in parameter loop, rather than loop_vinfo.
30304 (create_intersect_range_checks): Ditto.
30305 (vect_create_cond_for_alias_checks): Update call to above functions.
30307 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30309 PR tree-optimization/80815
30310 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30311 for merging runtime alias checks. Handle negative DR_STEPs.
30313 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30315 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30317 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30318 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30319 out code pruning runtime alias checks.
30320 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30321 factored out from above.
30322 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30324 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30326 (prune_runtime_alias_test_list): New decalaration.
30328 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30330 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30331 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30332 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30333 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30334 (operator==, comp_dr_with_seg_len_pair): Ditto.
30335 (vect_prune_runtime_alias_test_list): Ditto.
30337 2017-05-26 Martin Liska <mliska@suse.cz>
30340 * params.def: Bound partial-inlining-entry-probability param.
30342 2017-05-26 Marek Polacek <polacek@redhat.com>
30345 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30348 2017-05-26 Richard Biener <rguenther@suse.de>
30350 PR tree-optimization/80842
30351 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30354 2017-05-26 Richard Biener <rguenther@suse.de>
30356 PR tree-optimization/80844
30357 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30359 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30361 * doc/md.texi (Machine Constraints): Update x86 family
30362 machine constraints section to match 'config/i386/constraints.md'.
30364 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30366 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30368 2017-05-25 Nathan Sidwell <nathan@acm.org>
30370 * doc/invoke.texi (--enable-languages): Update documentation.
30372 2017-05-25 Martin Liska <mliska@suse.cz>
30374 * dumpfile.c: Add TDF_FOLDING.
30375 * dumpfile.h (enum dump_kind): Likewise.
30376 * genmatch.c (dt_simplify::gen_1): Use it.
30378 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30380 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30382 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30384 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30385 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30387 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30389 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30390 * match.pd (X == C): Rewrite it here.
30391 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30392 with_certain_nonzero_bits2): New predicates.
30393 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30395 2017-05-24 Nathan Sidwell <nathan@acm.org>
30397 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30400 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30403 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30405 * config/powerpcspe: New port. Files are copied from the rs6000
30406 port, with "rs6000" in filenames replaced by "powerpcspe".
30408 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30410 PR rtl-optimization/80754
30411 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30413 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30415 * config/sparc/sparc.md (length): Return the correct value for -mflat
30416 sibcalls to match output_sibcall.
30418 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30422 * config/rs6000/rs6000.c (struct machine_function): Add new field
30424 (rs6000_get_separate_components): Init that field, use it.
30425 (rs6000_components_for_bb): Use the field.
30427 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30429 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30431 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30433 PR middle-end/80823
30434 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30436 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30439 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30440 address against address_operand predicate.
30441 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30443 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30445 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30446 parameters passed indirectly.
30448 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30450 * config/i386/i386.md (*movdi_internal): Remove SSE4
30451 alternative 18 (?r, *v). Update insn attributes.
30452 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30453 Update insn attributes.
30454 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30455 Update insn attributes.
30456 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30457 alternative 1 (r, v). Remove isa attribute.
30458 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30459 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30460 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30462 2017-05-23 Tom de Vries <tom@codesourcery.com>
30464 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30467 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30469 * cgraphunit.c (symbol_table::process_new_functions): Update.
30470 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30471 (inline_generate_summary): Rename to ...
30472 (ipa_fn_summary_generate): ... this one.
30473 (inline_read_summary): Rename to ...
30474 (ipa_fn_summary_read): ... this one.
30475 (inline_write_summary): Rename to ...
30476 (ipa_fn_summary_write): ... this one.
30477 (inline_free_summary): Rename to ...
30478 (ipa_free_fn_summary): ... this one.
30479 (pass_data_local_fn_summary, pass_local_fn_summary,
30480 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30481 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30482 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30483 make_pass_ipa_fn_summary): New.
30484 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30485 inline_write_summary, inline_free_summary): Remove.
30486 (ipa_free_fn_summary) : New.
30487 * ipa-inline.c (ipa_inline): Update.
30488 (pass_ipa_inline): Do not generate summaries.
30489 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30491 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30492 and add pass_ipa_fn_summary.
30493 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30495 (make_pass_inline_parameters): Remove.
30497 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30499 * omp-low.c (struct omp_context): Remove "default_kind" member.
30502 * omp-offload.c (execute_oacc_device_lower): Remove the
30503 parallelism dimensions function attributes for unparallelized
30504 OpenACC kernels constructs.
30506 2017-05-23 Martin Liska <mliska@suse.cz>
30508 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30510 (cgraph_edge::make_speculative): Likewise.
30511 (cgraph_edge::resolve_speculation): Likewise.
30512 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30513 (cgraph_node::dump): Likewise.
30514 * cgraph.h: Likewise.
30515 * cgraphunit.c (analyze_functions): Likewise.
30516 (symbol_table::compile): Likewise.
30517 * ipa-cp.c (print_all_lattices): Likewise.
30518 (determine_versionability): Likewise.
30519 (initialize_node_lattices): Likewise.
30520 (ipcp_verify_propagated_values): Likewise.
30521 (estimate_local_effects): Likewise.
30522 (update_profiling_info): Likewise.
30523 (create_specialized_node): Likewise.
30524 (perhaps_add_new_callers): Likewise.
30525 (decide_about_value): Likewise.
30526 (decide_whether_version_node): Likewise.
30527 (identify_dead_nodes): Likewise.
30528 (ipcp_store_bits_results): Likewise.
30529 * ipa-devirt.c (dump_targets): Likewise.
30530 (ipa_devirt): Likewise.
30531 * ipa-icf.c (sem_item::dump): Likewise.
30532 (sem_function::equals): Likewise.
30533 (sem_variable::equals): Likewise.
30534 (sem_item_optimizer::read_section): Likewise.
30535 (sem_item_optimizer::execute): Likewise.
30536 (congruence_class::dump): Likewise.
30537 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30538 (dump_inline_summary): Likewise.
30539 (estimate_node_size_and_time): Likewise.
30540 (inline_analyze_function): Likewise.
30541 * ipa-inline-transform.c (inline_call): Likewise.
30542 * ipa-inline.c (report_inline_failed_reason): Likewise.
30543 (want_early_inline_function_p): Likewise.
30544 (edge_badness): Likewise.
30545 (update_edge_key): Likewise.
30546 (inline_small_functions): Likewise.
30547 * ipa-profile.c (ipa_profile): Likewise.
30548 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30549 (ipa_make_edge_direct_to_target): Likewise.
30550 (remove_described_reference): Likewise.
30551 (ipa_impossible_devirt_target): Likewise.
30552 (propagate_controlled_uses): Likewise.
30553 (ipa_print_node_params): Likewise.
30554 (ipcp_transform_function): Likewise.
30555 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30556 (propagate_pure_const): Likewise.
30557 * ipa-reference.c (generate_summary): Likewise.
30558 (read_write_all_from_decl): Likewise.
30559 (propagate): Likewise.
30560 (ipa_reference_read_optimization_summary): Likewise.
30561 * ipa-utils.c (ipa_merge_profiles): Likewise.
30562 * ipa.c (walk_polymorphic_call_targets): Likewise.
30563 (symbol_table::remove_unreachable_nodes): Likewise.
30564 (ipa_single_use): Likewise.
30565 * passes.c (execute_todo): Likewise.
30566 * predict.c (drop_profile): Likewise.
30567 * symtab.c (symtab_node::get_dump_name): New function.
30568 (symtab_node::dump_name): Likewise.
30569 (symtab_node::dump_asm_name): Likewise.
30570 (symtab_node::dump_references): Likewise.
30571 (symtab_node::dump_referring): Likewise.
30572 (symtab_node::dump_base): Likewise.
30573 (symtab_node::debug_symtab): Likewise.
30574 * tree-sra.c (convert_callers_for_node): Likewise.
30575 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30576 * value-prof.c (init_node_map): Likewise.
30578 2017-05-23 Martin Liska <mliska@suse.cz>
30580 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30581 and symtab_node::debug_symtab to symbol_table::debug.
30582 * cgraphunit.c (analyze_functions): Use the renamed function.
30583 (symbol_table::compile): Likewise.
30584 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30585 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30586 * passes.c (execute_todo): Likewise.
30587 * symtab.c (symbol_table::dump): New function.
30588 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30590 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30592 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30593 that nonconst implies exec.
30595 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30597 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30598 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30599 (inline_edge_summary_vec): Turn into ...
30600 (ipa_call_summaries): ... this one.
30601 (redirect_to_unreachable, edge_set_predicate,
30602 evaluate_properties_for_edge, inline_summary_alloc,
30603 reset_ipa_call_summary, reset_inline_summary,
30604 inline_summary_t::duplicate): Update.
30605 (inline_edge_duplication_hook): Turn to ...
30606 (ipa_call_summary_t::duplicate): ... this one.
30607 (inline_edge_removal_hook): Turn to ...
30608 (ipa_call_summary_t::remove): ... this one.
30609 (dump_inline_edge_summary): Turn to ...
30610 (dump_ipa_call_summary): ... this one.
30611 (estimate_function_body_sizes): Update.
30612 (inline_update_callee_summaries): Update.
30613 (remap_edge_change_prob): Update.
30614 (remap_edge_summaries): Update.
30615 (inline_merge_summary): Update.
30616 (do_estimate_edge_time): Update.
30617 (inline_generate_summary): Update.
30618 (inline_read_section): Update.
30619 (inline_read_summary): Update.
30620 (inline_free_summary): Update.
30621 * ipa-inline.c (can_inline_edge_p): Update.
30622 (compute_inlined_call_time): Update.
30623 (want_inline_small_function_p): Update.
30624 (edge_badness): Update.
30625 (early_inliner): Update.
30626 * ipa-inline.h (inline_edge_summary): Turn to ...
30627 (ipa_call_summary): ... this one.
30628 (ipa_call_summary_t): New class.
30629 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30630 (ipa_call_summaries): New.
30631 (inline_edge_summary): Remove.
30632 (estimate_edge_growth): Update.
30633 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30634 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30635 * ipa-split.c (execute_split_functions): Update.
30636 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30638 2017-05-23 Tom de Vries <tom@codesourcery.com>
30640 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30641 attributes): Document rdrand effective target.
30643 2017-05-23 Tom de Vries <tom@codesourcery.com>
30645 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30646 attributes): Sort alphabetically.
30648 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30650 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30652 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30655 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30656 V2DF/V2DI splat into two separate patterns, one that handles
30657 registers, and the other that only handles memory. Drop support
30658 for splatting from a GPR on ISA 2.07 and then splitting the
30659 splat into direct move and splat.
30660 (vsx_splat_<mode>_reg): Likewise.
30661 (vsx_splat_<mode>_mem): Likewise.
30663 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30665 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30667 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30669 PR middle-end/80809
30670 * omp-low.c (finish_taskreg_remap): New function.
30671 (finish_taskreg_scan): If unit size of ctx->record_type
30672 is non-constant, unshare the size expression and replace
30673 decls in it with possible outer var refs.
30675 PR middle-end/80809
30676 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30677 GOVD_SHARED rather than GOVD_PRIVATE with it.
30678 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30679 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30681 PR middle-end/80853
30682 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30683 as last argument to build_outer_var_ref for pointer bases of array
30684 section reductions.
30686 2017-05-19 Martin Sebor <msebor@redhat.com>
30688 * print-tree.c (print_node): Print DECL_READ_P flag.
30690 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30692 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30693 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30694 * cgraph.c: Likewise.
30695 * cgraphunit.c: Likewise.
30696 * gengtype.c: Likewise.
30697 * ipa-cp.c: Likewise.
30698 * ipa-devirt.c: Likewise.
30699 * ipa-icf.c: Likewise.
30700 * ipa-predicate.c: Likewise.
30701 * ipa-profile.c: Likewise.
30702 * ipa-prop.c: Likewise.
30703 * ipa-split.c: Likewise.
30705 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30706 edge_predicate_pool, dump_inline_hints,
30707 inline_summary::account_size_time, redirect_to_unreachable,
30708 edge_set_predicate, set_hint_predicate,
30709 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30710 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30711 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30712 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30713 ipa_call_summary_t::remove, initialize_growth_caches,
30714 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30715 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30716 mark_modified, unmodified_parm_1, unmodified_parm,
30717 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30718 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30719 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30720 will_be_nonconstant_predicate, record_modified_bb_info,
30721 get_minimal_bb, record_modified, param_change_prob,
30722 phi_result_unknown_predicate, predicate_for_phi_result,
30723 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30724 estimate_function_body_sizes, compute_inline_parameters,
30725 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30726 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30727 inline_update_callee_summaries, remap_edge_change_prob,
30728 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30729 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30730 inline_analyze_function, inline_summary_t::insert,
30731 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30732 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30733 inline_free_summary): Move to ipa-fnsummary.h
30734 (predicate_t): Remove.
30735 * ipa-fnsummary.c: New file.
30736 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30737 (enum inline_hints_vals, inline_hints, agg_position_info,
30738 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30739 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30740 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30741 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30742 inline_read_summary, inline_write_summary, inline_free_summary,
30743 inline_analyze_function, initialize_inline_failed,
30744 inline_merge_summary, inline_update_overall_summary,
30745 compute_inline_parameters): Move to ipa-fnsummary.h
30746 * ipa-fnsummary.h: New file.
30747 * ipa-inline-transform.h: Include ipa-inline.h.
30748 * ipa-inline.c: LIkewise.
30750 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30752 * ipa-inline.c (edge_badness): Use inlined_time instead of
30753 inline_summaries->get.
30755 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30757 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30759 2017-05-22 Nathan Sidwell <nathan@acm.org>
30761 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30762 (fdump-lang): Document 'raw' option.
30763 * dumpfile.h (TDI_tu): Delete.
30764 * dumpfile.c (dump_files): Remove translation-unit.
30765 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30767 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30769 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30770 command option from $(AWK) call.
30771 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30772 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30773 [FORMAT]: Remove handling of variable.
30774 * config/avr/t-multilib: Regenerate.
30776 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30778 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30780 (dump_inline_summary): Do not print self_time.
30781 (estimate_function_body_sizes): Do not set self_time.
30782 (compute_inline_parameters): Likewise.
30783 (inline_read_section, inline_write_summary): Do not stream self_time.
30784 * ipa-inline.h (inline_summary): Drop self_time.
30786 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30788 * ipa-inline-analysis.c (account_size_time): Rename to ...
30789 (inline_summary::account_size_time): ... this one.
30790 (reset_ipa_call_summary): Turn to ...
30791 (ipa_call_summary::reset): ... this one.
30792 (reset_inline_summary): Turn to ...
30793 (inline_summary::reset): ... this one.
30794 (inline_summary_t::remove): Update.
30795 (inline_summary_t::duplicate): Update.
30796 (ipa_call_summary_t::remove): Update.
30797 (dump_inline_summary): Update.
30798 (estimate_function_body_sizes): Update.
30799 (compute_inline_parameters): Update.
30800 (estimate_node_size_and_time): Update.
30801 (inline_merge_summary): Update.
30802 (inline_update_overall_summary): Update.
30803 (inline_read_section): Update.
30804 (inline_write_summary): Update.
30805 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30806 add account_size_time and reset member functions.
30807 (ipa_call_summary): Add reset function.
30808 * ipa-predicate.h (predicate::operator &): Constify.
30810 2017-05-22 Richard Biener <rguenther@suse.de>
30812 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30814 2017-05-19 Jason Merrill <jason@redhat.com>
30816 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30818 2017-05-19 Marek Polacek <polacek@redhat.com>
30821 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30822 TYPE_OVERFLOW_WRAPS checks.
30824 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30826 * tree-core.h (enum omp_clause_default_kind): Add
30827 "OMP_CLAUSE_DEFAULT_PRESENT".
30828 * tree-pretty-print.c (dump_omp_clause): Handle it.
30829 * gimplify.c (enum gimplify_omp_var_data): Add
30830 "GOVD_MAP_FORCE_PRESENT".
30831 (gimplify_adjust_omp_clauses_1): Map it to
30832 "GOMP_MAP_FORCE_PRESENT".
30833 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30835 * gimplify.c (oacc_default_clause): Clarify.
30837 2017-05-19 Nathan Sidwell <nathan@acm.org>
30839 LANG_HOOK_REGISTER_DUMPS
30840 * toplev.c (general_init): Call register dump lang hook.
30841 * doc/invoke.texi: Document -fdump-lang option family.
30842 * dumpfile.c (dump_files): Remove class dump here.
30843 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30844 * dumpfile.h (tree_dump_index): Remove TDI_class.
30845 * langhooks-def.h (lhd_register_dumps): Declare.
30846 (LANG_HOOKS_REGISTER_DUMPS): Define.
30847 (LANG_HOOKS_INITIALIZER): Add it.
30848 * langhooks.c (lhd_register_dumps): Define.
30849 * langhooks.h (struct lang_hooks): Add register_dumps.
30851 2017-05-19 Nathan Sidwell <nathan@acm.org>
30853 * context.h (context::set_passes): New.
30854 * context.c (context::context): Do not create pass manager.
30855 * toplev.c (general_init): Create pass manager here.
30857 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30859 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30860 use this splitter if two add or or instructions would also work for
30861 the constant we want to generate.
30863 2017-05-19 Richard Biener <rguenther@suse.de>
30866 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30867 predicate evaluation.
30869 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30871 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30873 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30876 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30878 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30879 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30880 (inline_edge_summary_vec): Turn into ...
30881 (ipa_call_summaries): ... this one.
30882 (redirect_to_unreachable, edge_set_predicate,
30883 evaluate_properties_for_edge, inline_summary_alloc,
30884 reset_ipa_call_summary, reset_inline_summary,
30885 inline_summary_t::duplicate): Update.
30886 (inline_edge_duplication_hook): Turn to ...
30887 (ipa_call_summary_t::duplicate): ... this one.
30888 (inline_edge_removal_hook): Turn to ...
30889 (ipa_call_summary_t::remove): ... this one.
30890 (dump_inline_edge_summary): Turn to ...
30891 (dump_ipa_call_summary): ... this one.
30892 (estimate_function_body_sizes): Update.
30893 (inline_update_callee_summaries): Update.
30894 (remap_edge_change_prob): Update.
30895 (remap_edge_summaries): Update.
30896 (inline_merge_summary): Update.
30897 (do_estimate_edge_time): Update.
30898 (inline_generate_summary): Update.
30899 (inline_read_section): Update.
30900 (inline_read_summary): Update.
30901 (inline_free_summary): Update.
30902 * ipa-inline.c (can_inline_edge_p): Update.
30903 (compute_inlined_call_time): Update.
30904 (want_inline_small_function_p): Update.
30905 (edge_badness): Update.
30906 (early_inliner): Update.
30907 * ipa-inline.h (inline_edge_summary): Turn to ...
30908 (ipa_call_summary): ... this one.
30909 (ipa_call_summary_t): New class.
30910 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30911 (ipa_call_summaries): New.
30912 (inline_edge_summary): Remove.
30913 (estimate_edge_growth): Update.
30914 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30915 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30916 * ipa-split.c (execute_split_functions): Update.
30917 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30919 2017-05-19 Richard Biener <rguenther@suse.de>
30921 PR middle-end/80764
30922 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30924 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30926 * config/rs6000/rs6000.c (struct machine_function): Add field
30927 fpr_is_wrapped_separately.
30928 (rs6000_get_separate_components): Use 64 components. Handle the
30929 new FPR components.
30930 (rs6000_components_for_bb): Handle the FPR components.
30931 (rs6000_emit_prologue_components): Handle the FPR components.
30932 (rs6000_emit_epilogue_components): Handle the FPR components.
30933 (rs6000_set_handled_components): Handle the FPR components.
30934 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30935 that are already separately shrink-wrapped.
30936 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30937 that are already separately shrink-wrapped.
30939 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30942 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30945 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30946 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30947 cases where the register allocator uses a move and an offsettable
30948 memory operation to/from a FPR register on ISA 2.06/2.07.
30949 (define_peephole2 for Altivec d-form store): Likewise.
30951 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30954 * config/i386/mmx.md (*mov<mode>_internal): Enable
30955 alternatives 11, 12, 13 and 14 also for 32bit targets.
30956 Remove alternatives 15, 16, 17 and 18.
30957 * config/i386/sse.md (vec_concatv2di): Change
30958 alternative (!x, *y) to (x, ?!*Yn).
30960 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30962 * dumpfile.h (enum dump_kind): Remove stray comma.
30964 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30966 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30967 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30968 predicate::num_conditions
30969 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30970 (CHANGED): turn into predicate::changed.
30971 (agg_position_info): Move to ipa-predicate.h
30972 (add_condition, predicate::add_clause, predicate::operator &=,
30973 predicate::or_with, predicate::evaluate, predicate::probability,
30974 dump_condition, dump_clause, predicate::dump,
30975 predicate::remap_after_duplication, predicate::remap_after_inlining,
30976 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30977 (evaluate_conditions_for_known_args): Update.
30978 (set_cond_stmt_execution_predicate): Update.
30979 * ipa-inline.h: Include ipa-predicate.h
30980 (condition, inline_param_summary, conditions, agg_position_info,
30981 predicate): Move to ipa-predicate.h
30982 * ipa-predicate.c: New file.
30983 * ipa-predicate.h: New file.
30985 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30987 * final.c (leaf_function_p): Check we are not in a sequence.
30989 2017-05-18 Martin Liska <mliska@suse.cz>
30991 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30992 * dumpfile.c (dump_register): Use new enum dump_kind.
30993 (get_dump_file_name): Likewise.
30994 (dump_enable_all): Likewise.
30995 (dump_switch_p_1): Likewise.
30996 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30997 * dumpfile.h (enum dump_kind): New enum type.
30998 (struct dump_file_info): Create constructor and
30999 format fields and comments.
31000 * passes.c (pass_manager::register_one_dump_file):
31002 * statistics.c (statistics_early_init): Likewise.
31003 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
31004 TDF_TREE with TDF_SLIM.
31005 (gather_memory_references_ref): Likewise.
31007 2017-05-18 Martin Liska <mliska@suse.cz>
31009 * vec.h (struct vnull): Use it.
31011 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31013 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
31014 (true_predicate, false_predicate, true_predicate_p,
31015 false_predicate_p): Remove.
31016 (single_cond_predicate, not_inlined_predicate): Turn to member function
31018 (add_condition): Update.
31019 (add_clause): Turn to...
31020 (predicate::add_clause): ... this one; update; allow passing NULL
31022 (and_predicates): Turn to ...
31023 (predicate::operator &=): ... this one.
31024 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
31025 (or_predicates): Turn to ...
31026 (predicate::or_with): ... this one.
31027 (evaluate_predicate): Turn to ...
31028 (predicate::evaluate): ... this one.
31029 (predicate_probability): Turn to ...
31030 (predicate::probability): ... this one.
31031 (dump_condition): Update.
31032 (dump_predicate): Turn to ...
31033 (predicate::dump): ... this one.
31034 (account_size_time): Update.
31035 (edge_set_predicate): Update.
31036 (set_hint_predicate): UPdate.
31037 (evaluate_conditions_for_known_args): Update.
31038 (evaluate_properties_for_edge): Update.
31039 (remap_predicate_after_duplication): Turn to...
31040 (predicate::remap_after_duplication): ... this one.
31041 (remap_hint_predicate_after_duplication): Update.
31042 (inline_summary_t::duplicate): UPdate.
31043 (dump_inline_edge_summary): Update.
31044 (dump_inline_summary): Update.
31045 (set_cond_stmt_execution_predicate): Update.
31046 (set_switch_stmt_execution_predicate): Update.
31047 (compute_bb_predicates): Update.
31048 (will_be_nonconstant_expr_predicate): Update.
31049 (will_be_nonconstant_predicate): Update.
31050 (phi_result_unknown_predicate): Update.
31051 (predicate_for_phi_result): Update.
31052 (array_index_predicate): Update.
31053 (estimate_function_body_sizes): Update.
31054 (estimate_node_size_and_time): Update.
31055 (estimate_ipcp_clone_size_and_time): Update.
31056 (remap_predicate): Rename to ...
31057 (predicate::remap_after_inlining): ... this one.
31058 (remap_hint_predicate): Update.
31059 (inline_merge_summary): Update.
31060 (inline_update_overall_summary): Update.
31061 (estimate_size_after_inlining): Update.
31062 (read_predicate): Rename to ...
31063 (predicate::stream_in): ... this one.
31064 (read_inline_edge_summary): Update.
31065 (write_predicate): Rename to ...
31066 (predicate::stream_out): ... this one.
31067 (write_inline_edge_summary): Update.
31068 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31069 (clause_t): Turn to uint32_t
31070 (predicate): Turn to class; implement constructor and operators
31072 (size_time_entry): Update.
31073 (inline_summary): Update.
31074 (inline_edge_summary): Update.
31076 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31078 * fold-const.c (fold_binary_loc): Move transformation...
31079 * match.pd (C - X CMP X): ... here.
31081 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31083 * config/sparc/sparc.c (sparc_option_override): Set function
31084 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31085 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31087 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31089 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31091 2017-05-18 Marek Polacek <polacek@redhat.com>
31094 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31095 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31096 gimple_assign_load_p.
31098 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31100 PR middle-end/80692
31101 * real.c (do_compare): Give decimal_do_compare preference over
31102 comparing just the signs.
31104 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31106 * doc/md.texi (Canonicalization of Instructions): Describe the
31107 canonical form of instructions that inherently set a condition
31110 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31112 PR middle-end/80775
31113 * tree-cfg.c: Move deletion of unreachable case statements to after
31114 the merging of consecutive case labels.
31116 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31118 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31119 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31120 restoring of callee-saved registers.
31122 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31124 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31125 * config/visium/visium.c (single_set_and_flags): Likewise.
31126 * config/visium/visium.md (Substitutions): Likewise.
31128 2017-05-17 Martin Liska <mliska@suse.cz>
31130 * cfg.c: Introduce dump_flags_t type and
31131 use it instead of int type.
31133 * cfghooks.c: Likewise.
31134 * cfghooks.h (struct cfg_hooks): Likewise.
31135 * cfgrtl.c: Likewise.
31136 * cfgrtl.h: Likewise.
31137 * cgraph.c (cgraph_node::get_body): Likewise.
31138 * coretypes.h: Likewise.
31139 * domwalk.c: Likewise.
31140 * domwalk.h: Likewise.
31141 * dumpfile.c (struct dump_option_value_info): Likewise.
31142 (dump_enable_all): Likewise.
31143 (dump_switch_p_1): Likewise.
31144 (opt_info_switch_p): Likewise.
31145 * dumpfile.h (enum tree_dump_index): Likewise.
31146 (struct dump_file_info): Likewise.
31147 * genemit.c: Likewise.
31148 * generic-match-head.c: Likewise.
31149 * gengtype.c (open_base_files): Likewise.
31150 * gimple-pretty-print.c: Likewise.
31151 * gimple-pretty-print.h: Likewise.
31152 * graph.c (print_graph_cfg): Likewise.
31153 * graphite-scop-detection.c (dot_all_sese): Likewise.
31154 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31155 * loop-unroll.c (report_unroll): Likewise.
31156 * passes.c (pass_manager::register_one_dump_file): Likewise.
31157 * print-tree.c: Likewise.
31158 * statistics.c: Likewise.
31159 * tree-cfg.c: Likewise.
31160 * tree-cfg.h: Likewise.
31161 * tree-dfa.c: Likewise.
31162 * tree-dfa.h: Likewise.
31163 * tree-dump.c (dump_function): Likewise.
31164 * tree-dump.h (struct dump_info): Likewise.
31165 * tree-pretty-print.c: Likewise.
31166 * tree-pretty-print.h: Likewise.
31167 * tree-ssa-live.c: Likewise.
31168 * tree-ssa-live.h: Likewise.
31169 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31170 * tree-vect-loop.c: Likewise.
31171 * tree-vect-slp.c: Likewise.
31173 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31174 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31176 PR tree-optimization/80457
31177 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31178 of all arguments to a statement as scalar_to_vec operations.
31179 (vectorizable_call): Adjust call to vect_model_simple_cost for
31181 (vectorizable_conversion): Likewise.
31182 (vectorizable_assignment): Likewise.
31183 (vectorizable_shift): Likewise.
31184 (vectorizable_operation): Likewise.
31185 (vectorizable_comparison): Likewise.
31186 (vect_is_simple_cond): Record the def types for operands.
31187 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31188 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31189 for statement argument count.
31191 2017-05-16 Carl Love <cel@us.ibm.com>
31193 * config/rs6000/rs6000-c: Add support for built-in functions
31194 vector unsigned long long vec_bperm (vector unsigned long long,
31195 vector unsigned char)
31196 vector signed long long vec_mule (vector signed int,
31198 vector unsigned long long vec_mule (vector unsigned int,
31199 vector unsigned int)
31200 vector signed long long vec_mulo (vector signed int,
31202 vector unsigned long long vec_mulo (vector unsigned int,
31203 vector unsigned int)
31204 vector signed char vec_sldw (vector signed char,
31205 vector signed char,
31207 vector unsigned char vec_sldw (vector unsigned char,
31208 vector unsigned char,
31210 vector signed short vec_sldw (vector signed short,
31211 vector signed short,
31213 vector unsigned short vec_sldw (vector unsigned short,
31214 vector unsigned short,
31216 vector signed int vec_sldw (vector signed int,
31219 vector unsigned int vec_sldw (vector unsigned int,
31220 vector unsigned int,
31222 vector signed long long vec_sldw (vector signed long long,
31223 vector signed long long,
31225 vector unsigned long long vec_sldw (vector unsigned long long,
31226 vector unsigned long long,
31228 * config/rs6000/rs6000-c: Add support for built-in functions
31229 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31230 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31231 * doc/extend.texi: Update the built-in documentation for the
31232 new built-in functions.
31234 2017-05-16 Marek Polacek <polacek@redhat.com>
31238 * tree.c (save_expr): Don't fold the expression.
31240 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31242 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31243 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31244 and (?*y,m). Update insn attributes.
31246 2017-05-16 Martin Liska <mliska@suse.cz>
31248 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31249 flags argument of print_gimple_stmt, print_gimple_expr,
31250 print_generic_stmt and print_generic_expr.
31251 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31252 * coretypes.h: Likewise.
31253 * except.c (dump_eh_tree): Likewise.
31254 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31255 * gimple-pretty-print.h: Likewise.
31256 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31257 (backprop::push_to_worklist): Likewise.
31258 (backprop::pop_from_worklist): Likewise.
31259 (backprop::process_use): Likewise.
31260 (backprop::intersect_uses): Likewise.
31261 (note_replacement): Likewise.
31262 * gimple-ssa-store-merging.c
31263 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31264 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31265 (pass_store_merging::execute): Likewise.
31266 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31267 (ssa_base_cand_dump_callback): Likewise.
31268 (dump_incr_vec): Likewise.
31269 (replace_refs): Likewise.
31270 (replace_mult_candidate): Likewise.
31271 (create_add_on_incoming_edge): Likewise.
31272 (create_phi_basis): Likewise.
31273 (insert_initializers): Likewise.
31274 (all_phi_incrs_profitable): Likewise.
31275 (introduce_cast_before_cand): Likewise.
31276 (replace_one_candidate): Likewise.
31277 * gimplify.c (gimplify_expr): Likewise.
31278 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31279 (set_rename): Likewise.
31280 (rename_uses): Likewise.
31281 (copy_loop_phi_nodes): Likewise.
31282 (add_close_phis_to_merge_points): Likewise.
31283 (copy_loop_close_phi_args): Likewise.
31284 (copy_cond_phi_args): Likewise.
31285 (graphite_copy_stmts_from_block): Likewise.
31286 (translate_pending_phi_nodes): Likewise.
31287 * graphite-poly.c (print_pdr): Likewise.
31288 (dump_gbb_cases): Likewise.
31289 (dump_gbb_conditions): Likewise.
31290 (print_scop_params): Likewise.
31291 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31292 (build_cross_bb_scalars_use): Likewise.
31293 (gather_bbs::before_dom_children): Likewise.
31294 * hsa-dump.c (dump_hsa_immed): Likewise.
31295 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31296 (get_replacement_map): Likewise.
31297 * ipa-inline-analysis.c (dump_condition): Likewise.
31298 (estimate_function_body_sizes): Likewise.
31299 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31300 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31301 * ipa-prop.c (ipa_dump_param): Likewise.
31302 (ipa_print_node_jump_functions_for_edge): Likewise.
31303 (ipa_modify_call_arguments): Likewise.
31304 (ipa_modify_expr): Likewise.
31305 (ipa_dump_param_adjustments): Likewise.
31306 (ipa_dump_agg_replacement_values): Likewise.
31307 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31308 * ipa-pure-const.c (check_stmt): Likewise.
31309 (pass_nothrow::execute): Likewise.
31310 * ipa-split.c (execute_split_functions): Likewise.
31311 * omp-offload.c (dump_oacc_loop_part): Likewise.
31312 (dump_oacc_loop): Likewise.
31313 * trans-mem.c (tm_log_emit): Likewise.
31314 (tm_memopt_accumulate_memops): Likewise.
31315 (dump_tm_memopt_set): Likewise.
31316 (dump_tm_memopt_transform): Likewise.
31317 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31318 (print_loop): Likewise.
31319 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31320 (chkp_gather_checks_info): Likewise.
31321 (chkp_get_check_result): Likewise.
31322 (chkp_remove_check_if_pass): Likewise.
31323 (chkp_use_outer_bounds_if_possible): Likewise.
31324 (chkp_reduce_bounds_lifetime): Likewise.
31325 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31326 (chkp_mark_completed_bounds): Likewise.
31327 (chkp_register_incomplete_bounds): Likewise.
31328 (chkp_mark_invalid_bounds): Likewise.
31329 (chkp_maybe_copy_and_register_bounds): Likewise.
31330 (chkp_build_returned_bound): Likewise.
31331 (chkp_get_bound_for_parm): Likewise.
31332 (chkp_build_bndldx): Likewise.
31333 (chkp_get_bounds_by_definition): Likewise.
31334 (chkp_generate_extern_var_bounds): Likewise.
31335 (chkp_get_bounds_for_decl_addr): Likewise.
31336 * tree-chrec.c (chrec_apply): Likewise.
31337 * tree-data-ref.c (dump_data_reference): Likewise.
31338 (dump_subscript): Likewise.
31339 (dump_data_dependence_relation): Likewise.
31340 (analyze_overlapping_iterations): Likewise.
31341 * tree-inline.c (expand_call_inline): Likewise.
31342 (tree_function_versioning): Likewise.
31343 * tree-into-ssa.c (dump_defs_stack): Likewise.
31344 (dump_currdefs): Likewise.
31345 (dump_names_replaced_by): Likewise.
31346 (dump_update_ssa): Likewise.
31347 (update_ssa): Likewise.
31348 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31349 * tree-parloops.c (build_new_reduction): Likewise.
31350 (try_create_reduction_list): Likewise.
31351 (ref_conflicts_with_region): Likewise.
31352 (oacc_entry_exit_ok_1): Likewise.
31353 (oacc_entry_exit_single_gang): Likewise.
31354 * tree-pretty-print.h: Likewise.
31355 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31356 (get_scalar_evolution): Likewise.
31357 (add_to_evolution): Likewise.
31358 (get_loop_exit_condition): Likewise.
31359 (analyze_evolution_in_loop): Likewise.
31360 (analyze_initial_condition): Likewise.
31361 (analyze_scalar_evolution): Likewise.
31362 (instantiate_scev): Likewise.
31363 (number_of_latch_executions): Likewise.
31364 (gather_chrec_stats): Likewise.
31365 (final_value_replacement_loop): Likewise.
31366 (scev_const_prop): Likewise.
31367 * tree-sra.c (dump_access): Likewise.
31368 (disqualify_candidate): Likewise.
31369 (create_access): Likewise.
31370 (reject): Likewise.
31371 (maybe_add_sra_candidate): Likewise.
31372 (create_access_replacement): Likewise.
31373 (analyze_access_subtree): Likewise.
31374 (analyze_all_variable_accesses): Likewise.
31375 (sra_modify_assign): Likewise.
31376 (initialize_constant_pool_replacements): Likewise.
31377 (find_param_candidates): Likewise.
31378 (decide_one_param_reduction): Likewise.
31379 (replace_removed_params_ssa_names): Likewise.
31380 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31381 * tree-ssa-copy.c (dump_copy_of): Likewise.
31382 (copy_prop_visit_cond_stmt): Likewise.
31383 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31384 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31385 (record_equivalences_from_stmt): Likewise.
31386 * tree-ssa-dse.c (compute_trims): Likewise.
31387 (delete_dead_call): Likewise.
31388 (delete_dead_assignment): Likewise.
31389 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31390 (forward_propagate_into_cond): Likewise.
31391 (pass_forwprop::execute): Likewise.
31392 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31393 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31395 (move_computations_worker): Likewise.
31396 (execute_sm): Likewise.
31397 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31398 (remove_exits_and_undefined_stmts): Likewise.
31399 (remove_redundant_iv_tests): Likewise.
31400 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31401 (adjust_iv_update_pos): Likewise.
31402 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31403 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31404 (value_replacement): Likewise.
31405 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31406 * tree-ssa-pre.c (print_pre_expr): Likewise.
31407 (get_representative_for): Likewise.
31408 (create_expression_by_pieces): Likewise.
31409 (insert_into_preds_of_block): Likewise.
31410 (eliminate_insert): Likewise.
31411 (eliminate_dom_walker::before_dom_children): Likewise.
31412 (eliminate): Likewise.
31413 (remove_dead_inserted_code): Likewise.
31414 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31415 * tree-ssa-reassoc.c (get_rank): Likewise.
31416 (eliminate_duplicate_pair): Likewise.
31417 (eliminate_plus_minus_pair): Likewise.
31418 (eliminate_not_pairs): Likewise.
31419 (undistribute_ops_list): Likewise.
31420 (eliminate_redundant_comparison): Likewise.
31421 (update_range_test): Likewise.
31422 (optimize_range_tests_var_bound): Likewise.
31423 (optimize_vec_cond_expr): Likewise.
31424 (rewrite_expr_tree): Likewise.
31425 (rewrite_expr_tree_parallel): Likewise.
31426 (linearize_expr): Likewise.
31427 (break_up_subtract): Likewise.
31428 (linearize_expr_tree): Likewise.
31429 (attempt_builtin_powi): Likewise.
31430 (attempt_builtin_copysign): Likewise.
31431 (transform_stmt_to_copy): Likewise.
31432 (transform_stmt_to_multiply): Likewise.
31433 (dump_ops_vector): Likewise.
31434 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31435 (print_scc): Likewise.
31436 (set_ssa_val_to): Likewise.
31437 (visit_reference_op_store): Likewise.
31438 (visit_use): Likewise.
31439 (sccvn_dom_walker::before_dom_children): Likewise.
31440 (run_scc_vn): Likewise.
31441 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31443 (expr_hash_elt::print): Likewise.
31444 (const_and_copies::pop_to_marker): Likewise.
31445 (const_and_copies::record_const_or_copy_raw): Likewise.
31446 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31447 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31448 (dump_predicates): Likewise.
31449 (find_uninit_use): Likewise.
31450 (warn_uninitialized_phi): Likewise.
31451 (pass_late_warn_uninitialized::execute): Likewise.
31452 * tree-ssa.c (verify_vssa): Likewise.
31453 (verify_ssa): Likewise.
31454 (maybe_optimize_var): Likewise.
31455 * tree-vrp.c (dump_value_range): Likewise.
31456 (dump_all_value_ranges): Likewise.
31457 (dump_asserts_for): Likewise.
31458 (register_edge_assert_for_2): Likewise.
31459 (vrp_visit_cond_stmt): Likewise.
31460 (vrp_visit_switch_stmt): Likewise.
31461 (vrp_visit_stmt): Likewise.
31462 (vrp_visit_phi_node): Likewise.
31463 (simplify_cond_using_ranges_1): Likewise.
31464 (fold_predicate_in): Likewise.
31465 (evrp_dom_walker::before_dom_children): Likewise.
31466 (evrp_dom_walker::push_value_range): Likewise.
31467 (evrp_dom_walker::pop_value_range): Likewise.
31468 (execute_early_vrp): Likewise.
31470 2017-05-16 Richard Biener <rguenther@suse.de>
31472 * dwarf2out.c (loc_list_from_tree_1): Do not create
31473 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31475 2017-05-16 Richard Biener <rguenther@suse.de>
31477 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31479 (note_variable_value_in_expr): If we resolved the decl ref
31480 do not push to the stack.
31482 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31484 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31485 operations in fast-math mode.
31486 (vaddq_f16): Likewise.
31487 (vmul_f16): Likewise.
31488 (vmulq_f16): Likewise.
31489 (vsub_f16): Likewise.
31490 (vsubq_f16): Likewise.
31491 * config/arm/neon.md (add<mode>3): New.
31493 (fma:<VH:mode>3): New. Also remove outdated comment.
31496 2017-05-16 Martin Liska <mliska@suse.cz>
31500 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31501 (odr_types_equivalent_p): Likewise.
31503 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31505 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31507 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31510 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31511 non-interunit SSE move alternatives with '?'.
31512 (zero-extendsidi peephole2): New peephole to skip intermediate
31513 general register in SSE zero-extend sequence.
31515 2017-05-15 Jeff Law <law@redhat.com>
31517 * reorg.c (relax_delay_slots): Create a new variable to hold
31518 the temporary target rather than clobbering TARGET_LABEL.
31520 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31521 missing argument to extract_bit_field call.
31522 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31524 2017-05-15 Martin Liska <mliska@suse.cz>
31527 * gcc.c (process_command): Do not allow empty argument of -o option.
31529 2017-05-15 Renlin Li <renlin.li@arm.com>
31531 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31532 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31533 * config/aarch64/constraints.md (Usf): Add long call check.
31534 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31535 (call_value): Likewise.
31536 (sibcall): Likewise.
31537 (sibcall_value): Likewise.
31539 (call_value_insn): New.
31540 (sibcall_insn): Update rtx pattern.
31541 (sibcall_value_insn): Likewise.
31542 (call_internal): Remove.
31543 (call_value_internal): Likewise.
31544 (sibcall_internal): Likewise.
31545 (sibcall_value_internal): Likewise.
31546 (call_reg): Likewise.
31547 (call_symbol): Likewise.
31548 (call_value_reg): Likewise.
31549 (call_value_symbol): Likewise.
31551 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31554 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31556 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31558 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31559 compatible with CCGOCmode and with CCZmode.
31561 2017-05-14 Martin Sebor <msebor@redhat.com>
31563 PR middle-end/77671
31564 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31565 (gimple_fold_builtin_snprintf): Same.
31566 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31567 (gimple_fold_builtin_snprintf): Same.
31568 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31569 of character types.
31570 (is_call_safe): New function.
31571 (try_substitute_return_value): Call it.
31572 (try_simplify_call): New function.
31573 (pass_sprintf_length::handle_gimple_call): Call it.
31575 2017-05-14 Martin Sebor <msebor@redhat.com>
31577 PR middle-end/80669
31578 * builtins.c (expand_builtin_stpncpy): Simplify.
31580 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31582 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31583 * config/i386/i386.h
31584 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31585 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31586 (struct machine_function): Add new members call_ms2sysv,
31587 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31588 (struct machine_frame_state): New fields sp_realigned and
31589 sp_realigned_offset.
31590 * config/i386/i386.c
31591 (enum xlogue_stub): New enum.
31592 (enum xlogue_stub_sets): New enum.
31593 (class xlogue_layout): New class.
31594 (struct ix86_frame): New fields stack_realign_allocate_offset,
31595 stack_realign_offset and outlined_save_offset. Modify comments to
31596 detail stack layout when using out-of-line stubs.
31597 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31598 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31599 -mcall-ms2sysv-xlogues.
31600 (stub_managed_regs): New static variable.
31601 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31602 registers managed by out-of-line stub.
31603 (disable_call_ms2sysv_xlogues): New function.
31604 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31605 m->call_ms2sysv when appropriate and compute frame layout for
31607 (sp_valid_at, fp_valid_at): New inline functions.
31608 (choose_basereg): New function.
31609 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31611 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31612 Use align parameter of choose_baseaddr to generated aligned SSE movs
31614 (pro_epilogue_adjust_stack): Modify to track
31615 machine_frame_state::sp_realigned.
31616 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31617 (ix86_nsaved_sseregs): Likewise.
31618 (ix86_emit_save_regs): Likewise.
31619 (ix86_emit_save_regs_using_mov): Likewise.
31620 (ix86_emit_save_sse_regs_using_mov): Likewise.
31621 (get_scratch_register_on_entry): Likewise.
31622 (gen_frame_set): New function.
31623 (gen_frame_load): Likewise.
31624 (gen_frame_store): Likewise.
31625 (emit_outlined_ms2sysv_save): Likewise.
31626 (emit_outlined_ms2sysv_restore): Likewise.
31627 (ix86_expand_prologue): Modify stack re-alignment code and call
31628 emit_outlined_ms2sysv_save when appropriate.
31629 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31630 parameter rtx_insn *insn, which allows the function to be used to only
31631 generate the notes.
31632 (ix86_expand_epilogue): Modify validity checks of frame and stack
31633 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31634 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31635 * config/i386/predicates.md
31636 (save_multiple): New predicate.
31637 (restore_multiple): Likewise.
31638 * config/i386/sse.md
31639 (save_multiple<mode>): New pattern.
31640 (save_multiple_realign<mode>): Likewise.
31641 (restore_multiple<mode>): Likewise.
31642 (restore_multiple_and_return<mode>): Likewise.
31643 (restore_multiple_leave_return<mode>): Likewise.
31644 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31646 2017-05-14 Julia Koval <julia.koval@intel.com>
31648 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31649 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31650 (__builtin_ia32_xsetbv): New builtins.
31651 * config/i386/i386.c (ix86_expand_special_args_builtin):
31653 (ix86_expand_builtin): Special expand for new intrinsics.
31654 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31655 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31656 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31658 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31660 * cfganal.c (inverted_post_order_compute): Change argument type
31662 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31663 * df-core.c (rest_of_handle_df_initialize): Adjust.
31664 (rest_of_handle_df_finish): Likewise.
31665 (df_analyze_1): Likewise.
31666 (df_analyze): Likewise.
31667 (loop_inverted_post_order_compute): Change argument to be a vec *.
31668 (df_analyze_loop): Adjust.
31669 (df_get_n_blocks): Likewise.
31670 (df_get_postorder): Likewise.
31671 * df.h (struct df_d): Change field to be a vec.
31672 * lcm.c (compute_laterin): Adjust.
31673 (compute_available): Likewise.
31674 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31675 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31676 * tree-ssa-pre.c (compute_antic): Likewise.
31678 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31680 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31681 (depth_first_search::depth_first_search): Change structure init
31682 function to this constructor.
31683 (depth_first_search::add_bb): Rename function to this member.
31684 (depth_first_search::execute): Likewise.
31685 (flow_dfs_compute_reverse_finish): Adjust.
31687 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31689 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31690 (longest_simple_path): Likewise.
31691 * shrink-wrap.c (spread_components): Likewise.
31692 (disqualify_problematic_components): Likewise.
31693 (emit_common_heads_for_components): Likewise.
31694 (emit_common_tails_for_components): Likewise.
31695 (insert_prologue_epilogue_for_components): Likewise.
31697 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31699 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31702 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31704 * df-core.c (df_set_blocks): Start using auto_bitmap.
31705 (df_compact_blocks): Likewise.
31706 * df-problems.c (df_rd_confluence_n): Likewise.
31707 * df-scan.c (df_insn_rescan_all): Likewise.
31708 (df_process_deferred_rescans): Likewise.
31709 (df_update_entry_block_defs): Likewise.
31710 (df_update_exit_block_uses): Likewise.
31711 (df_entry_block_bitmap_verify): Likewise.
31712 (df_exit_block_bitmap_verify): Likewise.
31713 (df_scan_verify): Likewise.
31714 * lra-constraints.c (lra_constraints): Likewise.
31715 (undo_optional_reloads): Likewise.
31716 (lra_undo_inheritance): Likewise.
31717 * lra-remat.c (calculate_gen_cands): Likewise.
31718 (do_remat): Likewise.
31719 * lra-spills.c (assign_spill_hard_regs): Likewise.
31720 (spill_pseudos): Likewise.
31721 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31722 (bitmap_set_subtract_values): Likewise.
31724 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31726 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31727 management with auto_bitmap.
31728 (fix_inter_tick): Likewise.
31729 (fix_recovery_deps): Likewise.
31730 * ira.c (add_store_equivs): Likewise.
31731 (find_moveable_pseudos): Likewise.
31732 (split_live_ranges_for_shrink_wrap): Likewise.
31733 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31734 (rtx_reuse_manager::seen_def_p): Likewise.
31735 (rtx_reuse_manager::set_seen_def): Likewise.
31736 * print-rtl.h (class rtx_reuse_manager): Likewise.
31738 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31740 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31742 (migrate_btr_def): Likewise.
31743 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31744 * df-core.c (loop_post_order_compute): Likewise.
31745 (loop_inverted_post_order_compute): Likewise.
31746 * hsa-common.h: Likewise.
31747 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31748 * init-regs.c (initialize_uninitialized_regs): Likewise.
31749 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31750 (inline_small_functions): Likewise.
31751 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31752 * ira.c (combine_and_move_insns): Likewise.
31753 (build_insn_chain): Likewise.
31754 * loop-invariant.c (find_invariants): Likewise.
31755 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31756 * predict.c (tree_predict_by_opcode): Likewise.
31757 (predict_paths_leading_to): Likewise.
31758 (predict_paths_leading_to_edge): Likewise.
31759 (estimate_loops_at_level): Likewise.
31760 (estimate_loops): Likewise.
31761 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31762 (spread_components): Likewise.
31763 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31764 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31765 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31766 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31767 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31768 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31769 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31770 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31771 (mark_threaded_blocks): Likewise.
31772 (thread_through_all_blocks): Likewise.
31773 * tree-ssa.c (verify_ssa): Likewise.
31774 (execute_update_addresses_taken): Likewise.
31775 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31777 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31779 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31781 (post_order_compute): Likewise.
31782 (inverted_post_order_compute): Likewise.
31783 (pre_and_rev_post_order_compute_fn): Likewise.
31785 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31787 * genrecog.c (int_set::int_set): Explicitly construct our
31788 auto_vec base class.
31789 * vec.h (auto_vec::auto_vec): New constructor.
31791 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31793 * bitmap.h (class auto_bitmap): New constructor taking
31794 bitmap_obstack * argument.
31796 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31798 * bitmap.h (class auto_bitmap): Change type of m_bits to
31799 bitmap_head, and adjust ctor / dtor and member operators.
31801 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31803 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31804 when returned register mode doesn't match original mode.
31806 2017-05-12 Jeff Law <law@redhat.com>
31807 Jakub Jelinek <jakub@redhat.com>
31809 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31810 we look for cc setter after the compare-elim changes.
31811 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31812 within the vector to match what compare-elim now expects.
31813 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31814 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31816 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31817 after the compare-elim changes.
31818 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31819 the vector to match what compare-elim now expects.
31820 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31821 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31822 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31823 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31824 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31826 * config/visium/visium.c (single_set_and_flags): Fix where
31827 we look for cc setter after the compare-elim changes.
31828 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31829 with the vector to match what compare-elim now expects.
31830 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31831 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31832 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31833 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31834 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31835 (neg<mode>2_insn_set_overflow): Likewise.
31837 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31839 PR middle-end/79794
31840 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31841 maybe_expand_insn call, set ops[0].target. If still set after call,
31842 set alt_rtl. Add extra arg to recursive calls.
31843 (extract_bit_field): Add alt_rtl argument. Pass to
31845 * expmed.h (extract_bit_field): Fix prototype.
31846 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31847 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31848 to extract_bit_field_calls.
31849 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31850 Pass alt_rtl to extract_bit_field calls.
31851 * calls.c (store_unaligned_arguments_into_psuedos)
31852 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31853 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31855 * optabs.h (struct expand_operand): Add target bitfield.
31857 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31859 * compare-elim.c (try_eliminate_compare): Canonicalize
31860 operation with embedded compare to
31861 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31862 (set (reg) (operation)].
31864 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31866 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31869 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31870 cost of adding a carry flag for ADC instruction.
31871 [case MINUS]: Ignore the cost of subtracting a carry flag
31872 for SBB instruction.
31874 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31876 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31878 * config/rs6000/bmiintrin.h: New file.
31879 * config/rs6000/bmi2intrin.h: New file.
31880 * config/rs6000/x86intrin.h: New file.
31882 2017-05-12 Jeff Law <law@redhat.com>
31884 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31887 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31889 PR middle-end/80707
31890 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31892 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31894 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31895 early expansion of vector divide builtins.
31896 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31897 builtins identified as having unsigned arguments.
31899 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31901 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31902 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31903 expansion of vector logical operations (and, andc, or, xor,
31906 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31908 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31909 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31911 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31913 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31914 early GIMPLE expansion of vector multiplies.
31916 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31918 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31919 TARGET_HAVE_MOVT conditional.
31920 (movt splitter): Likewise.
31922 2017-05-12 Richard Biener <rguenther@suse.de>
31924 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31925 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31926 Fold all stmts not inplace.
31928 2017-05-12 Richard Biener <rguenther@suse.de>
31930 PR tree-optimization/80713
31931 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31932 inserted_exprs bit for not removed stmts.
31934 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31936 PR middle-end/69921
31937 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31938 parallelized" attribute for parallelized OpenACC kernels.
31939 * omp-offload.c (execute_oacc_device_lower): Use it.
31941 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31942 Set "oacc kernels" attribute.
31943 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31944 parameter. Adjust all users.
31945 (oacc_fn_attrib_kernels_p): Remove function.
31946 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31947 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31948 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31949 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31950 assert "oacc kernels" attribute is set.
31952 2017-05-11 Carl Love <cel@us.ibm.com>
31954 * config/rs6000/rs6000-c: Add support for built-in functions
31955 vector unsigned char vec_popcnt (vector signed char)
31956 vector unsigned char vec_popcnt (vector unsigned char)
31957 vector unsigned short vec_popcnt (vector signed short)
31958 vector unsigned short vec_popcnt (vector unsigned short)
31959 vector unsigned int vec_popcnt (vector signed int)
31960 vector unsigned int vec_popcnt (vector unsigned int)
31961 vector unsigned long long vec_popcnt (vector signed long long)
31962 vector unsigned long long vec_popcnt (vector unsigned long long)
31963 vector signed long long vec_slo (vector signed long long,
31964 vector signed char)
31965 vector signed long long vec_slo (vector signed long long,
31966 vector unsigned char)
31967 vector unsigned long long vec_slo (vector unsigned long long,
31968 vector signed char)
31969 vector unsigned long long vec_slo (vector unsigned long long,
31970 vector unsigned char)
31971 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31972 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31973 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31974 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31975 * doc/extend.texi: Update the built-in documentation file for the
31976 new built-in functions.
31978 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31980 * attribs.h (sorted_attr_string): Move machine independent
31981 functions for target clone support from the i386 port to common
31982 code. Rename ix86_function_versions to common_function_versions.
31983 Rename make_name to make_unique_name.
31984 (common_function_versions): Likewise.
31985 (make_unique_name): Likewise.
31986 (make_dispatcher_decl): Likewise.
31987 (is_function_default_version): Likewise.
31988 * attribs.c (attr_strcmp): Likewise.
31989 (sorted_attr_string): Likewise.
31990 (common_function_versions): Likewise.
31991 (make_unique_name): Likewise.
31992 (make_dispatcher_decl): Likewise.
31993 (is_function_default_version): Likewise.
31994 * config/i386/i386.c (attr_strcmp): Likewise.
31995 (sorted_attr_string): Likewise.
31996 (ix86_function_versions): Likewise.
31997 (make_name): Likewise.
31998 (make_dispatcher_decl): Likewise.
31999 (is_function_default_version): Likewise.
32000 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
32002 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32005 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
32006 Account for direct move costs for vec_construct of integer
32009 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
32012 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
32013 (UNSPEC_STX_ATOMIC): Ditto.
32014 (loaddi_via_sse): New insn.
32015 (storedi_via_sse): Ditto.
32016 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
32017 Update corresponding peephole2 patterns.
32018 (atomic_storedi_fpu): Ditto.
32020 2017-05-11 Julia Koval <julia.koval@intel.com>
32022 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
32023 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
32025 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
32026 (__builtin_ia32_rsqrt14ss_mask): New builtins.
32027 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
32029 2017-05-11 Nathan Sidwell <nathan@acm.org>
32031 * graphite-poly.c: Include dumpfile.h.
32033 * dumpfle.h (dump_function): Declare here ...
32034 * tree-dump.h (dump_function): ... not here.
32035 * dumpfile.c: #include tree-cfg.h.
32036 (dump_function): Move here from ...
32037 * tree-dump.c (dump_function): ... here.
32038 * gimplify.c: #include splay-tree.h, not tree-dump.h.
32039 * graphite-poly.c: Don't include tree-dump.h.
32040 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
32041 * print-tree.c: Likewise.
32042 * stor-layout.c: Likewise.
32043 * tree-nested.c: Likewise.
32045 * dumpfile.c (dump_start): Use TDF_FLAGS.
32046 (dump_enable_all): Fix TDF_KIND check thinko.
32048 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
32050 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32051 array entries to represent two legal parameterizations of the
32052 overloaded __builtin_cmpb function, as represented by the
32053 P6_OV_BUILTIN_CMPB constant.
32054 (altivec_resolve_overloaded_builtin): Add special case handling
32055 for the __builtin_cmpb function, as represented by the
32056 P6_OV_BUILTIN_CMPB constant.
32057 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
32058 (BU_P6_64BIT_2): New macro.
32059 (BU_P6_OVERLOAD_2): New macro
32060 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32061 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32062 (CMPB): Add overload support to represent both 32-bit and 64-bit
32063 compare-bytes function.
32064 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32065 support for TARGET_CMPB.
32066 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32067 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32068 documentation of the __builtin_cmpb overloaded built-in function.
32070 2017-05-11 Richard Biener <rguenther@suse.de>
32072 PR tree-optimization/80705
32073 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32074 bases are not vectorizable.
32076 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32078 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32079 when counting register pressure.
32081 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32083 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32084 (iv_ca_more_deps): Renamed to ...
32085 (iv_ca_compare_deps): ... this.
32086 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32088 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32090 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32092 (determine_group_iv_costs): ... here.
32093 (find_inv_vars_cb): Record inv var if it's not recorded before.
32095 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32097 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32098 (get_shiftadd_cost): Ditto.
32100 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32102 * tree-ssa-address.c: Include header file.
32103 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32105 (add_to_parts): Refactor.
32106 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32107 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32110 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32112 PR tree-optimization/53090
32113 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32115 (extract_cond_operands): Detect condition with IV on both sides
32116 and return COMP_IV_EXPR_2.
32117 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32118 (rewrite_use_compare): Simplify by removing call to function
32119 extract_cond_operands.
32121 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32123 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32124 (extract_cond_operands): Detect condition comparing against non-
32125 invariant bound and return appropriate enum value.
32126 (find_interesting_uses_cond): Update use of extract_cond_operands.
32127 Handle its return value accordingly.
32128 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32130 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32132 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32133 nonlinear iv_use computation in loop invariant sensitive way.
32135 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32137 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32138 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32140 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32142 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32143 (dump_cand): Support iv_cand.inv_exprs.
32144 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32146 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32149 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32151 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32153 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32154 as local function. Include necessary header files.
32155 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32157 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32159 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32161 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32163 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32164 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32165 RSHIFT_EXPR and BIT_NOT_EXPR.
32167 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32169 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32170 (adjust_setup_cost): New parameter supporting round up adjustment.
32171 (struct address_cost_data): Delete.
32172 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32173 (split_address_cost, ptr_difference_cost): Delete.
32174 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32175 (struct ainc_cost_data): New struct.
32176 (get_address_cost_ainc): New function.
32177 (get_address_cost, get_computation_cost): Reimplement.
32178 (determine_group_iv_cost_address): Record inv_expr for all uses of
32180 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32181 (iv_ca_has_deps): Reimplemented to ...
32182 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32184 (iv_ca_extend): Call iv_ca_more_deps.
32186 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32188 * tree-ssa-address.c (struct mem_address): Move to header file.
32189 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32190 * tree-ssa-address.h (struct mem_address): Move from C file.
32191 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32193 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32195 * tree-affine.h (aff_combination_type): New interface.
32196 (aff_combination_zero_p): Remove static.
32197 (aff_combination_const_p): New interface.
32198 (aff_combination_singleton_var_p): New interfaces.
32200 2017-05-11 Richard Biener <rguenther@suse.de>
32202 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32203 Skip unreachable blocks and destinations.
32204 (eliminate): Move stmt removal and fixup ...
32205 (fini_eliminate): ... here. Skip inserted exprs.
32206 (pass_pre::execute): Move fini_pre after fini_eliminate.
32207 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32208 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32209 PRE to get rid of dead code that has invalid SSA form and
32210 split critical edges again.
32212 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32214 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32216 2017-05-11 Richard Biener <rguenther@suse.de>
32218 * passes.c (execute_function_todo): Verify loops if they are
32219 said to be up-to-date.
32220 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32221 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32223 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32226 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32227 handle calling assemble_external ourself.
32230 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32231 modes with zero size. Enhance comment.
32233 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32235 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32236 built-ins for vec_xl and vec_xst with short and char pointer
32239 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32241 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32242 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32243 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32244 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32245 (_mm_maskz_min_round_ss): New intrinsics.
32246 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32247 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32248 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32249 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32250 (__builtin_ia32_minss_mask_round): New builtins.
32251 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32252 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32253 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32255 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32256 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32258 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32261 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32263 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32264 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32265 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32266 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32267 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32268 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32269 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32270 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32271 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32272 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32273 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32274 (__builtin_ia32_mulss_mask_round): New builtins.
32275 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32276 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32277 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32279 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32280 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32282 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32285 2017-05-10 Julia Koval <julia.koval@intel.com>
32287 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32288 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32289 (_mm256_setr_m128i): New intrinsics.
32291 2017-05-10 Julia Koval <julia.koval@intel.com>
32293 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32294 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32295 (_mm_maskz_rcp14_ss): New intrinsics.
32296 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32297 (__builtin_ia32_rcp14ss_mask): New builtins.
32298 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32300 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32302 PR tree-optimization/51513
32303 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32304 (assert_unreachable_fallthru_edge_p): Use it.
32305 (group_case_labels_stmt): Likewise.
32306 * tree-cfg.h: Prototype it.
32307 * stmt.c: Include cfghooks.h and tree-cfg.h.
32308 (emit_case_dispatch_table) <gap_label>: New local variable.
32309 Use it to fill dispatch table gaps.
32310 Test for default_label before updating probabilities.
32311 (expand_case) <default_label>: Remove unneeded initialization.
32312 Test for unreachable default case statement and remove its edge.
32313 Set default_label accordingly.
32314 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32316 2017-05-10 Carl Love <cel@us.ibm.com>
32318 * config/rs6000/rs6000-c: Add support for built-in functions
32319 vector signed char vec_neg (vector signed char)
32320 vector signed short int vec_neg (vector short int)
32321 vector signed int vec_neg (vector signed int)
32322 vector signed long long vec_neg (vector signed long long)
32323 vector float vec_neg (vector float)
32324 vector double vec_neg (vector double)
32325 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32327 * config/rs6000/altivec.h: Add define for vec_neg
32328 * doc/extend.texi: Update the built-in documentation for the
32329 new built-in functions.
32331 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32333 PR tree-optimization/77644
32334 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32336 2017-05-10 Nathan Sidwell <nathan@acm.org>
32338 * dumpfile.h (TDI_lang_all): New.
32339 (TDF_KIND): New. Renumber others
32340 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32342 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32344 (get_dump_file_name): Adjust suffix generation.
32345 (dump_enable_all): Use TDF_KIND.
32346 * doc/invoke.texi (-fdump-lang-all): Document.
32348 * dumpfile.h: Tabify.
32350 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32353 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32354 Move member access before delete.
32356 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32358 * tree-inline.c (expand_call_inline): Split block at stmt
32361 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32364 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32365 are now unused after splitting mov{sf,sd}_hardfloat.
32366 (f32_lr2): Likewise.
32367 (f32_lm): Likewise.
32368 (f32_lm2): Likewise.
32369 (f32_li): Likewise.
32370 (f32_li2): Likewise.
32371 (f32_lv): Likewise.
32372 (f32_sr): Likewise.
32373 (f32_sr2): Likewise.
32374 (f32_sm): Likewise.
32375 (f32_sm2): Likewise.
32376 (f32_si): Likewise.
32377 (f32_si2): Likewise.
32378 (f32_sv): Likewise.
32379 (f32_dm): Likewise.
32380 (f32_vsx): Likewise.
32381 (f32_av): Likewise.
32382 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32383 For movsf, order stores so the VSX stores occur before the GPR
32384 store which encourages the register allocator to use a traditional
32385 FPR instead of a GPR. For movsd, order the stores so that the GPR
32386 store comes before the VSX stores to allow the power6 to work.
32387 This is due to the power6 not having a 32-bit integer store
32388 instruction from a FPR.
32389 (movsf_hardfloat): Likewise.
32390 (movsd_hardfloat): Likewise.
32392 2017-05-09 Martin Sebor <msebor@redhat.com>
32394 PR translation/80280
32395 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32398 PR translation/80280
32399 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32400 data member added in r247778.
32401 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32403 2017-05-09 Nathan Sidwell <nathan@acm.org>
32405 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32407 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32410 2017-05-09 Marek Polacek <polacek@redhat.com>
32412 * doc/invoke.texi: Fix typo.
32414 2017-05-09 Richard Biener <rguenther@suse.de>
32416 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32417 (vrp_val_is_min): Likewise.
32418 (set_value_range_to_value): Likewise.
32419 (set_value_range_to_nonnegative): Likewise.
32420 (gimple_assign_nonzero_p): Likewise.
32421 (gimple_stmt_nonzero_p): Likewise.
32422 (vrp_int_const_binop): Likewise. Remove unreachable case.
32423 (adjust_range_with_scev): Adjust comments.
32424 (compare_range_with_value): Likewise.
32425 (extract_range_from_phi_node): Likewise.
32426 (test_for_singularity): Likewise.
32428 2017-05-09 Richard Biener <rguenther@suse.de>
32430 * tree-vrp.c (get_single_symbol): Add assert that we don't
32431 get overflowed constants as invariant part.
32432 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32433 checks. Use wi::cmp instead of recursing for integer constants.
32434 (compare_values): Just ignore whether we assumed undefined
32435 overflow instead of failing the compare.
32436 (extract_range_for_var_from_comparison_expr): Add comment before the
32437 TREE_NO_WARNING sets.
32438 (test_for_singularity): Likewise.
32439 (extract_range_from_comparison): Do not disable optimization
32440 when we assumed undefined overflow.
32441 (extract_range_basic): Remove init of unused var.
32443 2017-05-09 Richard Biener <rguenther@suse.de>
32445 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32446 (extract_range_from_multiplicative_op_1): Adjust.
32447 (extract_range_from_binary_expr_1): Use int_const_binop.
32449 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32452 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32453 rs6000_store_data_bypass_p in seven define_bypass directives and
32454 in several comments.
32455 * config/rs6000/rs6000-protos.h: Add prototype for
32456 rs6000_store_data_bypass_p function.
32457 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32458 function implements slightly different (rs6000-specific) semantics
32459 than store_data_bypass_p, returning false rather than aborting
32460 with assertion error when arguments do not satisfy the
32461 requirements of store data bypass.
32462 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32463 rs6000_store_data_bypass_p.
32465 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32467 * config/xtensa/xtensa-protos.h
32468 (xtensa_initial_elimination_offset): New declaration.
32469 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32470 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32471 macro definition, add case for FRAME_POINTER_REGNUM when
32472 FRAME_GROWS_DOWNWARD.
32473 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32474 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32475 xtensa_initial_elimination_offset.
32477 2017-05-08 Nathan Sidwell <nathan@acm.org>
32479 * doc/invoke.texi: Alphabetize -fdump options.
32481 2017-05-08 Martin Sebor <msebor@redhat.com>
32483 PR translation/80280
32484 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32486 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32488 * target.def (compute_frame_layout): New optional target hook.
32489 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32490 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32491 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32493 * reload1.c (verify_initial_elim_offsets): Likewise.
32494 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32495 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32496 (arm_get_frame_offsets): Split up into this ...
32497 (arm_compute_frame_layout): ... and this function.
32499 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32501 * config/aarch64/constraints.md (Usa): New constraint.
32502 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32504 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32506 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32507 with_multilib_list after it has been checked.
32509 2017-05-08 Richard Biener <rguenther@suse.de>
32511 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32512 (bitmap_set_subtract_values): Likewise.
32514 2017-05-08 Richard Biener <rguenther@suse.de>
32516 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32517 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32519 (gimple_stmt_nonzero_warnv_p): Rename to ...
32520 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32522 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32523 (extract_range_basic): Adjust, do not disable propagation on
32524 strict overflow sensitive simplification.
32525 (vrp_visit_cond_stmt): Likewise.
32527 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32529 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32530 body size unconditionally.
32532 2017-05-07 Jeff Law <law@redhat.com>
32535 2017-05-06 Jeff Law <law@redhat.com>
32536 PR tree-optimization/78496
32537 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32540 PR tree-optimization/78496
32541 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32542 (simplify_stmt_using_ranges): Call it.
32543 (vrp_dom_walker::before_dom_children): Extract equivalences
32544 from an ASSERT_EXPR with an equality comparison against a
32547 2017-05-06 Jeff Law <law@redhat.com>
32549 PR tree-optimization/78496
32550 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32553 PR tree-optimization/78496
32554 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32555 (simplify_stmt_using_ranges): Call it.
32556 (vrp_dom_walker::before_dom_children): Extract equivalences
32557 from an ASSERT_EXPR with an equality comparison against a
32560 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32562 * lra-constraints.c (lra_copy_reg_equiv): New function.
32563 (split_reg): Use it to copy equivalence information from the
32564 original register to the spill register.
32566 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32568 PR rtl-optimization/75964
32569 * simplify-rtx.c (simplify_const_relational_operation): Remove
32570 invalid handling of comparisons of integer ABS.
32572 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32574 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32575 initialize to zero.
32576 (init_regs): Remove declaration.
32577 (function_arg_advance_32): Initialize error_p as boolean variable.
32579 2017-05-05 Nathan Sidwell <nathan@acm.org>
32581 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32582 lines. Use for (;;).
32584 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32586 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32587 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32588 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32589 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32590 VF=2 that require versioning.
32592 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32594 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32597 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32599 * diagnostic.h (diagnostic_override_option_index): Convert from
32600 macro to inline function.
32602 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32604 * diagnostic.c (last_module_changed_p): New function.
32605 (set_last_module): New function.
32606 (diagnostic_report_current_module): Convert macro usage to
32607 the above functions.
32608 * diagnostic.h (diagnostic_context::last_module): Strengthen
32609 from const line_map * to const line_map_ordinary *.
32610 (diagnostic_last_module_changed): Delete macro.
32611 (diagnostic_set_last_module): Delete macro.
32613 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32615 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32616 with diagnostic_report_diagnostic.
32617 (diagnostic_n_impl_richloc): Likewise.
32618 * diagnostic.h (report_diagnostic): Delete macro.
32619 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32620 with diagnostic_report_diagnostic.
32621 * substring-locations.c (format_warning_va): Likewise.
32623 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32625 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32626 save/restor of format_spec. Move option-printing code to...
32627 (print_option_information): ...this new function, and
32628 reimplement by simply printing to the pretty_printer,
32629 rather than appending to the format string.
32631 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32633 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32634 handling logic into...
32635 (update_effective_level_from_pragmas): ...this new function.
32637 2017-05-04 Andrew Waterman <andrew@sifive.com>
32639 * config/riscv/riscv.opt (mstrict-align): New option.
32640 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32641 (SLOW_UNALIGNED_ACCESS): Define.
32642 (riscv_slow_unaligned_access): Declare.
32643 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32645 (riscv_slow_unaligned_access): New variable.
32646 (rocket_tune_info): Set slow_unaligned_access to true.
32647 (optimize_size_tune_info): Set slow_unaligned_access to false.
32648 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32649 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32650 (riscv_option_override): Set riscv_slow_unaligned_access.
32651 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32653 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32655 * config/riscv/riscv.md: Unify indentation.
32657 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32662 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32664 (extendsi<mode>2): Add support for doing sign extension via
32665 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32666 don't have ISA 3.0 instructions.
32667 (extendsi<mode>2 splitter): Likewise.
32668 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32669 generate the normal insns since SImode can now go in vector
32670 registers. Disallow the special UNSPECs needed for previous
32671 machines to hide SImode being used. Add new insns
32672 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32673 (fix_trunc<mode>si2_stfiwx): Likewise.
32674 (fix_trunc<mode>si2_internal): Likewise.
32675 (fixuns_trunc<mode>si2): Likewise.
32676 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32677 (fctiw<u>z_<mode>_smallint): Likewise.
32678 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32679 of floating point to 32-bit integer from doing a direct move to
32680 the GPR registers to do a store.
32681 (fctiwz_<mode>): Break long line.
32683 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32685 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32686 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32687 (addr_list, addr_offset_valid_p): New.
32688 (split_address_groups): Check offset validity with above function.
32689 (gt-tree-ssa-loop-ivopts.h): Include header file.
32691 2017-05-05 Nathan Sidwell <nathan@acm.org>
32693 * config.gcc (arm*-*-*): Add missing 'fi'.
32695 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32697 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32698 included in -fopt-info does not matter.
32699 * doc/optinfo.texi (-fopt-info): Fix description of default
32700 behavour. Explicitly say order of options included in -fopt-info
32703 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32705 * config.gcc: Allow combinations of aprofile and rmprofile values for
32706 --with-multilib-list.
32707 * config/arm/t-multilib: New file.
32708 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32709 variables. Remove setting of ISA and floating-point ABI in
32710 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32711 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32712 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32713 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32715 * config/arm/t-rmprofile: Likewise except for the matches changes.
32716 * doc/install.texi (--with-multilib-list): Document the combination of
32717 aprofile and rmprofile values and warn about pitfalls in doing that.
32719 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32721 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32722 (movdi_aarch64): Likewise.
32724 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32726 PR tree-optimization/80632
32727 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32729 (build_arrays): Initialize it for virtual phis.
32730 (fix_phi_nodes): Use it for virtual phis.
32732 PR tree-optimization/80558
32733 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32734 [x, y] op z into [x op, y op z] for op & or | if conditions
32737 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32738 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32741 * config/arm/arm.md (use_literal_pool): Remove.
32742 (64-bit immediate split): No longer takes cost into consideration
32743 if arm_disable_literal_pool is enabled.
32744 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32745 used when arm_disable_literal_pool is enabled.
32746 (arm_max_const_double_inline_cost): Remove use of
32747 arm_disable_literal_pool.
32748 (push_minipool_fix): Add assert.
32749 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32750 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32751 (no_literal_pool_sf_immediate): New.
32753 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32755 PR tree-optimization/80613
32756 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32757 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32759 2017-05-05 Richard Biener <rguenther@suse.de>
32761 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32763 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32765 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32766 of this flag from insn conditions due to removal from r247495.
32768 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32770 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32772 (arm_early_store_addr_dep_ptr): Likewise.
32773 * config/arm/aarch-common-protos.h
32774 (arm_early_load_addr_dep_ptr): Add prototype.
32775 (arm_early_store_addr_dep_ptr): Likewise.
32776 * config/arm/cortex-a53.md: Add new bypasses.
32778 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32780 * tree.c (next_type_uid): Change type to unsigned.
32781 (type_hash_canon): Decrement back next_type_uid if
32782 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32783 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32786 2017-05-04 Martin Sebor <msebor@redhat.com>
32788 * builtins.c: Fix a trivial typo in a comment.
32790 PR middle-end/79234
32791 * builtins.c (check_sizes): Adjust to handle reading past the end.
32792 Avoid printing excessive upper bound of ranges. Use %E to print
32793 tree nodes instead of converting them to %wu.
32794 (expand_builtin_memchr): New function.
32795 (compute_dest_size): Rename...
32796 (compute_objsize): ...to this.
32797 (expand_builtin_memcpy): Adjust.
32798 (expand_builtin_mempcpy): Adjust.
32799 (expand_builtin_strcat): Adjust.
32800 (expand_builtin_strcpy): Adjust.
32801 (check_strncat_sizes): Adjust.
32802 (expand_builtin_strncat): Adjust.
32803 (expand_builtin_strncpy): Adjust and simplify.
32804 (expand_builtin_memset): Adjust.
32805 (expand_builtin_bzero): Adjust.
32806 (expand_builtin_memcmp): Adjust.
32807 (expand_builtin): Handle memcmp.
32808 (maybe_emit_chk_warning): Check strncat just once.
32810 2017-05-04 Martin Sebor <msebor@redhat.com>
32812 PR preprocessor/79214
32813 PR middle-end/79222
32814 PR middle-end/79223
32815 * builtins.c (check_sizes): Add inlining context and issue
32816 warnings even when -Wno-system-headers is set.
32817 (check_strncat_sizes): Same.
32818 (expand_builtin_strncat): Same.
32819 (expand_builtin_memmove): New function.
32820 (expand_builtin_stpncpy): Same.
32821 (expand_builtin): Handle memmove and stpncpy.
32823 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32825 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32826 which is not used any more.
32828 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32830 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32832 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32834 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32835 (cortexa53_tunings): Likewise.
32836 (cortexa57_tunings): Likewise.
32837 (cortexa72_tunings): Likewise.
32838 (cortexa73_tunings): Likewise.
32840 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32842 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32843 Set loop alignment to 8.
32845 2017-05-04 Martin Sebor <msebor@redhat.com>
32847 PR translation/80280
32848 * builtins.c (expand_builtin_object_size): Add missing quoting to
32849 %D and like directives.
32850 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32851 (hsa_type_for_tree_type): Same.
32852 (verify_function_arguments): Same.
32853 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32854 * varasm.c (get_section): Same.
32857 2017-05-04 Martin Sebor <msebor@redhat.com>
32859 PR translation/80280
32860 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32862 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32864 * config/aarch64/aarch64.c (generic_addrcost_table):
32865 Change HI/TI mode setting.
32867 2017-05-04 Martin Jambor <mjambor@suse.cz>
32869 PR tree-optimization/80622
32870 * tree-sra.c (comes_initialized_p): New function.
32871 (build_accesses_from_assign): Only set write lazily when
32872 comes_initialized_p is false.
32873 (analyze_access_subtree): Use comes_initialized_p.
32874 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32875 instead of testing for PARM_DECL.
32877 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32879 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32880 constraint on operand 0 to allow more general addressing modes.
32881 Adjust output template.
32882 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32884 * config/aarch64/aarch64-protos.h
32885 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32886 * config/aarch64/constraints.md (Dp): New address constraint.
32887 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32890 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32892 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32893 update use of estimate_ipcp_clone_size_and_time.
32894 (estimate_local_effects): Update use of
32895 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32896 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32897 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32898 Return nonspecialized time.
32900 2017-05-04 Richard Biener <rguenther@suse.de>
32902 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32903 for the last VUSE which def dominates the PHI. Directly call
32905 (get_continuation_for_phi_1): Remove.
32907 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32909 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32910 to explain the use of truncating division. Cap the number of
32911 iterations to the maximum given by nb_iterations_upper_bound,
32914 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32916 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32917 * configure: Regenerate.
32918 * config.in: Regenerate.
32919 * config/i386/driver-mingw32.c: new file.
32920 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32921 * config.host: Link driver-mingw32.o on MinGW host.
32922 * doc/install.texi: Document new --enable-mingw-wildcard configure
32925 2017-05-04 Marek Polacek <polacek@redhat.com>
32927 PR tree-optimization/80612
32928 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32930 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32931 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32933 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32934 (movt splitter): Likewise.
32935 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32936 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32937 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32938 block for Thumb-1 with MOVT.
32939 (thumb2_legitimate_address_p): Move code block ...
32940 (can_avoid_literal_pool_for_label_p): ... into this new function.
32941 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32943 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32944 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32945 "M-profile targets with the MOVT instruction".
32947 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32949 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32950 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32951 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32953 2017-05-04 Martin Liska <mliska@suse.cz>
32955 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32956 variable cond_code.
32958 2017-05-04 Richard Biener <rguenther@suse.de>
32960 * tree.c (array_at_struct_end_p): Handle arrays at struct
32961 end with flexarrays more conservatively. Refactor and treat
32962 arrays of arrays or aggregates more strict. Fix
32963 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32964 * tree.c (array_at_struct_end_p): Adjust prototype.
32965 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32966 * gimple-fold.c (get_range_strlen): Likewise.
32967 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32969 2017-05-04 Richard Biener <rguenther@suse.de>
32971 PR tree-optimization/31130
32972 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32974 (supports_overflow_infinity): Likewise.
32975 (is_negative_overflow_infinity): Likewise.
32976 (is_positive_overflow_infinity): Likewise.
32977 (is_overflow_infinity): Likewise.
32978 (stmt_overflow_infinity): Likewise.
32979 (overflow_infinity_range_p): Likewise.
32980 (usable_range_p): Remove as always returning true.
32981 (make_overflow_infinity): Remove.
32982 (negative_overflow_infinity): Likewise.
32983 (positive_overflow_infinity): Likewise.
32984 (avoid_overflow_infinity): Likewise.
32985 (set_value_range): Adjust accordingly.
32986 (set_value_range_to_nonnegative): Likewise, remove now unused
32987 overflow_infinity arg.
32988 (vrp_operand_equal_p): Adjust.
32989 (update_value_range): Likewise.
32990 (range_int_cst_singleton_p): Likewise.
32991 (operand_less_p): Likewise.
32992 (compare_values_warnv): Likewise.
32993 (extract_range_for_var_from_comparison_expr): Likewise.
32994 (vrp_int_const_binop): Likewise.
32995 (zero_nonzero_bits_from_vr): Likewise.
32996 (extract_range_from_multiplicative_op_1): Likewise.
32997 (extract_range_from_binary_expr_1): Likewise.
32998 (extract_range_from_unary_expr): Likewise.
32999 (extract_range_from_comparison): Likewise.
33000 (extract_range_basic): Likewise.
33001 (adjust_range_with_scev): Likewise.
33002 (compare_ranges): Likewise.
33003 (compare_range_with_value): Likewise.
33004 (dump_value_range): Likewise.
33005 (test_for_singularity): Likewise, remove strict_overflow_p parameter
33007 (simplify_cond_using_ranges): Adjust.
33009 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33011 * brig-builtins.def: Added a builtin for class_f64.
33012 * builtin-types.def: Added a builtin type needed by class_f64.
33014 2017-05-03 Jason Merrill <jason@redhat.com>
33016 * timevar.def: Add TV_CONSTEXPR.
33018 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33020 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
33022 2017-05-03 Martin Jambor <mjambor@suse.cz>
33024 * ipa-prop.c (ipa_update_after_lto_read): Removed.
33025 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
33026 * ipa-cp.c (ipcp_propagate_stage): Do not call
33027 ipa_update_after_lto_read.
33028 * ipa-inline.c (ipa_inline): Likewise.
33030 2017-05-03 Martin Jambor <mjambor@suse.cz>
33032 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
33033 tag. Added a default constructor and a destructor.
33034 (ipa_edge_args_sum_t): New class;
33035 (ipa_edge_args_sum): Declare.
33036 (ipa_edge_args_vector): Remove declaration.
33037 (IPA_EDGE_REF): Use ipa_edge_args_sum.
33038 (ipa_free_edge_args_substructures): Remove declaration.
33039 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
33040 (ipa_edge_args_info_available_for_edge_p): Likewise.
33041 * ipa-prop.c (ipa_edge_args_vector): Removed.
33042 (edge_removal_hook_holder): Likewise.
33043 (edge_duplication_hook_holder): Likewise.
33044 (ipa_edge_args_sum): New variable.
33045 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
33046 ipa_edge_args_vector.
33047 (ipa_free_edge_args_substructures): Likewise.
33048 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
33049 ipa_edge_args_vector.
33050 (ipa_edge_removal_hook): Turned into method
33051 ipa_edge_args_sum_t::remove.
33052 (ipa_edge_duplication_hook): Turned into method
33053 ipa_edge_args_sum_t::duplicate.
33054 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
33055 registering edge hooks.
33056 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
33057 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
33058 ipa_edge_args_sum instead of ipa_edge_args_vector.
33059 * ipa-profile.c (ipa_profile): Likewise.
33061 2017-05-03 Martin Jambor <mjambor@suse.cz>
33063 * symbol-summary.h (function_summary): New method exists.
33064 (function_summary::symtab_removal): Deallocate through release.
33065 (call_summary): New class.
33066 (gt_ggc_mx): New overload.
33067 (gt_pch_nx): Likewise.
33068 (gt_pch_nx): Likewise.
33070 2017-05-03 Jeff Law <law@redhat.com>
33072 PR tree-optimization/78496
33073 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33074 from simplify_cond_using_ranges. Split off code to walk
33075 backwards through casts into ...
33076 (simplify_cond_using_ranges_2): New function.
33077 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33078 (execute_vrp): After identifying jump threads, call
33079 simplify_cond_using_ranges_2.
33081 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33084 * ipa-inline.h (inline_summary): Add ctor.
33085 (create_ggc): Do not use ggc_cleared_alloc.
33087 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33088 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33090 * gcc.c (handle_braces): Support escaping in switch matching
33092 * doc/invoke.texi (Spec Files): Document it.
33093 Remove superfluous @code markup in items.
33095 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33097 * diagnostic-show-locus.c (struct column_range): New struct.
33098 (get_affected_columns): New function.
33099 (get_printed_columns): New function.
33100 (struct correction): New struct.
33101 (correction::ensure_capacity): New function.
33102 (correction::ensure_terminated): New function.
33103 (struct line_corrections): New struct.
33104 (line_corrections::~line_corrections): New dtor.
33105 (line_corrections::add_hint): New function.
33106 (layout::print_trailing_fixits): Reimplement in terms of the new
33108 (selftest::test_overlapped_fixit_printing): New function.
33109 (selftest::diagnostic_show_locus_c_tests): Call it.
33111 2017-05-03 Nathan Sidwell <nathan@acm.org>
33113 Canonicalize canonical type hashing
33114 * tree.h (type_hash_canon_hash): Declare.
33115 * tree.c (type_hash_list, attribute_hash_list): Move into
33116 type_hash_canon_hash.
33117 (build_type_attribute_qual_variant): Break out hash code calc into
33118 type_hash_canon_hash.
33119 (type_hash_canon_hash): New. Generic type hash computation.
33120 (build_range_type_1, build_array_type_1, build_function_type,
33121 build_method_type_directly, build_offset_type, build_complex_type,
33122 make_vector_type): Call it.
33124 2017-05-03 Richard Biener <rguenther@suse.de>
33126 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33127 When all DRs have unknown misaligned do not always peel
33128 when there is a store but apply the same costing model as if
33129 there were only loads.
33131 2017-05-03 Richard Biener <rguenther@suse.de>
33134 PR tree-optimization/80492
33135 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33136 compare_base_decls returning dont-know properly.
33138 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33140 * config/arm/iterators.md (CCSI): New mode iterator.
33141 (arch): New mode attribute.
33142 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33143 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33144 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33145 code iterator for success result mode.
33146 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33147 the corresponding new insn generators.
33149 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33152 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33153 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33155 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33157 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33158 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33159 (DDR_A): Wrap DDR argument in brackets.
33160 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33161 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33162 (DDR_REVERSED_P): Likewise.
33164 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33166 PR tree-optimization/79472
33167 * tree-switch-conversion.c (struct switch_conv_info): Add
33168 contiguous_range and default_case_nonstandard fields.
33169 (collect_switch_conv_info): Compute contiguous_range and
33170 default_case_nonstandard fields, don't clear final_bb if
33171 contiguous_range and only the default case doesn't have the required
33173 (check_all_empty_except_final): Set default_case_nonstandard instead
33174 of failing if contiguous_range and the default case doesn't have empty
33176 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33177 and only the default case doesn't have the required constants. Skip
33179 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33180 if default_case_nonstandard.
33181 (build_constructors): Build constant 1 just once. Assert that default
33182 values aren't inserted in between cases if contiguous_range. Skip
33184 (build_arrays): Skip virtual phis.
33185 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33186 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33187 Handle virtual phis.
33188 (gen_inbound_check): Handle default_case_nonstandard case.
33189 (process_switch): Adjust check_final_bb caller. Call
33190 gather_default_values with the first non-default case instead of
33191 default case if default_case_nonstandard.
33193 2017-05-02 Nathan Sidwell <nathan@acm.org>
33195 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33196 check. Fix formatting.
33198 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33200 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33201 errors when comparing specialized and unspecialized times.
33203 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33205 * diagnostic-show-locus.c
33206 (layout::should_print_annotation_line_p): Make private.
33207 (layout::print_annotation_line): Make private.
33208 (layout::annotation_line_showed_range_p): Make private.
33209 (layout::show_ruler): Make private.
33210 (layout::print_source_line): Make private. Pass in line and
33211 line_width, rather than calling location_get_source_line. Drop
33213 (layout::print_leading_fixits): New method.
33214 (layout::print_any_fixits): Rename to...
33215 (layout::print_trailing_fixits): ...this, and make private.
33216 Don't print newline fixits.
33217 (diagnostic_show_locus): Move logic for printing one row into...
33218 (layout::print_line): ...this new function. Move the
33219 location_get_source_line call and error-handling from
33220 print_source_line to here. Call print_leading_fixits, and rename
33221 print_any_fixits to print_trailing_fixits.
33222 (selftest::test_fixit_insert_containing_newline): Update now that
33223 newlines are partially supported.
33224 (selftest::test_fixit_insert_containing_newline_2): New test.
33225 (selftest::test_fixit_replace_containing_newline): Update comments.
33226 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33227 * edit-context.c (class added_line): New class.
33228 (class edited_line): Describe newline handling in comment.
33229 (edited_line::actually_edited_p): New method.
33230 (edited_line::print_content): Delete redundant decl.
33231 (edited_line::m_predecessors): New field.
33232 (edited_file::print_content): Call edited_line::print_content.
33233 (edited_file::print_diff): Update to support newlines.
33234 (edited_file::print_diff_hunk): Likewise.
33235 (edited_file::print_run_of_changed_lines): New function.
33236 (edited_file::print_diff_line): Convert to...
33237 (print_diff_line): ...this.
33238 (edited_file::get_effective_line_count): New function.
33239 (edited_line::edited_line): Initialize new field m_predecessors.
33240 (edited_line::~edited_line): Clean up m_predecessors.
33241 (edited_line::apply_fixit): Handle newlines.
33242 (edited_line::get_effective_line_count): New function.
33243 (edited_line::print_content): New function.
33244 (edited_line::print_diff_lines): New function.
33245 (selftest::test_applying_fixits_insert_containing_newline): New
33247 (selftest::test_applying_fixits_replace_containing_newline): New
33249 (selftest::insert_line): New function.
33250 (selftest::test_applying_fixits_multiple_lines): Add example of
33252 (selftest::edit_context_c_tests): Call the new tests.
33254 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33256 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33257 parameter cand. Update dump information.
33258 (get_computation_cost): Update uses.
33260 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33262 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33263 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33264 (get_computation_at, rewrite_use_address): Update use of
33265 get_computation_aff.
33267 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33269 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33270 (get_computation): Delete.
33271 (get_computation_cost): Implement like get_computation_cost_at.
33272 Use get_computation_at.
33273 (get_computation_cost_at): Delete.
33274 (rewrite_use_nonlinear_expr): Use get_computation_at.
33275 (rewrite_use_compare, remove_unused_ivs): Ditto.
33277 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33279 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33281 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33283 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33284 (ivopts_global_cost_for_size): Rename parameter and update uses.
33285 (iv_ca_recount_cost): Update uses.
33286 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33287 candidates seperately in n_invs and n_cands.
33288 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33290 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33292 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33293 (find_inv_vars_cb): New.
33294 (find_depends): Renamed to ...
33295 (find_inv_vars): ... this.
33296 (add_candidate_1, force_var_cost): Call find_inv_vars.
33297 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33299 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33301 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33302 inv_vars. Add inv_exprs.
33303 (struct iv_cand): Rename depends_on to inv_vars.
33304 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33305 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33306 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33307 (dump_cand): Dump inv_vars.
33308 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33309 (record_invariant, find_depends, add_candidate_1): Ditto.
33310 (set_group_iv_cost, force_var_cost): Ditto.
33311 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33312 (get_computation_cost_at, get_computation_cost): Ditto.
33313 (determine_group_iv_cost_generic): Ditto.
33314 (determine_group_iv_cost_address): Ditto.
33315 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33316 (determine_group_iv_costs): Ditto.
33317 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33318 (iv_ca_set_remove_invariants): Renamed to ...
33319 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33320 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33321 (iv_ca_set_add_invariants): Renamed to ...
33322 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33323 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33324 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33325 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33326 (create_new_ivs): Remove useless dump.
33328 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33330 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33332 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33333 (iv_ca_set_no_cp, create_new_iv): Ditto.
33335 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33337 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33339 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33341 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33342 function tree_check2.
33344 2017-05-02 Martin Liska <mliska@suse.cz>
33346 * doc/gcov.texi: Add missing preposition.
33347 * gcov.c (function_info::function_info): Properly fill up
33348 all member variables.
33350 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33352 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33354 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33356 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33358 2017-05-02 Martin Liska <mliska@suse.cz>
33361 * lto-streamer-in.c (lto_read_tree_1): Remove
33362 LTO_STREAMER_DEBUG.
33363 * lto-streamer.c (struct tree_hash_entry): Likewise.
33364 (struct tree_entry_hasher): Likewise.
33365 (tree_entry_hasher::hash): Likewise.
33366 (tree_entry_hasher::equal): Likewise.
33367 (lto_streamer_init): Likewise.
33368 (lto_orig_address_map): Likewise.
33369 (lto_orig_address_get): Likewise.
33370 (lto_orig_address_remove): Likewise.
33371 * lto-streamer.h: Likewise.
33372 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33373 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33375 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33377 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33378 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33379 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33380 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33381 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33382 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33383 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33384 (mm_maskz_sub_ss): New intrinsics.
33385 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33386 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33387 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33388 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33389 (__builtin_ia32_subss_mask_round): New builtins.
33390 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33391 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33392 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33394 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33395 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33397 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33400 2017-05-02 Martin Jambor <mjambor@suse.cz>
33402 PR tree-optimization/78687
33403 * tree-sra.c (access): New field parent.
33404 (process_subtree_disqualification): New function.
33405 (disqualify_candidate): Call it.
33406 (build_accesses_from_assign): Reset write flag if creating an
33408 (build_access_subtree): Fill in parent field and also prpagate
33409 down grp_write flag.
33410 (create_artificial_child_access): New parameter set_grp_write, set
33411 grp_write to its value.
33412 (propagate_subaccesses_across_link): Also propagate grp_write flag
33414 (propagate_all_subaccesses): Push the closest parent back to work
33415 queue if add_access_to_work_queue returned true.
33417 2017-05-02 Richard Biener <rguenther@suse.de>
33419 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33420 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33421 -fstrict-overflow documentation.
33422 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33423 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33424 flag_strict_overflow.
33425 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33426 * lto-opts.c (lto_write_options): Do not stream it.
33427 * lto-wrapper.c (merge_and_complain): Do not handle it.
33428 * opts.c (default_options_table): Do not set -fstrict-overflow.
33429 (finish_options): Likewise do not clear it when sanitizing.
33430 * simplify-rtx.c (simplify_const_relational_operation): Do not
33431 test flag_strict_overflow.
33433 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33435 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33436 using enabled attribute.
33437 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33438 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33439 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33440 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33441 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33442 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33443 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33444 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33445 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33446 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33448 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33450 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33452 2017-05-02 Richard Biener <rguenther@suse.de>
33454 PR tree-optimization/80591
33456 2017-04-10 Richard Biener <rguenther@suse.de>
33458 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33461 2017-05-02 Richard Biener <rguenther@suse.de>
33463 PR tree-optimization/80549
33464 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33465 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33466 headers if they do not have a preheader.
33468 2017-05-02 Martin Liska <mliska@suse.cz>
33471 * common.opt: Fix typo.
33472 * doc/invoke.texi: Likewise.
33474 2017-05-01 Jan Beulich <jbeulich@suse.com>
33476 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33477 swapping, add (x,x,m,x,n) alternative.
33479 2017-05-01 Nathan Sidwell <nathan@acm.org>
33481 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33482 unnecessary unadjusted_alignment check.
33484 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33487 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33489 * gimplify.c (gimplify_cilk_detach): New function.
33490 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33491 * tree-core.h: Document EXPR_CILK_SPAWN.
33492 * tree.h (EXPR_CILK_SPAWN): Define.
33494 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33496 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33497 to use new fixit_hint representation, using the "replace" logic.
33498 (get_line_span_for_fixit_hint): Likewise.
33499 (layout::print_any_fixits): Likewise.
33500 (selftest::test_one_liner_many_fixits): Rename to...
33501 (selftest::test_one_liner_many_fixits_1): ...this, and update
33502 comment and expected output to reflect that the multiple fix-it
33503 hints are now consolidated into one insertion.
33504 (selftest::test_one_liner_many_fixits_2): New test.
33505 (selftest::test_diagnostic_show_locus_one_liner): Update for
33507 (selftest::test_fixit_consolidation): Update for fix-it API
33509 * diagnostic.c (print_parseable_fixits): Likewise.
33510 * edit-context.c (edited_line::m_line_events): Convert from
33511 auto_vec <line_event *> to auto_vec <line_event>.
33512 (class line_event): Convert from abstract base class to a concrete
33513 class, taking over the role of replace_event.
33514 (class insert_event): Delete.
33515 (class replace_event): Rename to class line_event. Convert to
33517 (edit_context::add_fixits): Reimplement.
33518 (edit_context::apply_insert): Delete.
33519 (edit_context::apply_replace): Rename to...
33520 (edit_context::apply_fixit): ...this. Convert to half-open range.
33521 (edited_file::apply_insert): Delete.
33522 (edited_file::apply_replace): Rename to...
33523 (edited_file::apply_fixit): ...this.
33524 (edited_line::~edited_line): Drop deletion of events.
33525 (edited_line::apply_insert): Delete.
33526 (edited_line::apply_replace): Rename to...
33527 (edited_line::apply_fixit): ...this. Convert to half-open range.
33528 Update for change to type of m_line_events.
33529 * edit-context.h (edit_context::apply_insert): Delete.
33530 (edit_context::apply_replace): Rename to...
33531 (edit_context::apply_fixit): ...this.
33533 2017-05-01 Martin Sebor <msebor@redhat.com>
33535 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33538 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33541 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33542 __get_cpuid_max returns 0.
33543 (__get_cpuid_count): Ditto.
33545 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33547 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33548 replacement expression is another instance of one of its arguments.
33550 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33553 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33554 check for stack push/pop autoinc.
33555 * config/i386/i386.c (ix86_agi_dependent): Return false
33556 if the only reason why modified_in_p returned true is that
33557 addr is SP based and set_insn is a push or pop.
33559 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33561 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33564 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33567 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33568 (account_size_time): Use two predicates - exec_pred and
33570 (evaluate_conditions_for_known_args): Compute both clause and
33572 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33573 (inline_summary_t::duplicate): Update.
33574 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33576 (compute_inline_parameters): Likewise.
33577 (estimate_edge_size_and_time): Update caluclation of time.
33578 (estimate_node_size_and_time): Compute both time and nonspecialized
33580 (estimate_ipcp_clone_size_and_time): Update.
33581 (inline_merge_summary): Update.
33582 (do_estimate_edge_time): Update.
33583 (do_estimate_edge_size): Update.
33584 (do_estimate_edge_hints): Update.
33585 (inline_read_section, inline_write_summary): Stream both new predicates.
33586 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33588 (compute_inlined_call_time): Cleanup.
33589 (big_speedup_p): Update.
33590 (edge_badness): Update.
33591 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33592 (size_time_entry): Replace predicate by exec_predicate and
33593 nonconst_predicate.
33594 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33595 (estimate_edge_time): Return also nonspec_time.
33596 (reset_edge_growth_cache): Update.
33598 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33600 PR rtl-optimization/80491
33601 * ifcvt.c (noce_process_if_block): When looking for x setter
33602 with missing else_bb, don't check only the insn right before
33603 cond_earliest, but look for the last insn that x is modified in
33604 within the same bb.
33606 PR rtl-optimization/80491
33607 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33609 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33611 PR tree-optimization/80487
33612 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33614 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33616 PR tree-optimization/79697
33617 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33618 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33619 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33621 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33622 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33624 2017-04-28 Martin Sebor <msebor@redhat.com>
33626 PR tree-optimization/80523
33627 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33628 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33630 (maybe_warn, format_directive, parse_directive): Use new functions.
33631 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33633 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33635 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33637 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33639 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33640 target_header_dir): Set correctly.
33641 * configure: Regenerated.
33642 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33643 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33644 instead of SYSTEM_HEADER_DIR.
33646 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33648 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33649 (estimate_local_effects): Likewise.
33650 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33651 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33652 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33653 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33654 do_estimate_edge_time, estimate_edge_time): Likewise.
33655 * ipa-inline-analysis.c (estimate_node_size_and_time,
33656 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33657 (estimate_time_after_inlining): Remove.
33659 2017-04-28 Martin Liska <mliska@suse.cz>
33661 * doc/gcov.texi: Enhance documentation of gcov.
33663 2017-04-28 Martin Liska <mliska@suse.cz>
33665 * doc/gcov.texi: Sort options in alphabetic order.
33666 * doc/gcov-dump.texi: Likewise.
33667 * doc/gcov-tool.texi: Likewise.
33668 * gcov.c (print_usage): Likewise.
33669 * gcov-dump.c (print_usage): Likewise.
33670 * gcov-tool.c (print_merge_usage_message): Likewise.
33671 (print_rewrite_usage_message): Likewise.
33672 (print_overlap_usage_message): Likewise.
33674 2017-04-28 Martin Liska <mliska@suse.cz>
33676 PR gcov-profile/53915
33677 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33679 2017-04-28 Martin Liska <mliska@suse.cz>
33681 PR gcov-profile/79891
33682 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33683 is marked by compiler as living on a line.
33684 (get_cycles_count): Remove usage of the union.
33685 (output_intermediate_file): Likewise.
33686 (find_source): Fix GNU coding style.
33687 (accumulate_line_counts): Remove old non-all block mode.
33688 (output_lines): Remove usage of the union.
33689 * profile.c (output_location): Include all BBs, even if
33690 belonging to a same line (and file) as a previous BB.
33692 2017-04-28 Martin Liska <mliska@suse.cz>
33694 * gcov.c (process_args): Handle new argument 'w'.
33695 (read_graph_file): Assign ID to BBs.
33696 (output_branch_count): Display BB # if verbose flag is set.
33697 (output_lines): Likewise for arcs.
33698 (print_usage): Add '--verbose' option help.
33699 * doc/gcov.texi: Document --verbose (-w) option.
33701 2017-04-28 Martin Liska <mliska@suse.cz>
33703 * gcov.c (struct block_location_info): New struct.
33704 (process_file): Fill up the new structure.
33705 (read_graph_file): Replace usage of encoding by the newly added
33707 (add_line_counts): Likewise.
33708 (accumulate_line_counts): Remove usage of the union.
33709 (function_info::function_info): New function.
33710 (function_info::~function_info): Likewise.
33711 (process_file): Call delete instead of release_function.
33712 (release_function): Release the function.
33713 (release_structures): Call delete instead of release_function.
33714 (solve_flow_graph): Replace usage of num_blocks.
33715 (find_exception_blocks): Likewise.
33716 (output_lines): Fix GNU coding style.
33718 2017-04-28 Martin Liska <mliska@suse.cz>
33721 * coverage.c (coverage_remove_note_file): New function.
33722 * coverage.h: Declare the function.
33723 * toplev.c (finalize): Clean if an error has been seen.
33725 2017-04-28 Martin Liska <mliska@suse.cz>
33727 PR gcov-profile/80031
33728 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33729 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33730 * gcov.c (read_graph_file): Read just number of blocks.
33731 * profile.c (branch_prob): Do not stream 0 flags per a basic
33734 2017-04-28 Martin Liska <mliska@suse.cz>
33736 * gcov-dump.c (tag_*): Add new argument to declarations.
33737 (dump_gcov_file): Likewise.
33738 (tag_blocks): Add and use new argument depth.
33739 (tag_arcs): Likewise.
33740 (tag_lines): Likewise.
33741 (tag_counters): Likewise.
33742 (tag_summary): Likewise.
33743 (dump_working_sets): Use depth to do a proper indentation.
33745 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33748 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33749 * symtab.c (symtab_node::debug_symtab): Move definition here.
33751 2017-04-28 Richard Biener <rguenther@suse.de>
33753 * lto-streamer.h (LTO_major_version): Bump to 7.
33755 2017-04-28 Richard Biener <rguenther@suse.de>
33757 * tree-vrp.c (assert_info): New struct.
33758 (add_assert_info): New helper.
33759 (register_edge_assert_for_2): Refactor to add asserts to a vector
33761 (register_edge_assert_for_1): Likewise.
33762 (register_edge_assert_for): Likewise.
33763 (finish_register_edge_assert_for): New helper actually registering
33764 asserts where live on edge.
33765 (find_conditional_asserts): Adjust.
33766 (find_switch_asserts): Likewise.
33767 (evrp_dom_walker::try_find_new_range): Generalize.
33768 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33770 2017-04-27 Marek Polacek <polacek@redhat.com>
33773 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33774 arg10 and arg11 to itype.
33776 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33778 * doc/extend.texi (Object Size Checking): Improve grammar.
33780 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33783 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33784 that the logic for permitting reciprocal estimates matches that
33787 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33790 * tree.c (type_cache_hasher::equal): Only compare
33791 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33792 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33793 non-aggregate element types.
33794 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33795 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33797 2017-04-27 Richard Biener <rguenther@suse.de>
33799 PR middle-end/80533
33800 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33801 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33802 keeping a reference to a trailing array.
33804 2017-04-27 Richard Biener <rguenther@suse.de>
33806 PR middle-end/80539
33807 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33808 being in loop-closed SSA form conservatively.
33809 (chrec_fold_multiply_poly_poly): Likewise.
33811 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33813 PR middle-end/79665
33814 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33815 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33817 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33820 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33821 (aarch64_function_arg_alignment): Return unsigned int again, but still
33822 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33823 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33824 Don't emit -Wpsabi note.
33825 (aarch64_function_arg_boundary): Likewise.
33826 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33829 2017-04-26 Nathan Sidwell <nathan@acm.org>
33831 * tree.h (crc32_unsigned_n): Declare.
33832 (crc32_unsigned, crc32_unsigned): Make inline.
33833 * tree.c (crc32_unsigned_bits): Replace with ...
33834 (crc32_unsigned_n): ... this.
33835 (crc32_unsigned, crc32_byte): Remove.
33836 (crc32_string): Remove unnecessary braces.
33838 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33840 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33841 * ipa-inline-analysis.c (MAX_TIME): Remove.
33842 (account_size_time): Use sreal for time.
33843 (dump_inline_summary): Update.
33844 (estimate_function_body_sizes): Update.
33845 (estimate_edge_size_and_time): Update.
33846 (estimate_calls_size_and_time): Update.
33847 (estimate_node_size_and_time): Update.
33848 (inline_merge_summary): Update.
33849 (inline_update_overall_summary): Update.
33850 (estimate_time_after_inlining): Update.
33851 (inline_read_section): Update.
33852 (inline_write_summary): Update.
33853 * ipa-inline.c (compute_uninlined_call_time): Update.
33854 (compute_inlined_call_time): Update.
33855 (recursive_inlining): Update.
33856 (inline_small_functions): Update.
33857 (dump_overall_stats): Update.
33858 * ipa-inline.h: Include sreal.h.
33859 (size_time_entry): Turn time to sreal.
33860 (inline_summary): Turn self_time nad time to sreal.
33862 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33864 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33866 (sreal::stream_out, sreal::stream_in): New.
33867 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33869 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33871 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33874 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33877 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33879 (dimode_scalar_chain::compute_convert_gain): Ditto.
33880 (dimode_scalar_chain::make_vector_copies): Ditto.
33881 (dimode_scalar_chain::convert_reg): Ditto.
33882 (dimode_scalar_chain::convert_insn): Ditto.
33883 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33884 (VI248_AVX512BW_1): New mode iterator.
33885 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33886 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33889 2017-04-25 Martin Sebor <msebor@redhat.com>
33891 PR tree-optimization/80497
33892 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33893 constants are representable in HOST_WIDE_INT.
33894 (parse_directive): Ditto.
33896 2017-04-25 Martin Sebor <msebor@redhat.com>
33899 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33900 (new_zero_array): Adjust signature.
33901 (dom_info::dom_init): Used unsigned rather that size_t.
33902 (dom_info::dom_info): Same.
33904 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33905 Jakub Jelinek <jakub@redhat.com>
33908 * config/arm/arm.c: Include gimple.h.
33909 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33910 returns negative, increment ncrn only if it returned positive.
33911 (arm_needs_doubleword_align): Return int instead of bool,
33912 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33913 members, but if there is any such non-FIELD_DECL
33914 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33915 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33916 returns negative, increment nregs only if it returned positive.
33917 (arm_setup_incoming_varargs): Likewise.
33918 (arm_function_arg_boundary): Emit -Wpsabi note if
33919 arm_needs_doubleword_align returns negative, return
33920 DOUBLEWORD_ALIGNMENT only if it returned positive.
33922 2017-04-25 Marek Polacek <polacek@redhat.com>
33925 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33926 first argument to type.
33928 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33931 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33932 type checks to test for compatibility instead of equality.
33934 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33935 Jakub Jelinek <jakub@redhat.com>
33938 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33940 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33941 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33942 the alignment computation, but return their maximum in warn_alignment.
33943 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33944 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33946 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33947 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33950 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33952 * config/arc/simdext.md (dmpyh): Fix typo.
33954 2017-04-25 Richard Biener <rguenther@suse.de>
33956 PR tree-optimization/80492
33957 * alias.c (compare_base_decls): Handle registers with asm
33958 specification conservatively.
33959 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33960 compare_base_decls returning dont-know properly.
33962 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33964 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33965 (legitimate_offset_address_p): New function.
33966 (arc_legitimate_address_p): Use above function.
33968 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33970 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33972 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33974 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33975 ACCH registers whenever they are available.
33977 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33979 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33980 double regs fix when not used.
33982 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33984 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33986 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33987 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33989 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33991 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33992 when emitting short ADD instructions.
33994 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33996 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33998 (cmpsi_cc_c_insn): Likewise.
33999 (cbranchsi4_scratch): Compute proper instruction length using
34000 compact_hreg_operand.
34001 * config/arc/predicates.md (compact_hreg_operand): New predicate.
34003 2017-04-25 Richard Biener <rguenther@suse.de>
34005 PR middle-end/80509
34006 * passes.c (pass_manager::pass_manager): Initialize
34007 m_name_to_pass_map.
34009 2017-04-25 Richard Biener <rguenther@suse.de>
34011 PR tree-optimization/79201
34012 * tree-ssa-sink.c (statement_sink_location): Handle calls.
34014 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34017 * config/s390/vector.md: Split MEM->GPR vector moves for
34018 non-s_operand addresses.
34020 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34023 * config/s390/predicates.md (reload_const_wide_int_operand): New
34025 * config/s390/s390.md ("movti"): Remove d/P alternative.
34026 ("movti_bigconst"): New pattern definition.
34028 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34031 * s390-protos.h (s390_expand_cs_hqi): Removed.
34032 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
34033 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
34034 modes as well as CCZ1mode and CCZmode.
34035 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
34036 signature of s390_emit_compare_and_swap.
34037 (s390_expand_cs_hqi): Likewise, make static.
34038 (s390_expand_cs_tdsi): Generate an explicit compare before trying
34039 compare-and-swap, in some cases.
34040 (s390_expand_cs): Wrapper function.
34041 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
34043 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
34044 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
34045 patterns for small and large integers. Forbid symref memory operands.
34046 Move expander to s390.c. Require cc register.
34047 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
34048 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
34049 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
34050 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
34051 symref memory operands. Remove CC mode and call s390_match_ccmode
34053 ("atomic_exchange<mode>"): Allow and implement all integer modes.
34055 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34057 * config/s390/s390.md (define_peephole2): New peephole to help
34058 combining the load-and-test pattern with volatile memory.
34060 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34062 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34063 with CCZmode for TARGET_Z196.
34065 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34067 PR rtl-optimization/80501
34068 * combine.c (make_compound_operation_int): Set subreg_code to SET
34069 even for AND with mask of the sign bit of mode.
34071 PR rtl-optimization/80500
34072 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34073 sum's initial value.
34075 2017-04-25 Julian Brown <julian@codesourcery.com>
34076 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34078 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34080 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34082 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34084 2017-04-25 Julian Brown <julian@codesourcery.com>
34085 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34087 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34088 (thunderx2t99_sha): New Reservation.
34090 2017-04-25 Julian Brown <julian@codesourcery.com>
34091 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34093 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34094 type for 1-element load.
34096 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34098 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34100 2017-04-24 Martin Jambor <mjambor@suse.cz>
34102 PR tree-optimization/80293
34103 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34104 char arrays not totally scalarizable if it is false.
34105 (analyze_all_variable_accesses): Pass correct value in the new
34106 parameter. Add a statistics counter.
34108 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34110 PR middle-end/79931
34111 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34113 2017-04-24 Richard Biener <rguenther@suse.de>
34115 PR tree-optimization/80494
34116 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34117 out for complex types.
34119 2017-04-24 Richard Biener <rguenther@suse.de>
34121 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34122 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34123 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34124 (DFS): Adjust and never fail.
34125 (sccvn_dom_walker::fail): Remove.
34126 (sccvn_dom_walker::before_dom_children): Adjust.
34127 (run_scc_vn): Likewise and never fail.
34128 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34129 (pass_fre::execute): Likewise.
34131 2017-04-24 Richard Biener <rguenther@suse.de>
34133 PR tree-optimization/79725
34134 * tree-ssa-sink.c (statement_sink_location): Return whether
34135 failure reason was zero uses. Move that check later.
34136 (sink_code_in_bb): Deal with zero uses by removing the stmt
34139 2017-04-24 Richard Biener <rguenther@suse.de>
34142 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34143 pointer-based references.
34145 2017-04-24 Richard Biener <rguenther@suse.de>
34148 * pass_manager.h (pass_manager::operator new): Remove.
34149 (pass_manager::operator delete): Likewise.
34150 * passes.c (pass_manager::operator new): Remove.
34151 (pass_manager::operator delete): Likewise.
34152 (pass_manager::pass_manager): Zero individual pass members.
34154 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34157 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34158 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34159 Check "XEXP (src, 1)" operand here.
34160 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34161 Check "XEXP (src, 1)" operand here.
34162 (dimode_scalar_chain::make_vector_copies): Detect count register
34163 of a shift instruction. Zero extend count register from QImode
34164 to DImode to satisfy vector shift pattern count operand predicate.
34165 Substitute vector shift count operand with a DImode copy.
34166 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34169 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34171 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34172 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34173 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34174 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34175 (UNSPEC_NOREX_MEM): Remove definition.
34177 2017-04-21 Richard Biener <rguenther@suse.de>
34179 PR tree-optimization/79547
34180 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34181 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34182 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34183 without any constraints.
34185 2017-04-21 Richard Biener <rguenther@suse.de>
34187 PR tree-optimization/78847
34188 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34190 2017-04-21 Richard Biener <rguenther@suse.de>
34192 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34193 (build_distinct_type_copy): Likewise.
34194 (build_variant_type_copy): Likewise.
34195 * tree.c (build_qualified_type): Pass down mem-stat info.
34196 (build_distinct_type_copy): Likewise.
34197 (build_variant_type_copy): Likewise.
34199 2017-04-21 Richard Biener <rguenther@suse.de>
34201 PR tree-optimization/80237
34202 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34204 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34205 for a simplified result.
34207 2016-04-21 Richard Biener <rguenther@suse.de>
34209 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34210 sth as strict as a simple_iv but a chrec without symbols and an
34211 operand defined in the loop we are peeling (and not some subloop).
34212 (propagate_constants_for_unrolling): Propagate all constants.
34214 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34217 * config/i386/i386.c (print_reg): Remove assert for disalowed
34218 regno values, call output_operand_lossage instead.
34220 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34223 * config/i386/constraints.md (Yc): New register constraint.
34224 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34225 Use Yc constraint for alternative 2 of operand 0. Remove
34226 preferred_for_speed attribute.
34228 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34230 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34231 lastprivate clauses in SIMT case.
34233 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34235 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34237 2017-04-20 Richard Biener <rguenther@suse.de>
34239 PR tree-optimization/57796
34240 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34241 as N scalar stores.
34242 (vect_model_load_cost): Cost gathers as N scalar loads.
34244 2017-04-20 Richard Biener <rguenther@suse.de>
34246 * ggc-page.c (ggc_allocated_p): Rename to ...
34247 (safe_lookup_page_table_entry): ... this and return the lookup
34249 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34251 2017-04-20 Richard Biener <rguenther@suse.de>
34253 PR tree-optimization/80453
34254 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34255 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34256 from the conditions.
34257 (vn_phi_eq): Pass them down.
34258 (vn_phi_lookup): Record them.
34259 (vn_phi_insert): Likewise.
34261 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34263 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34264 uninitialized variable warning to avoid buffer overrun.
34266 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34269 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34270 is suppressed for '{ 0 }' in C.
34272 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34274 * BASE-VER: Set to 8.0.0.
34276 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34278 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34279 priority .init_array and .fini_array section with SECTION_NOTYPE
34282 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34284 PR middle-end/80423
34285 * tree.h (build_array_type): Add typeless_storage default argument.
34286 * tree.c (type_cache_hasher::equal): Also compare
34287 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34288 (build_array_type): Add typeless_storage argument, set
34289 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34291 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34292 (build_array_type): Likewise. Add typeless_storage argument.
34294 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34295 Jakub Jelinek <jakub@redhat.com>
34297 PR tree-optimization/80426
34298 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34299 operation on symbolic operands, also compute the overflow for the
34300 invariant part when the operation degenerates into a negation.
34302 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34305 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34306 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34309 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34311 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34314 * config/avr/avr.c (tree.h): Include it.
34315 (cgraph.h): Include it.
34316 (avr_encode_section_info): Don't warn for uninitialized progmem
34317 variable if it's just an alias.
34319 2017-04-19 Richard Biener <rguenther@suse.de>
34322 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34323 when needed by AutoPGO.
34325 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34328 * doc/lto.texi: Remove an extra 'that'.
34330 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34332 PR rtl-optimization/80429
34333 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34334 are only used in debug insns.
34336 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34337 Vladimir Makarov <vmakarov@redhat.com>
34339 * config/sparc/predicates.md (input_operand): Add comment. Return
34340 true for any memory operand when LRA is in progress.
34341 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34343 2017-04-18 Jeff Law <law@redhat.com>
34346 * mips.md ({return,simple_return}_internal): Do not overwrite
34349 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34351 PR tree-optimization/80443
34352 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34353 instead of adding 1, subtract -1 and similarly instead of subtracting
34356 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34358 PR rtl-optimization/80357
34359 * haifa-sched.c (tmp_bitmap): New variable.
34360 (model_recompute): Handle duplicate use records.
34361 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34362 (free_global_sched_pressure_data): Free it.
34364 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34367 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34368 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34369 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34370 instead of SYSTEM_HEADER_DIR.
34372 2017-04-18 Jeff Law <law@redhat.com>
34374 PR middle-end/80422
34375 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34376 predecessors after walking up the insn chain.
34378 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34381 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34382 sizetype type into debug info.
34384 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34387 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34388 unneeded test for TARGET_UPPER_REGS_SF.
34389 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34391 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34394 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34395 instead of gsi_after_labels.
34397 2017-04-18 Jeff Law <law@redhat.com>
34399 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34403 2017-04-13 Jeff Law <law@redhat.com>
34404 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34405 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34407 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34410 * config/avr/avr.c (intl.h): Include it.
34411 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34413 2017-04-18 Martin Liska <mliska@suse.cz>
34415 PR gcov-profile/78783
34416 * gcov-tool.c (gcov_output_files): Validate that destination
34417 file is either removed by the tool or by a user.
34419 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34420 Guy Benyei <guybe@mellanox.com>
34422 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34423 block, and do not negate it, the stored id is already negative.
34425 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34427 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34429 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34432 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34433 masks of options that should be turned off if the VSX vector
34434 options are turned off.
34435 (OTHER_P8_VECTOR_MASKS): Likewise.
34436 (OTHER_VSX_VECTOR_MASKS): Likewise.
34437 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34438 rs6000_disable_incompatible_switches to validate no type switches
34440 (rs6000_incompatible_switch): New function to disallow turning on
34441 other vector options if -mno-vsx, -mno-power8-vector, or
34442 -mno-power9-vector are specified.
34444 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34446 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34448 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34450 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34451 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34452 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34453 (ARG_POINTER_CFA_OFFSET): Likewise.
34455 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34457 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34458 conditions to take advantage of various optimizations.
34460 2017-04-13 Jeff Law <law@redhat.com>
34462 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34463 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34464 (zero_extendsidi2_dext): Likewise.
34466 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34469 * fold-const.c (fold_ternary_loc): Revert
34470 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34473 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34475 PR rtl-optimization/80343
34476 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34477 new scratch pseudo.
34479 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34482 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34483 to ubsan_encode_value.
34485 2017-04-13 Jeff Law <law@redhat.com>
34487 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34488 appearing in DEBUG_INSNs.
34490 2017-04-13 Martin Liska <mliska@suse.cz>
34492 PR gcov-profile/80413
34493 * gcov-io.c (gcov_write_string): Copy to buffer just when
34494 allocated size is greater than zero.
34496 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34499 * dwarf2out.c (decls_for_scope): Ignore declarations of
34500 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34502 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34505 * ipa-visibility.c (non_local_p): Fix typos.
34506 (localize_node): When localizing symbol in same comdat group,
34507 dissolve the group only when we know external symbols are going
34509 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34511 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34513 PR tree-optimization/79390
34514 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34515 order does not result in usable sequence, retry with reversed operand
34521 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34522 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34523 op0 instead of fold_convert_loc (loc, type, arg0).
34525 2017-04-12 Jeff Law <law@redhat.com>
34527 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34528 has a delay slot in the generated code.
34530 * config/cris/cris.md (cris_preferred_reload_class): Return
34531 GENNONACR_REGS rather than GENERAL_REGS.
34533 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34536 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34537 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34538 signedness of the result type.
34540 2017-04-12 Richard Biener <rguenther@suse.de>
34541 Jeff Law <law@redhat.com>
34543 PR tree-optimization/80359
34544 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34545 trim stores to TARGET_MEM_REFs.
34547 2017-04-12 Richard Biener <rguenther@suse.de>
34549 PR tree-optimization/79390
34550 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34551 threading case even more.
34553 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34556 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34557 for quad_address_p for TImode, instead of just not indexed_address.
34559 2017-04-12 Richard Biener <rguenther@suse.de>
34560 Bernd Edlinger <bernd.edlinger@hotmail.de>
34562 PR middle-end/79671
34563 * alias.c (component_uses_parent_alias_set_from): Handle
34564 TYPE_TYPELESS_STORAGE.
34565 (get_alias_set): Likewise.
34566 * tree-core.h (tree_type_common): Add typeless_storage flag.
34567 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34568 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34569 for types containing members with TYPE_TYPELESS_STORAGE.
34570 (place_field): Likewise.
34571 (layout_type): Likewise for ARRAY_TYPE.
34572 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34573 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34574 TYPE_TYPELESS_STORAGE.
34575 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34577 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34580 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34581 first argument to type.
34583 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34587 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34588 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34589 (rs6000_expand_binop_builtin): Likewise.
34590 (rs6000_expand_ternop_builtin): Likewise; also add missing
34591 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34592 vshasigma built-ins.
34593 * doc/extend.texi: Document that vec_xxpermdi's third argument
34594 must be a constant.
34596 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34598 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34599 Use shift_const cost parameter when calculating gain of STV shifts.
34601 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34603 PR rtl-optimization/70478
34604 * lra-constraints.c (process_alt_operands): Check memory for
34605 disfavoring memory insn operand.
34607 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34609 PR middle-end/80100
34610 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34611 left shift in unsigned HOST_WIDE_INT type.
34613 PR rtl-optimization/80385
34614 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34615 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34618 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34619 if they have any depend clauses.
34621 2017-04-11 Martin Liska <mliska@suse.cz>
34624 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34625 * ipa-split.c (split_function): Create a local comdat symbol
34626 if caller is in a comdat group.
34628 2017-04-11 Martin Liska <mliska@suse.cz>
34631 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34634 2017-04-11 Martin Sebor <msebor@redhat.com>
34636 PR middle-end/80364
34637 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34638 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34640 (directive::set_width, directive::set_precision, format_character):
34642 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34645 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34648 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34649 conflict, set target->arch_name instead of target->cpu_name.
34651 2017-04-11 Richard Biener <rguenther@suse.de>
34653 PR tree-optimization/80374
34654 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34655 build_zero_cst, remove fold_convertible_p check again.
34657 2017-04-11 Martin Liska <mliska@suse.cz>
34660 * ubsan.c (instrument_object_size): Do not instrument register
34663 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34666 * config/i386/i386-builtin-types.def
34667 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34668 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34669 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34670 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34671 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34672 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34673 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34674 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34675 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34676 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34677 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34678 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34679 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34680 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34681 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34682 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34683 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34684 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34685 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34686 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34687 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34688 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34689 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34690 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34691 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34692 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34693 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34694 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34695 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34696 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34697 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34698 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34699 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34700 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34701 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34702 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34703 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34704 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34705 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34706 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34707 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34708 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34709 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34710 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34711 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34712 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34714 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34715 flag to second_arg_count, handle 4 argument function type _COUNT
34716 aliases, handle second_arg_count on second argument rather than last.
34718 2017-04-10 Jeff Law <law@redhat.com>
34720 PR tree-optimization/80374
34721 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34722 record anything if we can not convert integer_zero_node to the
34725 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34728 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34729 Enhance special handling given to the TARGET_P9_MINMAX option in
34730 relation to certain other options.
34732 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34734 PR tree-optimization/80153
34735 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34736 remove POINTER_PLUS_EXPR's base part directly, rather than through
34739 2017-04-10 Richard Biener <rguenther@suse.de>
34740 Bin Cheng <bin.cheng@arm.com>
34742 PR tree-optimization/80153
34743 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34744 the first element of pointer type aff_tree. Build result expr in
34746 (add_elt_to_tree): Convert to type unconditionally. Remove other
34747 fold_convert calls.
34748 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34749 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34751 2017-04-10 Richard Biener <rguenther@suse.de>
34753 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34756 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34758 PR rtl-optimization/70478
34759 * lra-constraints.c (curr_small_class_check): New.
34760 (update_and_check_small_class_inputs): New.
34761 (process_alt_operands): Update curr_small_class_check. Disfavor
34762 alternative insn memory operands. Check available regs for small
34765 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34768 * config/mips/mips.opt (-mvirt): Update description.
34769 * doc/invoke.texi (-mvirt): Likewise.
34771 2017-04-10 Richard Biener <rguenther@suse.de>
34773 PR middle-end/80362
34774 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34775 looking for NEGATE_EXPR in -A / -B to A / B folding.
34777 2017-04-10 Martin Liska <mliska@suse.cz>
34779 PR gcov-profile/80224
34780 * gcov.c (print_usage): Fix usage string.
34781 (get_gcov_intermediate_filename): Remove.
34782 (output_gcov_file): Use both for normal and intermediate format.
34783 (generate_results): Do not initialize special file for
34784 intermediate format.
34786 2017-04-10 Richard Biener <rguenther@suse.de>
34788 PR tree-optimization/80304
34789 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34792 2017-04-10 Nathan Sidwell <nathan@acm.org>
34795 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34796 (rs6000_init_builtins): Use it.
34798 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34800 * config/arm/arm.md (<mrc>): Add mode to SET source.
34801 (<mrrc>): Likewise.
34803 2017-04-10 Richard Biener <rguenther@suse.de>
34805 PR middle-end/80344
34806 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34808 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34811 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34812 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34813 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34814 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34815 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34816 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34817 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34818 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34819 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34820 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34821 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34822 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34823 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34824 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34825 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34826 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34827 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34828 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34829 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34830 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34831 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34832 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34833 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34835 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34837 PR rtl-optimization/70478
34838 * lra-constraints.c: Reverse the last patch.
34840 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34842 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34843 Add comment for WCHAR_T.
34845 2017-04-08 Martin Liska <mliska@suse.cz>
34848 2017-04-07 Martin Liska <mliska@suse.cz>
34851 * ipa-split.c (split_function): Add function part to a same comdat
34854 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34857 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34859 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34861 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34862 (vec_pairing): ...to this.
34863 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34864 (rs6000_sched_init): Adjust for name change.
34865 (struct rs6000_sched_context): Likewise.
34866 (rs6000_init_sched_context): Likewise.
34867 (rs6000_set_sched_context): Likewise.
34869 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34875 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34877 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34878 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34879 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34881 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34883 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34885 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34887 PR rtl-optimization/70703
34888 * ira-color.c (update_conflict_hard_regno_costs): Use
34889 int64_t instead of HOST_WIDE_INT.
34891 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34893 PR rtl-optimization/70478
34894 * lra-constraints.c (process_alt_operands): Disfavor alternative
34895 insn memory operands.
34897 2017-04-07 Jeff Law <law@redhat.com>
34899 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34900 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34902 2017-04-07 Martin Liska <mliska@suse.cz>
34905 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34906 Show error message instead of an ICE.
34908 2017-04-07 Martin Liska <mliska@suse.cz>
34911 * ipa-split.c (split_function): Add function part to a same comdat
34914 2017-04-07 Richard Biener <rguenther@suse.de>
34916 PR middle-end/80341
34917 * tree.c (get_unwidened): Also handle ! for_type case for
34919 * convert.c (do_narrow): Split out from ...
34920 (convert_to_integer_1): ... here. Do not pass final truncation
34921 type to get_unwidened for TRUNC_DIV_EXPR.
34923 2017-04-07 Richard Biener <rguenther@suse.de>
34925 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34927 (aff_combination_const): Adjust.
34928 (aff_combination_scale): Likewise.
34929 (aff_combination_add_elt): Likewise.
34930 (aff_combination_add_cst): Likewise.
34931 (aff_combination_convert): Likewise.
34932 (add_elt_to_tree): Likewise. Remove unused argument.
34933 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34935 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34937 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34939 * config/arm/arm.c (arm_default_short_enums): Use
34940 ARM_DEFAULT_SHORT_ENUMS.
34941 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34943 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34946 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34947 members with redundant out-of-class redeclaration.
34949 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34952 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34953 * config/i386/i386.md (*zero_extendsidi2):
34954 Add (?*x,*x) and (?*v,*v) alternatives.
34956 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34959 * config/i386/i386.c (ix86_expand_builtin)
34960 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34961 mode from insn data. Convert operands to insn operand mode.
34962 Copy operands that don't satisfy insn predicate to a register.
34964 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34966 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34969 2017-04-06 Richard Biener <rguenther@suse.de>
34971 PR tree-optimization/80334
34972 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34973 preserve alignment of accesses.
34975 2017-04-06 Richard Biener <rguenther@suse.de>
34977 PR tree-optimization/80262
34978 * tree-sra.c (build_ref_for_offset): Preserve address-space
34980 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34981 Drop useless address-space information on MEM_REF offsets.
34983 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34985 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34987 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34989 PR rtl-optimization/70703
34990 * ira-color.c (update_conflict_hard_regno_costs): Use
34991 HOST_WIDE_INT instead of long.
34993 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34996 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34997 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34999 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
35000 for x86_64 target. Handle -m3dnowa option.
35002 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35004 PR rtl-optimization/70703
35005 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
35006 (update_conflict_hard_regno_costs): Use long instead of unsigned
35007 arithmetic for cost calculation.
35009 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35010 Bernd Edlinger <bernd.edlinger@hotmail.de>
35013 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
35016 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
35019 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
35020 ptr_mode with Pmode throughout.
35021 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
35022 into probe_stack_range and use DImode.
35024 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
35027 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
35028 call_eh_return is true.
35030 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35032 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
35033 Initialize last_match_fntype_index.
35035 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35038 * tree-nvr.c: Include internal-fn.h.
35039 (pass_return_slot::execute): Ignore internal calls without
35042 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35043 Richard Biener <rguenther@suse.de>
35046 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
35047 captures used multiple times, except for the last use.
35048 * generic-match-head.c: Include gimplify.h.
35050 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35052 PR tree-optimization/79390
35053 * target.h (struct noce_if_info): Declare.
35054 * targhooks.h (default_noce_conversion_profitable_p): Declare.
35055 * target.def (noce_conversion_profitable_p): New target hook.
35056 * ifcvt.h (struct noce_if_info): New type, moved from ...
35057 * ifcvt.c (struct noce_if_info): ... here.
35058 (noce_conversion_profitable_p): Renamed to ...
35059 (default_noce_conversion_profitable_p): ... this. No longer
35061 (noce_try_store_flag_constants, noce_try_addcc,
35062 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35063 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35064 instead of noce_conversion_profitable_p.
35065 * config/i386/i386.c: Include ifcvt.h.
35066 (ix86_option_override_internal): Don't override
35067 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35068 (ix86_noce_conversion_profitable_p): New function.
35069 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35070 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35071 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35072 * doc/tm.texi: Regenerated.
35074 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35076 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35079 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35082 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35083 instructions for small multiply cores.
35085 2017-04-04 Jeff Law <law@redhat.com>
35087 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35089 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35090 that are not set by the loop over the elements.
35092 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35095 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35096 int mode, convert_modes it to mode as unsigned, otherwise use
35097 lowpart_subreg to mode rather than SImode.
35098 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35099 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35100 Use DImode instead of SImode for the shift count operand.
35101 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35104 2017-04-04 Richard Biener <rguenther@suse.de>
35106 PR middle-end/80281
35107 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35108 arithmetic done for the negate or the plus. Simplify.
35109 (A - (-B) -> A + B): Likewise.
35110 * fold-const.c (split_tree): Make sure to not negate pointers.
35112 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35114 PR rtl-optimization/60818
35115 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35116 a compare of comparisons with the thing compared if this results
35117 in a different machine mode.
35119 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35121 * alias.c (base_alias_check): Fix typo in comment.
35122 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35123 * cgraphunit.c (symbol_table::compile): Likewise.
35124 * collect2.c (maybe_run_lto_and_relink): Likewise.
35125 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35126 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35127 * config/avr/avr.c (avr_map_op_t): Likewise.
35128 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35129 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35130 * config/epiphany/epiphany.md (movcc): Likewise.
35131 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35132 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35134 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35135 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35136 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35137 * config/sh/sh.c (sh_rtx_costs): Likewise.
35138 * fold-const.c (fold_truth_andor): Likewise.
35139 * genautomata.c (collapse_flag): Likewise.
35140 * gengtype.h (struct type::u::s): Likewise.
35141 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35142 * input.c (FORMAT_AMOUNT): Likewise.
35143 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35144 (known_aggs_to_agg_replacement_list): Likewise.
35145 * ipa-inline-analysis.c: Likewise.
35146 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35147 * ipa-polymorphic-call.c
35148 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35149 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35150 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35152 * modulo-sched.c (apply_reg_moves): Likewise.
35153 * omp-expand.c (build_omp_regions_1): Likewise.
35154 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35155 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35156 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35157 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35158 * value-prof.c: Likewise.
35159 * var-tracking.c (val_reset): Likewise.
35161 2017-04-03 Richard Biener <rguenther@suse.de>
35163 PR tree-optimization/80275
35164 * fold-const.c (split_address_to_core_and_offset): Handle
35167 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35169 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35170 descriptors is at least equal to that of functions.
35172 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35174 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35176 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35179 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35180 (mov<IMOD4:mode>): New expander.
35181 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35183 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35185 PR rtl-optimization/79405
35186 * fwprop.c (propagations_left): New variable.
35187 (forward_propagate_into): Decrement it.
35188 (fwprop_init): Initialize it.
35189 (fw_prop): If the variable has reached zero, stop propagating.
35190 (fwprop_addr): Ditto.
35192 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35195 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35196 a FUNCTION_DECL, pass it as decl instead of origin to
35199 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35201 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35204 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35207 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35208 TARGET_VSX_SMALL_INTEGER.
35210 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35212 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35213 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35215 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35217 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35218 extraction from odd-numbered MSA register.
35220 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35222 PR middle-end/80173
35223 * expmed.c (store_bit_field_1): Don't attempt to create
35224 a word subreg out of hard registers wider than word if they
35225 have HARD_REGNO_NREGS of 1 for their mode.
35227 PR middle-end/80163
35228 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35229 conversions to integer types wider than word and pointer.
35232 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35233 (rtx_equal_for_cselib_p): Pass 0 to it.
35234 * cselib.c (cselib_hasher::equal): Likewise.
35235 (rtx_equal_for_cselib_1): Add depth argument. If depth
35236 is 128, don't look up VALUE locs and punt. Increment
35237 depth in recursive calls when walking VALUE locs.
35239 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35241 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35242 (make_gcov_file_name): Use the canonical path name for generating
35244 (read_line): Fix handling of files with ascii null bytes.
35246 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35248 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35249 to initialise a vector register instead
35250 of using a const_int.
35252 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35254 PR translation/80189
35255 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35256 diagnostic messages.
35258 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35261 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35262 (dfp_diex_<mode>): Update mode of operand 1.
35263 * doc/extend.texi (dxex, dxexq): Document change to return type.
35264 (diex, diexq): Document change to argument type.
35266 2017-03-30 Martin Jambor <mjambor@suse.cz>
35269 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35270 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35271 it reflects the signature changes performed at the callee side.
35272 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35273 to cgraph_build_function_type_skip_args.
35274 (build_function_decl_skip_args): Adjust call to the above function.
35276 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35279 * config/i386/sse.md
35280 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35281 register as dest whenever it is a MEM not rtx_equal_p to the
35282 corresponding dup operand, and when forcing into reg move the
35283 reg into the memory afterwards.
35284 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35285 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35286 for the force_reg mode.
35287 (avx512vl_vextractf128<mode>): Use register as dest either
35288 always when a MEM, or when it is a MEM not rtx_equal_p to the
35289 corresponding dup operand, or even not when it is a CONST_VECTOR
35290 depending on the mode and lo vs. hi.
35291 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35293 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35294 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35295 Likewise. Require that operands[2] is even.
35296 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35297 Remove extraneous parens. Require that operands[2] is a multiple
35299 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35300 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35301 disallow memory then.
35303 2017-03-30 Richard Biener <rguenther@suse.de>
35305 PR tree-optimization/77498
35306 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35307 to non-constants over backedges.
35309 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35311 PR rtl-optimization/80233
35312 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35313 as last_combined_insn. Do not test for BARRIER_P separately.
35315 2017-03-29 Andreas Schwab <schwab@suse.de>
35318 * calls.c (prepare_call_address): Convert funexp to Pmode before
35319 copying to temp reg.
35321 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35323 PR tree-optimization/80158
35324 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35325 Handle possible future case of more than one alternate
35327 (replace_rhs_if_not_dup): Likewise.
35328 (replace_one_candidate): Likewise.
35330 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35332 PR rtl-optimization/80193
35333 * ira.c (ira): Do not check allocation for LRA.
35335 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35337 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35338 (nvptx_output_simt_exit): Declare.
35339 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35340 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35341 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35342 (nvptx_declare_function_name): ...here. Emit declaration of local
35343 memory space buffer for omp_simt_enter insn.
35344 (nvptx_output_unisimt_switch): New.
35345 (nvptx_output_softstack_switch): New.
35346 (nvptx_output_simt_enter): New.
35347 (nvptx_output_simt_exit): New.
35348 * config/nvptx/nvptx.h (struct machine_function): New fields
35349 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35350 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35351 (UNSPECV_SIMT_EXIT): Ditto.
35352 (omp_simt_enter_insn): New insn.
35353 (omp_simt_enter): New expansion.
35354 (omp_simt_exit): New insn.
35355 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35357 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35358 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35359 (expand_GOMP_SIMT_EXIT): New.
35360 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35361 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35362 (GOMP_SIMT_EXIT): Ditto.
35363 * target-insns.def (omp_simt_enter): New insn.
35364 (omp_simt_exit): Ditto.
35365 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35367 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35368 (lower_rec_input_clauses): Likewise.
35369 (lower_lastprivate_clauses): Handle SIMT privatization.
35371 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35372 (ompdevlow_adjust_simt_enter): New.
35373 (find_simtpriv_var_op): New.
35374 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35375 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35377 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35378 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35379 (copy_decl_for_dup_finish): Ditto.
35381 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35383 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35386 * config/i386/i386.c (ix86_option_override_internal): Always
35387 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35389 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35391 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35393 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35395 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35396 mark new edge's irreducible flag accordign to it.
35397 (vect_do_peeling): Check loop preheader edge's irreducible flag
35398 and pass it to function slpeel_add_loop_guard.
35400 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35402 PR tree-optimization/80218
35403 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35404 Update block frequencies and counts.
35406 2017-03-28 Richard Biener <rguenther@suse.de>
35408 PR tree-optimization/78644
35409 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35410 of a simplification result we may not use it at all.
35412 2017-03-28 Richard Biener <rguenther@suse.de>
35415 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35416 without arguments, generate default definition of a SSA name.
35418 2017-03-28 Richard Biener <rguenther@suse.de>
35420 PR middle-end/80222
35421 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35422 TYPE_REF_CAN_ALIAS_ALL references.
35423 * fold-const.c (fold_indirect_ref_1): Likewise.
35425 2017-03-28 Martin Liska <mliska@suse.cz>
35428 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35429 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35431 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35432 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35434 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35435 (EXTRA_SPECS): Define.
35436 (SUBTARGET_EXTRA_SPECS): Likewise.
35437 (SUBTARGET_CPP_SPEC): Likewise.
35438 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35439 SUBTARGET_EXTRA_SPECS.
35440 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35442 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35444 * config/arc/simdext.md (vst64_insn): Update pattern.
35445 (vld32wh_insn): Likewise.
35446 (vld32wl_insn): Likewise.
35447 (vld64_insn): Likewise.
35448 (vld32_insn): Likewise.
35450 2017-03-28 Marek Polacek <polacek@redhat.com>
35453 * fold-const.c (fold_comparison): Use protected_set_expr_location
35454 instead of SET_EXPR_LOCATION.
35456 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35458 * tree.c (add_expr): Avoid name lookup warning.
35460 2017-03-27 Jeff Law <law@redhat.com>
35462 PR tree-optimization/80216
35463 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35464 function name. Limit recursion depth.
35465 (record_temporary_equivalences): Corresponding changes.
35467 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35469 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35472 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35475 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35477 * cfgcleanup.c (reg_note_cfa_p): New array.
35478 (insns_have_identical_cfa_notes): New function.
35479 (old_insns_match_p): Don't cross-jump in between /f
35480 and non-/f instructions. If both i1 and i2 are frame related,
35481 verify all CFA notes, their order and content.
35483 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35486 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35487 HImode and SImode with zero extend to DImode to one insn.
35488 (bswap<mode>2_extenddi): Likewise.
35489 (bswapsi2_extenddi): Likewise.
35490 (bswaphi2_extendsi): Likewise.
35491 (bswaphi2): Combine bswap HImode and SImode into one insn.
35492 Separate memory insns from swapping register.
35493 (bswapsi2): Likewise.
35494 (bswap<mode>2): Likewise.
35495 (bswaphi2_internal): Delete, no longer used.
35496 (bswapsi2_internal): Likewise.
35497 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35498 store, and gpr<-gpr swap insns.
35499 (bswap<mode>2_store): Likewise.
35500 (bswaphi2_reg): Register only splitter, combine with the splitter.
35501 (bswaphi2 splitter): Likewise.
35502 (bswapsi2_reg): Likewise.
35503 (bswapsi2 splitter): Likewise.
35504 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35505 the insns into load, store, and register/register insns.
35506 (bswapdi2_ldbrx): Likewise.
35507 (bswapdi2_load): Likewise.
35508 (bswapdi2_store): Likewise.
35509 (bswapdi2_reg): Likewise.
35511 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35513 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35514 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35516 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35519 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35521 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35522 special handling for target option conflicts between dform
35523 options (-mpower9-dform, -mpower9-dform-vector,
35524 -mpower9-dform-scalar) and -mno-direct-move.
35526 2017-03-27 Richard Biener <rguenther@suse.de>
35528 PR tree-optimization/80181
35529 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35531 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35533 * config/arc/predicates.md (move_double_src_operand): Replace the
35534 call to move_double_src_operand with a call to address_operand.
35536 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35538 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35539 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35540 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35542 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35544 * config/arc/predicates.md (long_immediate_loadstore_operand):
35545 Consider scaled addresses cases.
35547 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35549 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35550 restored when in interrupt.
35551 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35552 doesn't have delay slot.
35554 2017-03-27 Richard Biener <rguenther@suse.de>
35557 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35558 inlined thunk clones.
35560 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35563 * asan.c (instrument_derefs): Copy over last operand from
35564 original COMPONENT_REF to the new COMPONENT_REF with
35565 DECL_BIT_FIELD_REPRESENTATIVE.
35566 * ubsan.c (instrument_object_size): Likewise.
35568 2017-03-27 Richard Biener <rguenther@suse.de>
35570 PR tree-optimization/80170
35571 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35572 sure DR/SCEV didnt fold in constants we do not see when looking
35573 at the reference base alignment.
35575 2017-03-27 Richard Biener <rguenther@suse.de>
35577 PR middle-end/80171
35578 * gimple-fold.c (fold_ctor_reference): Properly guard against
35579 NULL return value from canonicalize_constructor_val.
35581 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35584 * config/i386/i386.c (ix86_expand_builtin)
35585 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35586 flags reg setting and flags reg using instructions.
35587 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35588 clobbering instructions to zero extend op2.
35590 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35592 * doc/install.texi (Configuration) <--with-aix-soname>:
35593 Update link to AIX ld.
35595 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35597 PR rtl-optimization/80160
35598 PR rtl-optimization/80159
35599 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35600 reg_alternate_class into account.
35602 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35605 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35606 to consider in curr_insn_transform.
35608 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35610 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35611 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35612 and emit_mode_inner.
35614 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35616 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35617 argument to the overloaded builtin variants. Use the new flag to
35618 deprecate certain builtin variants.
35619 * config/s390/s390-builtin-types.def: Add new builtin types.
35620 * config/s390/s390-builtins.h: Support new flags field for
35621 overloaded builtins.
35622 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35623 (s390_macro_to_expand): Enable vector float data type.
35624 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35625 builtins by incrementing the __VEC__ version number.
35626 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35628 (s390_resolve_overloaded_builtin): Emit error messages depending
35629 on the builtin flags.
35630 * config/s390/s390.c (s390_expand_builtin): Support additional
35631 flags argument. Change error message to match the messages
35632 emitted in s390-c.c.
35633 * config/s390/s390.md: New UNSPEC_* constants.
35634 (op_type): Add new instruction types.
35635 * config/s390/vecintrin.h: Add new builtins and test data class
35637 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35638 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35639 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35640 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35641 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35642 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35644 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35645 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35646 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35647 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35649 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35650 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35651 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35652 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35654 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35655 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35656 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35657 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35658 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35659 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35660 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35662 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35663 ("vec_scatter_element<V_HW_4:mode>_DI")
35664 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35665 ("vec_fpint<mode>", "vflls")
35666 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35667 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35668 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35669 ("*vec_cmphe<mode>_cc"): ... these.
35671 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35672 mode constant instead of magic value.
35674 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35676 * config/s390/s390.c (s390_expand_vec_compare): Support other
35677 vector floating point modes than just V2DF.
35678 (s390_expand_vcond): Likewise.
35679 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35680 (s390_cannot_change_mode_class): Prevent mode changes between TF
35681 and V1TF in vector registers.
35682 * config/s390/s390.md (DF, SF): New mode attributes.
35683 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35684 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35685 SFmode support for VRs.
35686 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35688 (VFT, VF_HW): New mode iterators.
35689 (vw, sdx): New mode attributes.
35690 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35691 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35692 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35693 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35694 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35695 also the new vector floating point modes. Renaming to ...
35697 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35698 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35699 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35700 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35701 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35702 ("vec_unordered<mode>"): ... these.
35704 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35705 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35706 ("*vec_extendv2df"): New insn definitions.
35708 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35710 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35711 ("mulditi3_2", "*muldi3_sign"): New patterns.
35712 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35713 rename the pattern definition.
35715 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35717 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35719 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35721 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35723 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35724 instruction if possible.
35725 * config/s390/vector.md (vec_halfnumelts): New mode
35727 ("*vec_vllezlf<mode>"): New pattern.
35729 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35731 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35732 ("popcountv4si2", "popcountv2di2"): Rename to ...
35733 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35734 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35736 ("popcount<mode>2_vxe"): New pattern.
35738 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35740 * common/config/s390/s390-common.c (processor_flags_table): Add
35742 * config.gcc: Add arch12.
35743 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35744 Default to arch12 for unknown CPU model numbers.
35745 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35746 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35747 PROCESSOR_max sanity check.
35748 * config/s390/s390-opts.h (enum processor_type): Add
35750 * config/s390/s390.c (processor_table): Add arch12.
35751 (s390_expand_builtin): Add check for B_VXE flag.
35752 (s390_issue_rate): Add PROCESSOR_ARCH12.
35753 (s390_get_sched_attrmask): Likewise.
35754 (s390_get_unit_mask): Likewise.
35755 (s390_sched_score): Enable z13 scheduling for arch12.
35756 (s390_sched_reorder): Likewise.
35757 (s390_sched_variable_issue): Likewise.
35758 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35760 (s390_tune_attr): Use z13 scheduling also for arch12.
35761 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35762 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35763 (TARGET_VXE_P): New macros.
35764 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35765 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35766 * config/s390/s390.opt: Add arch12 as processor_type.
35768 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35770 * config/s390/s390.md
35771 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35772 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35773 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35775 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35776 Rename expanders to ...
35778 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35779 ("fixuns_truncdddi2_emu"): ... these.
35781 ("fixuns_trunc<mode>si2_emu"): New expander.
35783 ("*fixuns_truncdfdi2_z13"): Rename to ...
35784 ("*fixuns_truncdfdi2_vx"): ... this.
35786 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35788 * config/s390/2964.md: Remove the single element vector compare
35789 instructions which are no longer used.
35790 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35792 (s390_canonicalize_comparison): Remove handling of DFmode
35794 (s390_expand_vec_compare_scalar): Remove function.
35795 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35796 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35798 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35800 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35802 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35804 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35805 will anyway by matched by mov<mode>_64dfp.
35807 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35809 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35810 vlef/vstef. Add missing operand to vleif.
35812 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35814 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35815 pair for all vector types with 64 bit elements.
35816 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35817 * config/s390/vector.md (V_HW_64): ... here.
35818 (V_128_NOSINGLE): New mode iterator.
35819 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35820 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35821 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35822 ("*vec_load_pairv2di"): Change to ...
35823 ("*vec_load_pair<mode>"): ... this one.
35825 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35827 * config/s390/constraints.md: Add comments.
35828 (jKK): Reject element sizes > 8 bytes.
35829 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35831 * config/s390/s390.md: Add the s_operand checks formerly in
35832 s390_split_ok_p to various splitters where they are still
35834 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35835 for 128 bit vectors. Plus two splitters.
35837 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35839 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35842 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35845 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35846 error if the boundary argument is not constant.
35848 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35850 PR rtl-optimization/80112
35851 * loop-doloop.c (doloop_condition_get): Don't check condition
35852 if cmp isn't SET with IF_THEN_ELSE src.
35854 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35856 PR tree-optimization/80158
35857 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35858 replacing a candidate statement, also replace it for the
35859 candidate's alternate interpretation.
35860 (replace_rhs_if_not_dup): Likewise.
35861 (replace_one_candidate): Likewise.
35863 2017-03-24 Richard Biener <rguenther@suse.de>
35865 PR tree-optimization/80167
35866 * graphite-isl-ast-to-gimple.c
35867 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35869 (translate_isl_ast_to_gimple::get_rename): Likewise.
35871 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35873 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35874 handling of certain combinations of target options, including the
35875 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35876 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35878 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35881 * config/arm/arm.md (*load_multiple): Add reload_completed to
35882 matching condition.
35884 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35885 Richard Biener <rguenth@suse.de>
35887 PR tree-optimization/79908
35888 PR tree-optimization/80136
35889 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35890 been cast away, gimplify_and_add suffices.
35892 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35894 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35896 2017-03-23 Richard Biener <rguenther@suse.de>
35898 PR tree-optimization/80032
35899 * gimplify.c (gimple_push_cleanup): Forced unconditional
35900 cleanups still have to go to the conditional_cleanups
35903 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35905 PR tree-optimization/80072
35906 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35908 (next_operand_entry_id): Change type to unsigned int.
35909 (sort_by_operand_rank): Make sure to return the right return value
35910 even if unsigned fields are bigger than INT_MAX.
35911 (struct oecount): Change cnt and id type to unsigned int.
35912 (oecount_hasher::equal): Formatting fix.
35913 (oecount_cmp): Make sure to return the right return value
35914 even if unsigned fields are bigger than INT_MAX.
35915 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35918 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35919 TREE_READONLY on result if writing it more than once.
35922 * doc/invoke.texi (-fsanitize=thread): Document that with
35923 -fnon-call-exceptions atomics are not able to throw
35927 * tsan.c: Include tree-eh.h.
35928 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35929 maybe_clean_or_replace_eh_stmt where needed.
35930 (instrument_memory_accesses): Add cfg_changed argument.
35931 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35932 if it returned true.
35933 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35935 PR rtl-optimization/63191
35936 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35937 wrapper function, moved the whole old content into ...
35938 (ix86_delegitimize_address_1): ... this. New inline function.
35939 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35940 true as last argument instead of ix86_delegitimize_address.
35942 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35944 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35945 cortexa57_branch_cost.
35947 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35949 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35951 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35954 * doc/md.texi (Constraints): Document wA constraint.
35955 * config/rs6000/constraints.md (wA): New.
35956 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35957 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35958 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35959 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35961 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35964 * gimplify.c (is_oacc_declared): New function.
35965 (oacc_default_clause): Use it to set default flags for acc declared
35966 variables inside parallel regions.
35967 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35968 declared variables.
35969 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35970 declare attribute to any decl as necessary.
35972 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35975 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35976 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35977 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35978 (arm_arch_lpae): This.
35979 * config/arm/arm.c (arm_arch7ve): Rename into ...
35980 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35981 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35984 2017-03-22 Martin Liska <mliska@suse.cz>
35987 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35988 error message instead of an ICE.
35990 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35992 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35994 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35996 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35998 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36001 2017-03-21 Martin Sebor <msebor@redhat.com>
36003 * doc/extend.texi: Use "cannot" instead of "can't."
36004 * doc/hostconfig.texi: Same.
36005 * doc/install.texi: Same.
36006 * doc/invoke.texi: Same.
36007 * doc/loop.texi: Same.
36008 * doc/md.texi: Same.
36009 * doc/objc.texi: Same.
36010 * doc/rtl.texi: Same.
36011 * doc/tm.texi: Same.
36012 * doc/tm.texi.in: Same.
36013 * doc/trouble.texi: Same.
36015 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
36018 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
36019 (collect_checksum_attributes): Set it.
36020 (die_checksum_ordered): Use it.
36022 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36024 PR tree-optimization/79908
36025 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
36026 change: For a VA_ARG whose LHS has been cast away, use
36027 force_gimple_operand to construct the side effects.
36029 2017-03-21 David Malcolm <dmalcolm@redhat.com>
36031 PR translation/80001
36032 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
36033 more amenable to translation.
36034 (oacc_loop_auto_partitions): Likewise.
36036 2017-03-21 Marek Polacek <polacek@redhat.com>
36037 Martin Sebor <msebor@redhat.com>
36039 PR tree-optimization/80109
36040 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
36041 on INTEGRAL_TYPE_P.
36043 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36044 Segher Boessenkool <segher@kernel.crashing.org>
36047 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
36048 check reg_used_between_p between insn and one of succ or succ2
36049 depending on if succ is artificial insn not inserted into insn
36052 2017-03-21 Martin Liska <mliska@suse.cz>
36054 PR gcov-profile/80081
36055 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
36056 * doc/gcc.texi: Include gcov-dump stuff.
36057 * doc/gcov-dump.texi: New file.
36059 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36061 PR rtl-optimization/79150
36062 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36063 conditional jump, if the jump is the last insn of the loop.
36065 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36066 Richard Biener <rguenth@suse.de>
36068 PR tree-optimization/79908
36069 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36070 been cast away, use force_gimple_operand to construct the side
36073 2017-03-21 Martin Liska <mliska@suse.cz>
36075 PR libfortran/79956
36076 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36079 2017-03-21 Brad Spengler <spender@grsecurity.net>
36082 * plugin.c (htab_hash_plugin): New function.
36083 (add_new_plugin): Use it and adjust.
36084 (parse_plugin_arg_opt): Adjust.
36085 (init_one_plugin): Likewise.
36087 2017-03-21 Richard Biener <rguenther@suse.de>
36089 PR tree-optimization/80032
36090 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36091 if set force the cleanup to happen unconditionally.
36092 (gimplify_target_expr): Push inserted clobbers with force_uncond
36093 to avoid them being removed by control-dependent DCE.
36095 2017-03-21 Richard Biener <rguenther@suse.de>
36097 PR tree-optimization/80122
36098 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36099 va_arg_pack_len when the inlined call stmt requires pack
36101 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36103 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36106 * tsan.c (instrument_builtin_call): If the memory model argument
36107 is not a constant, assume it is valid.
36110 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36113 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36115 PR rtl-optimization/79910
36116 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36117 if its dest is used by an insn before I2 (other than the combined
36118 insns themselves, which are properly handled already).
36120 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36123 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36125 * combine.c (record_used_regs): New static function.
36126 (try_combine): Handle situations where there is an additional
36127 instruction between I2 and I3 which needs to have a LOG_LINK
36131 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36133 * combine.c (try_combine): Delete redundant i1 test. Call
36134 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36136 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36139 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36140 alternatives 13/14.
36142 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36144 PR tree-optimization/80054
36145 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36146 the optimization if a PHI or any of its arguments is not dominated
36147 by the candidate's basis. Use gphi* rather than gimple* as
36149 (replace_profitable_candidates): Clean up a gimple* variable that
36150 should be a gphi* variable.
36152 2017-03-20 Martin Sebor <msebor@redhat.com>
36155 * doc/extend.texi (attribute constructor): Document present limitation.
36157 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36160 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36161 __POWER9_VECTOR__ #ifdef control, change template definition to
36162 use Power9-specific built-in function.
36163 (vec_any_eq): Likewise.
36164 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36165 to control outcomes from this test.
36166 (vector_ae_<mode>p): For VEC_F modes, likewise.
36168 2017-03-20 Ian Lance Taylor <iant@google.com>
36170 * config/i386/i386.c (ix86_function_regparm): Save an extra
36171 register for -fsplit-stack with DECL_STATIC_CHAIN.
36173 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36176 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36177 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36179 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36181 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36183 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36186 2017-03-20 Marek Polacek <polacek@redhat.com>
36189 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36191 2017-03-20 Richard Biener <rguenther@suse.de>
36193 PR tree-optimization/80113
36194 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36195 allocate extra SSA name for PHI def.
36196 (add_close_phis_to_outer_loops): Likewise.
36197 (add_close_phis_to_merge_points): Likewise.
36198 (copy_loop_close_phi_args): Likewise.
36199 (copy_cond_phi_nodes): Likewise.
36201 2017-03-20 Martin Liska <mliska@suse.cz>
36203 PR middle-end/79753
36204 * tree-chkp.c (chkp_build_returned_bound): Do not build
36205 returned bounds for a LHS that's not a BOUNDED_P type.
36207 2017-03-20 Martin Liska <mliska@suse.cz>
36211 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36212 COMPLEX_CST and VECTOR_CST.
36214 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36217 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36218 target operand. A new splitter adds the clobber statement in case
36219 the target operand is dead anyway.
36221 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36223 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36224 to age-old versions of binutils and glibc.
36226 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36228 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36230 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36232 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36234 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36236 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36237 requirement for binutils 2.13.
36239 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36241 * combine.c (try_combine): Delete redundant i1 test. Call
36242 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36244 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36246 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36247 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36249 <riscv64-*-elf>: Re-arrange section
36250 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36251 <riscv32-*-linux>: Likewise.
36252 <riscv64-*-elf>: Likewise
36253 <riscv64-*-linux>: Likewise.
36255 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36258 * aarch64.opt(verbose-cost-dump): Fix typo.
36260 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36263 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36264 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36266 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36268 * reload.c (find_reloads): When reloading a nonoffsettable address,
36269 use RELOAD_OTHER for it and its address reloads.
36271 PR rtl-optimization/79910
36272 * combine.c (record_used_regs): New static function.
36273 (try_combine): Handle situations where there is an additional
36274 instruction between I2 and I3 which needs to have a LOG_LINK
36277 2017-03-17 Jeff Law <law@redhat.com>
36279 PR tree-optimization/71437
36280 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36281 conditional in the hash table first.
36282 (vrp_dom_walker::before_dom_children): Extract condition from
36283 ASSERT_EXPR. Record condition, its inverion and any implied
36284 conditions as well.
36286 2017-03-17 Marek Polacek <polacek@redhat.com>
36287 Markus Trippelsdorf <markus@trippelsdorf.de>
36289 PR tree-optimization/80079
36290 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36293 2017-03-17 Richard Biener <rguenther@suse.de>
36295 PR middle-end/80075
36296 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36297 Properly verify the LHS before the RHS possibly claims to be
36299 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36302 2017-03-17 Martin Jambor <mjambor@suse.cz>
36304 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36305 (List of -O2 options): Likewise.
36306 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36309 2017-03-17 Tom de Vries <tom@codesourcery.com>
36311 * gcov-dump.c (print_usage): Print bug_report_url.
36313 2017-03-17 Richard Biener <rguenther@suse.de>
36315 PR middle-end/80050
36316 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36317 (parser::peek): Likewise.
36319 2017-03-17 Richard Biener <rguenther@suse.de>
36321 PR tree-optimization/80048
36322 * sese.c (free_sese_info): Properly release rename_map and
36323 copied_bb_map elements.
36325 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36327 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36328 Add linked-list forward and backlinks. Insert on
36329 construction, remove on destruction.
36330 (class pass_store_merging): Add m_stores_head field.
36331 (pass_store_merging::terminate_and_process_all_chains):
36332 Iterate over m_stores_head list.
36333 (pass_store_merging::terminate_all_aliasing_chains):
36335 (pass_store_merging::execute): Check for debug stmts first.
36336 Push new chains onto the m_stores_head stack.
36338 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36341 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36342 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36343 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36345 2017-03-16 Jeff Law <law@redhat.com>
36347 PR tree-optimization/71437
36348 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36349 member function. Implementation moved into after_dom_children
36350 member function and into the threader's thread_outgoing_edges
36352 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36353 some code into new thread_outgoing_edges.
36354 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36355 definition. Simplify marker handling (do it here). Assume we always
36356 have the available expression and the const/copies tables.
36357 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36359 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36360 * tree-vrp.c (equiv_stack): No longer file scoped.
36361 (vrp_dom_walker): New class.
36362 (vrp_dom_walker::before_dom_children): New member function.
36363 (vrp_dom_walker::after_dom_children): Likewise.
36364 (identify_jump_threads): Setup domwalker. Use it rather than
36365 walking edges in a random order by hand. Simplify setup/finalization.
36366 (finalize_jump_threads): Remove.
36367 (vrp_finalize): Do not call identify_jump_threads here.
36368 (execute_vrp): Do it here instead and call thread_through_all_blocks
36371 PR tree-optimization/71437
36372 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36374 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36376 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36377 (dom_opt_dom_walker::thread_across_edge): Remove
36378 handle_dominating_asserts argument. All callers changed.
36379 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36380 changes. Remove calls to lhs_of_dominating_assert. Other
36381 uses of handle_dominating_asserts turn into unconditional code
36382 (simplify_control_stmt_condition_1): Likewise.
36383 (simplify_control_stmt_condition): Likewise.
36384 (thread_through_normal_block, thread_across_edge): Likewise.
36385 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36386 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36387 object if it is not an SSA_NAME.
36388 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36389 before calling into the VRP specific simplifiers.
36390 (identify_jump_threads): Remove handle_dominating_asserts
36393 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36396 * tree-diagnostic.c (default_tree_printer): No longer static.
36397 * tree-diagnostic.h (default_tree_printer): New prototype.
36399 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36401 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36402 Change ins into fmov.
36404 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36406 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36407 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36408 Use h_con constraint for operand 1.
36409 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36410 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36412 2017-03-15 Jeff Law <law@redhat.com>
36414 PR tree-optimization/71437
36415 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36416 (record_temporary_equivalences): Use it.
36418 PR tree-optimization/71437
36419 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36420 tree-ssa-scopedtables.
36421 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36422 (record_conditions, record_cond, vuse_eq): Likewise.
36423 (record_edge_info): Adjust to API tweak of record_conditions.
36424 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36425 (record_temporary_equivalences, optimize_stmt): Likewise.
36426 (eliminate_redundant_computations): Likewise.
36427 (record_equivalences_from_stmt): Likewise.
36428 * tree-ssa-scopedtables.c: Include options.h and params.h.
36429 (vuse_eq): New function, moved from tree-ssa-dom.c
36430 (build_and_record_new_cond): Likewise.
36431 (record_conditions): Likewise. Accept vector of conditions rather
36432 than edge_equivalence structure for first argument.
36433 for the first argument.
36434 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36435 from tree-ssa-dom.c.
36436 (avail_exprs_stack::record_cond): Likewise.
36437 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36438 from tree-ssa-dom.c.
36439 (avail_exprs_stack): Add new member functions lookup_avail_expr
36441 (record_conditions): Declare.
36443 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36446 * lra-constraints.c (process_alt_operands): Increase reject for
36447 reloading an input/output operand.
36449 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36452 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36453 insns to convert from signed/unsigned char/short to IEEE 128-bit
36455 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36457 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36460 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36461 subreg of inner mode for values already in registers.
36463 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36465 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36466 iteration reg is used after the loop.
36468 2017-03-14 Martin Sebor <msebor@redhat.com>
36470 PR tree-optimization/79800
36471 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36472 precision in negative-positive range.
36473 (format_floating): Call non-const overload with adjusted precision.
36475 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36478 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36481 2017-03-14 Martin Sebor <msebor@redhat.com>
36483 PR middle-end/80020
36484 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36485 * builtins.def (aligned_alloc): Use it.
36488 * Makefile.in (GTFILES): Add calls.c.
36489 * calls.c: Include "gt-calls.h".
36491 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36493 PR rtl-optimization/79728
36494 * regs.h (struct target_regs): New field
36495 x_contains_allocatable_regs_of_mode.
36496 (contains_allocatable_regs_of_mode): New macro.
36497 * reginfo.c (init_reg_sets_1): Initialize it, and change
36498 contains_reg_of_mode so it includes global regs as well.
36499 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36500 rather than contains_regs_of_mode.
36502 2017-03-14 Martin Liska <mliska@suse.cz>
36504 * doc/invoke.texi: Document options that can't be combined with
36505 -fcheck-pointer-bounds.
36507 2017-03-14 Martin Liska <mliska@suse.cz>
36509 PR middle-end/79831
36510 * doc/invoke.texi (-Wchkp): Document the option.
36512 2017-03-14 Martin Liska <mliska@suse.cz>
36514 * Makefile.in: Install gcov-dump.
36516 2017-03-14 Martin Liska <mliska@suse.cz>
36518 * multiple_target.c (expand_target_clones): Bail out for
36519 an invalid attribute.
36521 2017-03-14 Richard Biener <rguenther@suse.de>
36523 * alias.c (struct alias_set_entry): Pack properly.
36524 * cfgloop.h (struct loop): Likewise.
36525 * cse.c (struct set): Likewise.
36526 * ipa-utils.c (struct searchc_env): Likewise.
36527 * loop-invariant.c (struct invariant): Likewise.
36528 * lra-remat.c (struct cand): Likewise.
36529 * recog.c (struct change_t): Likewise.
36530 * rtl.h (struct address_info): Likewise.
36531 * symbol-summary.h (function_summary): Likewise.
36532 * tree-loop-distribution.c (struct partition): Likewise.
36533 * tree-object-size.c (struct object_size_info): Likewise.
36534 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36535 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36536 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36537 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36538 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36539 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36540 (struct _stmt_vec_info): Likewise.
36542 2017-03-14 Martin Liska <mliska@suse.cz>
36545 * multiple_target.c (create_dispatcher_calls): Check that
36546 a target can create a function dispatcher.
36548 2017-03-14 Martin Liska <mliska@suse.cz>
36551 * multiple_target.c (expand_target_clones): Drop local.local
36552 flag for default implementation.
36554 2017-03-14 Richard Biener <rguenther@suse.de>
36556 PR tree-optimization/80030
36557 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36559 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36561 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36562 gcc_fallthrough() instead of __attribute__((fallthrough));
36564 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36566 * doc/gcc.texi: Remove "up" link to (DIR).
36567 * doc/gccint.texi: Ditto.
36569 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36571 * doc/install.texi (Specific) <avr>: Remove reference to
36574 2017-03-13 Jeff Law <law@redhat.com>
36576 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36577 attribute rather than comments.
36579 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36580 match_scratch operand is highest.
36582 2017-03-13 Martin Liska <mliska@suse.cz>
36584 PR middle-end/78339
36585 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36586 is a CHKP clone, use original declaration.
36588 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36590 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36591 (arc_conditional_register_usage): Use a different allocation order
36592 when optimizing for size.
36593 * common/config/arc/arc-common.c (arc_option_optimization_table):
36594 Section anchors default on when optimizing for size.
36596 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36598 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36600 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36602 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36603 * config/arc/arc.md (cpu_facility): Add cd variant.
36604 (*movqi_insn): Add code density variant.
36605 (*movhi_insn): Likewise.
36606 (*movqi_insn): Likewise.
36607 (*addsi3_mixed): Likewise.
36608 (subsi3_insn): Likewise.
36610 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36612 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36614 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36616 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36617 expressions with MINUS and UNARY ops.
36619 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36622 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36624 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36625 between vec_select and vector argument.
36626 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36627 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36628 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36629 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36630 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36631 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36633 2017-03-13 Richard Biener <rguenther@suse.de>
36636 * params.def (vect-max-peeling-for-alignment): Fix typo.
36638 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36640 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36641 issue that only occurred with binutils below 2.18.
36643 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36645 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36646 refer to binutils 2.11/2.12 minimum.
36648 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36650 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36651 ftp.kernel.org and simplify binutils requirement.
36653 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36655 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36657 (Optimize Options): Ditto. Also remove redundancy.
36659 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36661 PR translation/79848
36662 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36664 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36665 to G_ to avoid double translation.
36667 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36669 PR translation/79923
36670 * auto-profile.c (get_combined_location): Convert leading
36671 character of diagnostics to lower case and remove trailing period.
36672 (read_profile): Likewise for various diagnostics.
36673 * config/arm/arm.c (arm_option_override): Remove trailing period
36674 from various diagnostics.
36675 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36676 (msp430_expand_delay_cycles): Likewise.
36678 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36681 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36682 full command-line argument, rather than just "str".
36683 (aarch64_validate_march): Likewise.
36684 (aarch64_validate_mtune): Likewise.
36686 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36688 PR rtl-optimization/78911
36689 * lra-assigns.c (must_not_spill_p): New function.
36690 (spill_for): Use it.
36692 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36694 PR tree-optimization/79981
36695 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36696 ATOMIC_COMPARE_EXCHANGE ifn result.
36697 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36698 IFN_ATOMIC_COMPARE_EXCHANGE.
36700 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36703 * opts.c (parse_sanitizer_options): Add missing question mark to
36704 "did you mean" message.
36706 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36708 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36710 (VMULEUH_UNS): Likewise.
36711 (VMULOUB_UNS): Likewise.
36712 (VMULOUH_UNS): Likewise.
36713 * config/rs6000/rs6000.c (builtin_function_type): Remove
36714 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36716 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36719 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36720 x with result of extra_parsing_for_operand_code_0.
36721 (function_reader::extra_parsing_for_operand_code_0): Convert
36722 return type from void to rtx, returning x. When reading
36723 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36724 larger size containing struct block_symbol.
36726 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36728 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36729 -mfloat128-hardware without -m64.
36731 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36734 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36735 entries to the case statement that marks unsigned arguments to
36736 overloaded functions.
36738 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36740 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36741 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36743 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36746 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36747 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36749 2017-03-10 Martin Liska <mliska@suse.cz>
36753 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36754 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36757 2017-03-10 Olivier Hainque <hainque@adacore.com>
36759 * tree-switch-conversion (array_value_type): Start by resetting
36760 candidate type to it's main variant.
36762 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36764 PR rtl-optimization/79909
36765 * combine.c (try_combine): Use simplify_replace_rtx on individual
36766 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36767 of the whole CALL_INSN_FUNCTION_USAGE.
36769 PR tree-optimization/79972
36770 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36771 get_range_info on SSA_NAMEs. Formatting fixes.
36773 2017-03-10 Richard Biener <rguenther@suse.de>
36774 Jakub Jelinek <jakub@redhat.com>
36776 PR tree-optimization/77975
36777 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36778 edge to be constant.
36779 (get_val_for): For constant x return it. Formatting fix.
36780 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36781 would use the same bases as the current one.
36783 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36785 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36786 instead of vec_select for V1TImode.
36787 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36789 (VSX_LE_128): Add V1TI to this mode iterator.
36790 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36791 (*vsx_le_perm_store_<mode>): Likewise.
36792 (pre-reload splitter for VSX stores): Likewise.
36793 (post-reload splitter for VSX stores): Likewise.
36794 (*vsx_xxpermdi2_le_<mode>): Likewise.
36795 (*vsx_lxvd2x2_le_<mode>): Likewise.
36796 (*vsx_stxvd2x2_le_<mode>): Likewise.
36798 2017-03-09 Michael Eager <eager@eagercon.com>
36800 Correct failures with --enable-checking=yes,rtl.
36802 * config/microblaze/microblaze.c (microblaze_expand_shift):
36803 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36804 test for const0_rtx.
36805 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36806 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36808 2017-03-09 Richard Biener <rguenther@suse.de>
36810 PR tree-optimization/79977
36811 * graphite-scop-detection.c (scop_detection::merge_sese):
36812 Handle the case of extra exits to blocks dominating the entry.
36814 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36816 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36819 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36821 PR rtl-optimization/79949
36822 * lra-constraints.c (process_alt_operands): Check memory when
36823 trying to predict a cycle. Print about the overall increase.
36825 2017-03-09 Richard Biener <rguenther@suse.de>
36827 PR middle-end/79971
36828 * gimple-expr.c (useless_type_conversion_p): Preserve
36829 TYPE_SATURATING for fixed-point types.
36831 2017-03-09 Richard Biener <rguenther@suse.de>
36834 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36835 alignment of BLKmode params.
36837 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36840 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36841 (VALL_NO_V2Q): Likewise.
36843 * config/aarch64/aarch64-simd.md
36844 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36846 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36847 VALL_NO_V2Q mode iterator.
36848 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36850 2017-03-09 Martin Liska <mliska@suse.cz>
36852 PR tree-optimization/79631
36853 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36854 tree_int_cst_sign_bit just for INTEGER constants.
36856 2017-03-09 Martin Liska <mliska@suse.cz>
36860 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36863 2017-03-09 Marek Polacek <polacek@redhat.com>
36866 * tree.c (inchash::add_expr): Handle TREE_VEC.
36868 2017-03-09 Martin Liska <mliska@suse.cz>
36871 (chkp_narrow_size_and_offset): New function.
36872 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36873 (void chkp_parse_bit_field_ref): New function.
36874 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36875 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36877 2017-03-09 Martin Liska <mliska@suse.cz>
36880 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36881 (chkp_find_bounds_1): Remove gcc_unreachable.
36883 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36886 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36887 BUILT_IN_SYNC*, determine the access type from the size suffix and
36888 always build a MEM_REF with that type. Handle forgotten
36889 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36892 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36893 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36894 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36895 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36896 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36897 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36898 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36899 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36900 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36901 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36902 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36903 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36904 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36905 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36906 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36907 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36908 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36909 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36910 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36911 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36912 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36913 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36914 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36915 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36916 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36917 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36918 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36919 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36920 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36921 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36922 definitions outside of __OPTIMIZE__ guarded section.
36925 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36926 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36927 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36928 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36931 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36933 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36934 ("vfenez<mode>"): Add missing constraints.
36936 2017-03-08 Martin Sebor <msebor@redhat.com>
36939 * config/nds32/nds32.c (nds32_option_override):
36940 Fix misspelled diagnostic.
36942 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36945 * gimplify.c (gimplify_omp_for): Replace index var in outer
36946 taskloop statement with an artificial variable and add
36947 OMP_CLAUSE_PRIVATE clause for it.
36949 2017-03-08 Richard Biener <rguenther@suse.de>
36951 PR tree-optimization/79955
36952 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36953 for accesses that are completely outside of the variable.
36955 2017-03-08 Andrew Haley <aph@redhat.com>
36957 PR tree-optimization/79943
36958 * tree-ssa-loop-split.c (compute_new_first_bound): When
36959 calculating the new upper bound, (END-BEG) should be added, not
36962 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36964 * config/avr/avr.md (setmemhi): Make sure match_dup
36965 operand number comes before match_scratch.
36967 2017-03-08 Richard Biener <rguenther@suse.de>
36969 PR tree-optimization/79920
36970 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36971 with ncopies == 1 to ...
36972 (vect_transform_slp_perm_load): ... here. Properly compute
36973 all element loads by iterating VF times over the group. Do
36974 not handle ncopies (computed in a broken way) in
36975 vect_create_mask_and_perm.
36977 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36980 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36981 is a uniform vector, use uniform_vector_p return value instead of
36982 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36984 2017-03-07 Marek Polacek <polacek@redhat.com>
36986 PR middle-end/79809
36987 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36988 (alloca_call_type): Likewise.
36990 2017-03-07 Martin Liska <mliska@suse.cz>
36992 * gcov.c (process_args): Put comment to correct location.
36994 2017-03-07 Martin Liska <mliska@suse.cz>
36996 PR middle-end/68270
36997 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36998 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36999 (chkp_narrow_bounds_for_field): Likewise.
37000 (chkp_parse_array_and_component_ref): Pass one more argument to
37003 2017-03-07 Richard Biener <rguenther@suse.de>
37005 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
37008 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
37010 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
37011 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
37013 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37016 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
37017 to end of description.
37018 (PARAM_MAX_STORES_TO_MERGE): Likewise.
37020 2017-03-07 Jakub Jelinek <jakub@redhat.com>
37022 PR rtl-optimization/79901
37023 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
37025 (*avx512f_<code><mode>3<mask_name>): ... this.
37026 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
37027 iterator instead of VI8_AVX2_AVX512BW.
37029 PR rtl-optimization/79901
37030 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
37031 min/max expander, expand it using expand_vec_cond_expr.
37034 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
37037 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37040 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
37041 to void * for PCH reasons.
37042 * dwarf2out.c (output_loc_operands, output_die): Cast
37043 v.val_vec.array to unsigned char *.
37045 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
37048 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
37051 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
37053 PR rtl-optimization/79571
37054 * lra-constraints.c (process_alt_operands): Calculate static
37055 reject and subtract it from overall when only addresses will be
37058 2017-03-06 Julia Koval <julia.koval@intel.com>
37061 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37062 incoming stack boundary to 128 for 64-bit targets.
37064 2017-03-06 Richard Biener <rguenther@suse.de>
37066 PR tree-optimization/79894
37067 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37068 to NULL after folding it.
37070 2017-03-06 Richard Biener <rguenther@suse.de>
37072 PR tree-optimization/79824
37073 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37074 check disabling peeling for gaps.
37076 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37078 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37079 attributes): Document gettimeofday.
37081 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37083 * config/s390/s390.c (s390_option_override_internal): Set
37084 PARAM_MIN_VECT_LOOP_BOUND
37086 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37088 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37089 * config/s390/s390.md: Likewise.
37091 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37094 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37095 (<avx2_avx512>_perm<mode>): Rename to ...
37096 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37098 (<avx512>_perm<mode>_mask): Rename to ...
37099 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37101 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37102 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37103 instead of VI8F_256_512.
37104 (avx512f_perm<mode>): New define_expand.
37105 (avx512f_perm<mode>_mask): Likewise.
37106 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37107 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37109 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37111 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37112 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37114 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37116 2017-03-06 Martin Liska <mliska@suse.cz>
37119 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37120 when having a SSA NAME w/o VAR_DECL assigned to it.
37122 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37124 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37125 msa_dpsub_<su>_d): Fix MODE for vec_select.
37127 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37129 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37131 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37133 2017-03-06 Richard Biener <rguenther@suse.de>
37135 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37136 * plugin.c (register_plugin_info): Likewise.
37137 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37139 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37141 * config/i386/sse.md (sse_storehps, sse_storelps,
37142 avx_<castmode><avxsizesuffix>_<castmode>,
37143 avx512f_<castmode><avxsizesuffix>_<castmode>,
37144 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37145 in condition that at least one operand is not a MEM.
37147 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37149 PR middle-end/79805
37150 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37151 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37153 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37154 gimple_call_nothrow_p flag based on whether original builtin can throw.
37155 If it can, emit following stmts on the fallthrough edge.
37156 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37157 don't create new bb if inserting just debug stmts on the edge, try to
37158 insert them on the fallthru bb or just reset debug stmts.
37160 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37163 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37164 restore recog_data (including the operand rtxes inside it) around
37165 the call to get_insn_template.
37167 2017-03-03 Martin Sebor <msebor@redhat.com>
37169 PR tree-optimization/79699
37170 * context.c (context::~context): Free MPFR caches to avoid
37171 a memory leak on program exit.
37173 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37175 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37176 Use wide_int::ulow () instead of .elt (0).
37178 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37180 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37181 (*pushxf): Limit oF constraint to 32bit targets and add oC
37182 constraint for 64bit targets.
37183 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37184 (*pushdf): Change rmF constraint to rmC.
37186 2017-03-03 Martin Liska <mliska@suse.cz>
37188 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37189 Remove unused variable.
37191 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37194 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37195 is a memory operand, increase num_memory.
37196 (ix86_expand_args_builtin): Likewise.
37198 2017-03-03 Jan Hubicka <jh@suse.cz>
37201 * ipa-devirt.c (maybe_record_node): Properly handle
37202 __cxa_pure_virtual visibility.
37204 2017-03-03 Martin Liska <mliska@suse.cz>
37206 PR tree-optimization/79803
37207 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37209 (pass_loop_prefetch::execute): Disabled optimization if an
37210 assumption about L1 cache size is not met.
37212 2017-03-03 Martin Liska <mliska@suse.cz>
37214 PR rtl-optimization/79574
37215 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37216 (hash_scan_set): Likewise.
37217 (dump_hash_table): Likewise.
37218 (hoist_code): Likewise.
37220 2017-03-03 Richard Biener <rguenther@suse.de>
37222 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37223 in place of uhigh ().
37224 (fixed_convert_from_real): Likewise.
37226 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37229 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37231 2017-03-03 Richard Biener <rguenther@suse.de>
37233 PR middle-end/79818
37234 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37235 TYPE_OVERFLOW_UNDEFINED check.
37237 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37239 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37241 (vector_ae_<mode>_p): Likewise.
37242 (vector_nez_<mode>_p): Likewise.
37243 (vector_ne_v2di_p): Likewise.
37244 (vector_ae_v2di_p): Likewise.
37245 (vector_ne_<mode>_p): Likewise.
37246 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37248 (vsx_tsqrt<mode>2_fe): Likewise.
37250 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37253 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37255 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37257 PR rtl-optimization/79780
37258 * cprop.c (one_cprop_pass): When second and further conditional trap
37259 in a single basic block is turned into an unconditional trap, turn it
37260 into a deleted note to avoid RTL verification failures.
37262 2017-03-02 Richard Biener <rguenther@suse.de>
37264 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37266 2017-03-02 Richard Biener <rguenther@suse.de>
37268 PR tree-optimization/79345
37270 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37271 param and abort the walk, returning -1 if it is hit.
37272 (walk_aliased_vdefs): Take a limit param and pass it on.
37273 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37274 defaulting to 0 and return a signed int.
37275 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37276 (check_defs): New helper.
37277 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37278 about uninitialized memory.
37279 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37280 bogus uninitialized warning.
37281 (fixed_convert_from_real): Likewise.
37283 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37285 PR tree-optimization/66768
37286 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37287 iv_use if base object can't be determined.
37289 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37291 PR tree-optimization/79345
37292 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37293 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37294 (get_pattern_stats): Initialize it.
37295 * genemit.c (gen_expand): Verify match_scratch numbers come after
37296 match_operand/match_dup numbers.
37297 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37298 match_scratch numbers.
37299 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37301 * config/s390/s390.md (trunctdsd2): Likewise.
37303 2017-03-02 Richard Biener <rguenther@suse.de>
37305 * wide-int.h (wide_int_storage::operator=): Implement in terms
37308 2017-03-02 Richard Biener <rguenther@suse.de>
37310 PR tree-optimization/79777
37311 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37312 the to insert expression to sth existing.
37314 2017-03-01 Martin Sebor <msebor@redhat.com>
37316 PR middle-end/79692
37317 * gimple-ssa-sprintf.c
37318 (directive::known_width_and_precision): New function.
37319 (format_integer): Use it.
37320 (get_mpfr_format_length): Consider the full range of precision
37321 when computing %g output with the # flag. Set the likely byte
37322 count to 3 rather than 1 when precision is indeterminate.
37323 (format_floating): Correct the lower bound of precision.
37325 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37327 * doc/invoke.texi: Document default code model for 64-bit Linux.
37329 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37332 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37333 udiv rather than div since input pattern is unsigned.
37335 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37337 * config/i386/i386.c (print_reg): Warn for values of
37338 unsupported size in integer register.
37340 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37343 * config/rs6000/predicates.md (current_file_function_operand): Do
37344 not allow self calls to be local if the function is replaceable.
37346 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37349 * config/rs6000/altivec.h (vec_ctz and others): Change the
37350 preprocessor macro that controls conditional compilation from
37351 _ARCH_PWR9 to __POWER9_VECTOR__.
37352 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37353 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37354 control (instead of _ARCH_PWR9 control) so that template
37355 definition uses power9-specific function.
37356 (vec_any_eq): Likewise.
37357 (vec_all_ne): Change macro definition to use a power9-specific
37358 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37359 _ARCH_PWR9 control).
37360 (vec_any_eq) Likewise.
37361 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37362 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37363 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37364 support for xvcmpnedp instruction.
37365 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37366 macro expansion so that Power9 implementation of vec_all_ne does
37367 not use the AltiVec predicate framework.
37368 (VCMPNEH_P): Likewise.
37369 (VCMPNEW_P): Likewise.
37370 (VCMPNED_P): Likewise.
37371 (VCMPNEFP_P): Likewise.
37372 (VCMPNEDP_P): Likewise.
37373 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37374 implementation of vec_any_eq to not use AltiVec predicate
37376 (VCMPAEH_P): Likewise.
37377 (VCMPAEW_P): Likewise.
37378 (VCMPAED_P): Likewise.
37379 (VCMPAEFP_P): Likewise.
37380 (VCMPAEDP_P): Likewise.
37381 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37382 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37383 not use the AltiVec predicate framework.
37384 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37385 of vec_any_eq to not use AltiVec predicate framework.
37386 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37387 support for predefined __POWER9_VECTOR__ macro to indicate that
37388 Power9 instruction selection is enabled.
37389 (altivec_overloaded_builtins): Remove extraneous
37390 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37391 function argument types RS6000_BTI_bool_V16QI and
37392 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37393 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37394 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37395 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37396 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37397 Power9 for implementations of vec_cmpne. Change the signature for
37398 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37399 (representing vec_all_ne) to remove the previously described first
37400 argument of type RS6000_BTI_INTSI, as this was an artifact of
37401 reliance on the AltiVec predicate framework, which is no longer
37402 used in the implementation of these functions. Add
37403 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37404 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37405 since, unlike the AltiVec predicate framework implementation, we
37406 do not share function descriptors between vec_alle and vec_anyeq.
37407 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37408 set of modes that receive special treatment even when
37409 TARGET_P9_VECTOR is true. The special treatment emits code that
37410 does not depend on Power9 instructions.
37411 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37412 define_expand to not rely on AltiVec predicate framework.
37413 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37415 (vector_ne_v2di_p): Change this define_expand to not rely on
37416 AltiVec predicate framework.
37417 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37419 (vector_ne_<mode>_p): Change this define_expand to not rely on
37420 AltiVec predicate framework.
37421 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37423 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37424 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37425 define_insn pattern.
37426 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37427 define_insn pattern because the xvcmpne<VSs>. instruction is not
37429 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37430 instruction is not supported.
37432 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37434 * config/nvptx/nvptx.c: Include intl.h.
37436 2017-03-01 Martin Jambor <mjambor@suse.cz>
37439 * ipa-prop.h (ipa_bits): Removed field known.
37440 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37441 to pointers. Adjusted their comments to warn about their sharing.
37442 (ipcp_transformation_summary): Change bits to a vector of pointers.
37443 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37444 (ipa_get_ipa_bits_for_value): Declare.
37445 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37446 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37447 (ipa_bits_hash_table): Likewise.
37448 (ipa_vr_ggc_hash_traits): Likewise.
37449 (ipa_vr_hash_table): Likewise.
37450 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37451 being pointers and vr_known being removed.
37452 (ipa_set_jf_unknown): Likewise.
37453 (ipa_get_ipa_bits_for_value): New function.
37454 (ipa_set_jfunc_bits): Likewise.
37455 (ipa_get_value_range): New overloaded functions.
37456 (ipa_set_jfunc_vr): Likewise.
37457 (ipa_compute_jump_functions_for_edge): Use the above functions to
37458 construct bits and vr parts of jump functions.
37459 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37460 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37462 (ipcp_grow_transformations_if_necessary): Also allocate
37463 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37465 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37466 them. Fix too long lines.
37467 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37468 vr_known being removed.
37469 (ipa_read_jump_function): Use new setter functions to construct bits
37470 and vr parts of jump functions or set them to NULL.
37471 (write_ipcp_transformation_info): Adjust for bits being pointers.
37472 (read_ipcp_transformation_info): Likewise.
37473 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37475 Include gt-ipa-prop.h.
37476 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37478 (ipcp_store_bits_results): Likewise.
37479 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37480 Do not write to existing jump functions but use a temporary instead.
37482 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37485 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37486 attempt to use its first operand as BIT_FIELD_REF base.
37488 2017-03-01 Richard Biener <rguenther@suse.de>
37490 PR middle-end/79721
37491 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37492 interpolating formula in wrapping arithmetic.
37493 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37495 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37497 PR tree-optimization/79734
37498 * tree-vect-generic.c (expand_vector_condition): Optimize
37499 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37500 Handle VEC_COND_EXPR where comparison has different inner width from
37501 type's inner width.
37503 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37505 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37506 markup, and similar issues. Remove @opindex entries for things
37507 that aren't options. Add missing -mmpy-option entries.
37509 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37511 PR tree-optimization/79737
37512 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37513 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37514 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37515 instead of byte_size. Formatting fix.
37516 (shift_bytes_in_array_right): Formatting fix.
37518 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37521 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37522 condition on optimize for the leaf function test.
37524 2017-02-28 Martin Liska <mliska@suse.cz>
37527 * read-rtl-function.c (function_reader::handle_unknown_directive):
37528 Bail out when one uses -flto.
37530 2017-02-28 Martin Liska <mliska@suse.cz>
37532 * common.opt: Replace space with tabular for options of <number>
37534 * config/i386/i386.opt: Show <number> value for
37535 -mlarge-data-threshold.
37536 * opts.c (print_filtered_help): Do not display number in hexadecimal
37539 2017-02-28 Martin Liska <mliska@suse.cz>
37541 * common.opt: Fix --help=option -Q for options which are of
37544 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37546 * config/i386/i386.c (print_reg): Error out for values
37547 of 8-bit size in invalid integer register.
37549 2017-02-28 Martin Sebor <msebor@redhat.com>
37551 PR tree-optimization/79691
37552 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37554 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37557 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37558 gcc_unreachable with output_operand_lossage.
37560 2017-02-28 Richard Biener <rguenther@suse.de>
37562 PR tree-optimization/79740
37563 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37565 (visit_nary_op): Insert the nary into the hashtable if we
37566 pattern-matched sth.
37567 * tree-ssa-pre.c (eliminate_insert): Robustify.
37569 2017-02-28 Richard Biener <rguenther@suse.de>
37571 PR middle-end/79731
37572 * fold-const.c (decode_field_reference): Reject out-of-bound
37575 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37577 * config/i386/i386.c: Include intl.h.
37578 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37579 instead of just cond ? "..." : "...".
37580 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37581 * coverage.c (read_counts_file): Likewise.
37582 * omp-offload.c: Include intl.h.
37583 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37584 of just cond ? "..." : "...".
37585 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37586 of just cond ? "..." : "...".
37588 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37591 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37593 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37594 'tune for' CPU name.
37595 * config/arm/arm-cpu-data.h: Regenerated.
37597 2017-02-28 Richard Biener <rguenther@suse.de>
37599 PR tree-optimization/79732
37600 * tree-inline.c (expand_call_inline): Do not shadow var.
37602 2017-02-28 Richard Biener <rguenther@suse.de>
37604 PR tree-optimization/79723
37605 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37606 address-space properly.
37608 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37610 * doc/optinfo.texi (Optimization groups): Fix option used for
37612 * doc/invoke.texi (-fopt-info): Document "omp".
37613 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37614 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37615 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37616 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37617 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37619 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37621 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37622 OPTGROUP_OMP with "omp".
37624 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37627 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37628 for arithmetic shift of unsigned V2DI.
37630 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37632 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37633 arc/linux.h headers.
37634 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37635 (LINK_SPEC): Likewise.
37636 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37637 (EXTRA_SPECS): Likewise.
37638 (STARTFILE_SPEC): Likewise.
37639 (ENDFILE_SPEC): Likewise.
37640 (LIB_SPEC): Likewise.
37641 (TARGET_SDATA_DEFAULT): Likewise.
37642 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37643 (MULTILIB_DEFAULTS): Likewise.
37644 (DWARF2_UNWIND_INFO): Likewise.
37645 * config/arc/big.h: New file.
37646 * config/arc/elf.h: Likewise.
37647 * config/arc/linux.h: Likewise.
37648 * config/arc/t-uClibc: Remove.
37650 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37652 PR tree-optimization/77536
37653 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37654 (tree_transform_and_unroll_loop): Use above function to compute the
37655 estimated niter of unrolled loop and use it when scaling profile.
37656 Also use count info rather than frequency if it's non-zero.
37657 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37658 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37659 (vect_transform_loop): Call above function.
37661 2017-02-27 Richard Biener <rguenther@suse.de>
37663 PR tree-optimization/45397
37664 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37665 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37666 (visit_nary_op): Add pattern matching for CSEing sign-changed
37667 or truncated operations with wider ones.
37669 2017-02-27 Richard Biener <rguenther@suse.de>
37671 PR tree-optimization/79690
37672 * tree-vect-stmts.c (vectorizable_store): Use vector type
37673 built from the DR with address-space.
37675 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37677 * doc/invoke.texi (Optimize Options): Refine the description
37678 of asan-use-after-return.
37680 2017-02-25 Alan Modra <amodra@gmail.com>
37682 PR rtl-optimization/79584
37683 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37684 base, not ad->base_term, the reg within base. Remove assertion
37685 that ad->base == ad->base_term. Replace gen_int_mode using
37686 bogus mode with const0_rtx.
37688 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37690 PR middle-end/79396
37691 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37692 FMA_EXPR like tcc_binary or tcc_unary.
37694 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37697 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37699 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37700 (output_loc_operands): Handle DW_OP_call_ref and
37701 DW_OP_GNU_variable_value.
37702 (struct variable_value_struct): New type.
37703 (struct variable_value_hasher): Likewise.
37704 (variable_value_hash): New variable.
37705 (string_types): Remove.
37706 (copy_loc_descr): New function.
37707 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37708 (prepend_loc_descr_to_each): New function.
37709 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37710 instead of add_loc_descr_to_each if the first argument is single
37711 location list and the second has multiple.
37712 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37713 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37714 when looking for variable value which doesn't have other location info.
37715 (loc_list_from_tree): Formatting fix.
37716 (gen_array_type_die): Simplify DW_AT_string_length handling.
37717 (adjust_string_types): Remove.
37718 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37719 string_types. Call resolve_variable_values.
37720 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37721 (resolve_addr_in_expr): Likewise. Add A argument.
37722 (copy_deref_exprloc): Remove deref argument. Adjust for the
37723 original expression being DW_OP_GNU_variable_value with optionally
37724 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37725 optionally after it.
37726 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37727 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37728 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37729 (variable_value_hasher::hash, variable_value_hasher::equal): New
37731 (resolve_variable_value_in_expr, resolve_variable_value,
37732 resolve_variable_values, note_variable_value_in_expr,
37733 note_variable_value): New functions.
37734 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37737 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37740 * opts.h (handle_generated_option): Add GENERATED_P argument.
37741 * opts-common.c (handle_option): Adjust function comment.
37742 (handle_generated_option): Add GENERATED_P argument, pass it to
37744 (control_warning_option): Pass false to handle_generated_option
37746 * opts.c (maybe_default_option): Pass true to handle_generated_option
37748 * optc-gen.awk: Likewise.
37750 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37752 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37753 a REG, look at the REG it is a SUBREG of.
37754 (splitter for cmpeqsi_t): Ditto.
37756 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37758 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37759 the special USEs with the pattern of the insn, not the insn itself.
37761 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37764 * doc/invoke.texi: Document -mload-store-pairs.
37766 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37767 Sandra Loosemore <sandra@codesourcery.com>
37769 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37770 argument isn't a CONST_INT.
37771 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37772 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37773 (nios2_validate_compare): Bypass alternate compare logic if *op2
37774 is not a CONST_INT.
37775 (ldstwm_operation_p): Return false if first_base is not a REG or
37776 if first_offset is not a CONST_INT.
37778 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37780 * config/cris/cris.md: Use correct operand in a define_peephole2.
37782 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37784 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37786 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37788 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37789 this_insn if it is an INSN or JUMP_INSN.
37790 (force_offsettable): Look at base, not at addr.
37791 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37792 on things that aren't necessarily CONST_INTs.
37794 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37796 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37797 -mfpmath=sse is the default also for x86-32 targets with SSE2
37798 instruction set when @option{-ffast-math} is enabled
37800 2017-02-24 Jeff Law <law@redhat.com>
37802 PR rtl-optimizatoin/79286
37803 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37806 2017-02-24 Richard Biener <rguenther@suse.de>
37808 PR tree-optimization/79389
37809 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37812 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37814 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37815 function comment to reflect reality.
37816 (loop_exits_before_overflow): Fix typo in function description.
37818 2017-02-24 Richard Biener <rguenther@suse.de>
37820 PR tree-optimization/79389
37821 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37822 properly that a threading opportunity exists. Detect conditional
37823 copy/constant propagation opportunities.
37825 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37827 * config/visium/visium.md (type): Add trap.
37828 (b): New mode attribute.
37829 (*btst): Rename into...
37830 (*btst<mode>): ...this and adjust.
37831 (*cbranchsi4_btst_insn): Rename into...
37832 (*cbranch<mode>4_btst_insn): ...this and adjust.
37833 (trap): New define_insn.
37835 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37837 PR tree-optimization/79389
37838 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37839 (noce_reversed_cond_code): New function.
37840 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37841 reversed_comparison_code. Formatting fix.
37842 (noce_try_store_flag): Test rev_cond != NULL in addition to
37843 reversed_comparison_code.
37844 (noce_try_store_flag_constants): Likewise.
37845 (noce_try_store_flag_mask): Likewise.
37846 (noce_try_addcc): Use rev_cond if non-NULL instead of
37847 reversed_comparison_code.
37848 (noce_try_cmove_arith): Likewise. Formatting fixes.
37849 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37850 (noce_find_if_block): Initialize rev_cond.
37851 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37852 instead of false as last argument never attempt to reverse it
37855 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37857 PR tree-optimization/79663
37858 * tree-predcom.c (combine_chains): Process refs in reverse order
37859 only for ZERO length chains, and add explaining comment.
37861 2017-02-23 Jeff Law <law@redhat.com>
37863 PR tree-optimization/79578
37864 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37865 in call to operand_equal_p.
37867 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37870 * config/i386/cpuid.h: Fix another undefined behavior.
37872 2017-02-23 Richard Biener <rguenther@suse.de>
37874 PR tree-optimization/79683
37875 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37876 vector types for data-refs.
37878 2017-02-23 Martin Liska <mliska@suse.cz>
37880 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37882 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37884 PR middle-end/79665
37885 * internal-fn.c (get_range_pos_neg): Moved to ...
37886 * tree.c (get_range_pos_neg): ... here. No longer static.
37887 * tree.h (get_range_pos_neg): New prototype.
37888 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37889 are known to be in between 0 and signed maximum inclusive, try to
37890 expand both unsigned and signed divmod and use the cheaper one from
37893 2017-02-22 Jeff Law <law@redhat.com>
37895 PR tree-optimization/79578
37896 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37897 to compare base operands.
37899 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37902 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37903 gpc_reg_operand instead of fpr_reg_operand.
37905 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37907 * config/mips/mips.c (mips_return_in_memory): Force FP
37908 vector types to be returned in memory for o32 ABI.
37910 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37912 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37913 instead of DW_TAG_member for static data member declarations and don't
37914 set no_linkage_name for static inline data members.
37915 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37918 2017-02-22 Martin Liska <mliska@suse.cz>
37920 * doc/invoke.texi: Replace inequality signs with square brackets
37923 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37926 * lra-constraints.c (simplify_operand_subreg): Handle
37927 WORD_REGISTER_OPERATIONS targets.
37929 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37932 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37933 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37934 elimination by swapping fld*.
37936 2017-02-22 Richard Biener <rguenther@suse.de>
37938 PR tree-optimization/79673
37939 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37940 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37941 irrelevant address-space qualifiers and avoiding a
37942 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37944 2017-02-22 Richard Biener <rguenther@suse.de>
37946 PR tree-optimization/79666
37947 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37948 to not symbolically negate if that may introduce undefined
37951 2017-02-22 Martin Liska <mliska@suse.cz>
37954 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37955 * data-streamer-out.c (streamer_write_gcov_count_stream):
37957 * value-prof.c (stream_out_histogram_value): Make assert more
37958 precise based on type of counter.
37960 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37963 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37964 Use nonimmediate_operand instead of memory_operand for operand 1.
37965 (float-extend standard_x87sse_constant_load splitter): Ditto.
37967 2017-02-21 Jeff Law <law@redhat.com>
37969 PR tree-optimization/79621
37970 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37971 blocks with edges to themselves.
37973 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37976 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37977 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37978 Use gimple_call_builtin_p.
37981 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37982 on temporarily removed DEBUG_INSNs.
37984 PR tree-optimization/79649
37985 * tree-loop-distribution.c (classify_partition): Give up on
37986 non-generic address space loads/stores.
37988 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37990 * doc/loop.texi (Loop manipulation): Remove nonexistent
37991 tree_ssa_loop_version from the documentation.
37992 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37994 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37997 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37998 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37999 * config/rs6000/rs6000.c: Include except.h.
38000 (rs6000_expand_split_stack_prologue): Call
38001 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
38003 2017-02-21 Martin Jambor <mjambor@suse.cz>
38006 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
38007 have been analyzed.
38009 2017-02-21 Martin Jambor <mjambor@suse.cz>
38011 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
38012 for backward compatibility only.
38013 * doc/invoke.texi (Option Summary): Remove all references to
38014 -fipa-cp-alignment.
38016 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
38020 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38022 * lra-constraints.c (curr_insn_transform): Handle
38023 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38025 2017-02-21 Martin Liska <mliska@suse.cz>
38027 * config/i386/i386.opt: Replace -masm-dialect with -masm.
38029 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
38031 PR translation/79638
38032 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
38034 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
38037 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
38038 (arm_function_ok_for_sibcall): Return false for an indirect call by
38039 descriptor if all the argument registers are used.
38040 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
38041 alignment of the function.
38043 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38045 PR tree-optimization/61441
38046 * simplify-rtx.c (simplify_const_unary_operation): For
38047 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
38048 the sNaN unmodified.
38050 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
38052 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
38053 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
38054 instead of SYSTEM_HEADER_DIR.
38056 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
38057 Martin Liška <mliska@suse.cz>
38059 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
38060 Fix typos and grammar, use active voice, and clarify.
38062 2017-02-20 Marek Polacek <polacek@redhat.com>
38064 PR middle-end/79537
38065 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38068 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38070 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38073 * config/i386/i386.c (ix86_expand_builtin): Handle
38074 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38075 ix86_builtins_isa[fcode].isa as a requirement of those
38076 flags and any other flag in the bitmask.
38077 (ix86_init_mmx_sse_builtins): Use 0 instead of
38078 ~OPTION_MASK_ISA_64BIT as mask.
38079 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38080 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38081 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38082 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38084 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38087 * lra-constraints.c (split_reg): Check requested split mode
38088 is supported by the register.
38090 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38092 * lra-constraints.c (simplify_operand_subreg): Remove early
38095 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38098 * lra-constraints.c (curr_insn_transform): Tighten condition
38099 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38101 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38104 * lra-constraints.c (curr_insn_transform): Handle
38105 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38107 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38110 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38112 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38114 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38117 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38120 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38122 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38125 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38127 * final.c (last_columnnum, override_columnnum): New variables.
38128 (final_start_function): Set last_columnnum, pass it to begin_prologue
38129 hook and pass 0 to dwarf2out_begin_prologue.
38130 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38131 to source_line debug hook.
38132 (notice_source_line): Compute last_columnnum and for debug_column_info
38133 return true on column changes.
38134 * debug.h (struct gcc_debug_hooks): Add column argument to
38135 source_line and begin_prologue hooks.
38136 (debug_nothing_int_charstar_int_bool): Remove prototype.
38137 (debug_nothing_int_int_charstar,
38138 debug_nothing_int_int_charstar_int_bool): New prototypes.
38139 (dwarf2out_begin_prologue): Add column argument.
38140 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38141 begin_prologue hooks.
38142 (debug_nothing_int_charstar_int_bool): Remove.
38143 (debug_nothing_int_int_charstar,
38144 debug_nothing_int_int_charstar_int_bool): New functions.
38145 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38146 through to dwarf2out_source_line.
38147 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38148 (dwarf2out_source_line): Add column argument, emit it if requested.
38149 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38151 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38152 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38153 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38154 through to dwarf2out_begin_prologue.
38155 (vmsdbgout_source_line): Add column argument, pass it through to
38156 dwarf2out_source_line.
38157 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38158 dbxout_source_line caller.
38159 (dbxout_source_line): Add column argument.
38161 * common.opt (gno-column-info, gcolumn-info): New options.
38162 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38163 (check_die): Also test for multiple DW_AT_decl_column attributes.
38164 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38165 DW_AT_decl_column if requested.
38166 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38168 (gen_variable_die): Likewise.
38169 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38170 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38173 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38174 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38175 (ix86_handle_option): Handle OPT_m3dnowa.
38176 * doc/invoke.texi (-m3dnowa): Document.
38177 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38178 -m3dnowa instead of -m3dnow -march=athlon.
38181 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38182 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38184 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38187 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38188 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38189 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38190 generator for vsx_xxpermdi_<mode>_be.
38191 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38192 force big-endian semantics.
38193 (vsx_xxpermdi_<mode>_be): New define_expand with same
38194 implementation as previous version of vsx_xxpermdi_<mode>.
38196 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38198 PR tree-optimization/79327
38199 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38200 variable, its initialization and use.
38202 2017-02-17 Julia Koval <julia.koval@intel.com>
38204 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38205 (OPTION_MASK_ISA_PKU_UNSET): New.
38206 (ix86_handle_option): Handle -mrdpid.
38207 * config/i386/cpuid.h (bit_RDPID): New.
38208 * config/i386/driver-i386.c (host_detect_local_cpu):
38209 Detect RDPID feature.
38210 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38211 * config/i386/i386-c.c (ix86_target_macros_internal):
38213 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38214 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38215 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38216 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38217 * config/i386/i386.md (define_insn "rdpid"): New.
38218 * config/i386/i386.opt Add -mrdpid.
38219 * config/i386/immintrin.h (_rdpid_u32): New.
38221 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38223 PR rtl-optimization/79541
38224 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38225 instead of transforming it into USE.
38227 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38229 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38230 If HONOR_SNANS (SFmode) force the input to a register.
38231 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38232 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38233 an frsp or similar insn.
38235 2017-02-17 Martin Liska <mliska@suse.cz>
38237 PR rtl-optimization/79577
38238 * params.def (selsched-max-sched-times): Increase minimum to 1.
38240 2017-02-17 Martin Liska <mliska@suse.cz>
38242 PR rtl-optimization/79574
38243 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38245 2017-02-17 Martin Liska <mliska@suse.cz>
38247 PR tree-optimization/79529
38248 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38249 ssa_defined_default_def_p to handle cases which are implicitly
38251 * tree-ssa.c (ssa_defined_default_def_p): New function.
38252 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38253 which are implicitly defined.
38254 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38256 2017-02-17 Richard Biener <rguenther@suse.de>
38258 PR middle-end/79576
38259 * params.def (max-ssa-name-query-depth): Limit to 10.
38261 2017-02-17 Richard Biener <rguenther@suse.de>
38263 PR tree-optimization/79552
38264 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38267 2017-02-17 Richard Biener <rguenther@suse.de>
38270 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38272 2017-02-17 Marek Polacek <polacek@redhat.com>
38274 PR middle-end/79536
38275 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38276 (fold_negate_expr): New wrapper.
38278 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38280 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38281 Correct terminology and de-emphasize pre-standard behavior.
38283 2017-02-16 Alan Modra <amodra@gmail.com>
38285 PR rtl-optimization/79286
38286 * ira.c (def_dominates_uses): New function.
38287 (update_equiv_regs): Don't create an equivalence for insns that
38288 may trap where the register def does not dominate the use.
38290 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38292 PR rtl-optimization/78127
38293 * lra.c (lra): Call lra_eliminate before finish the loop after
38296 2017-02-16 Richard Biener <rguenther@suse.de>
38298 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38300 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38301 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38302 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38303 (isl_val_int_from_wi): New function.
38304 (extract_affine_gmp): Rename to ...
38305 (extract_affine_wi): ... this, take a widest_int.
38306 (extract_affine_int): Just wrap extract_affine_wi.
38307 (add_param_constraints): Use isl_val_int_from_wi.
38308 (add_loop_constraints): Likewise, and extract_affine_wi.
38310 2017-02-15 Jeff Law <law@redhat.com>
38312 PR middle-end/79521
38313 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38314 ira_init_register_move_cost_if_necessary.
38316 2017-02-15 Martin Sebor <msebor@redhat.com>
38318 PR middle-end/32003
38319 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38320 removed in a prior commit.
38322 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38324 PR tree-optimization/79347
38325 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38326 counters during peeling.
38328 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38330 * Makefile.in (site.exp): Remove "set ISLVER".
38332 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38335 * real.c (real_from_integer): Call real_convert even for decimal.
38337 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38340 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38342 2017-02-14 Andrew Pinski <apinski@cavium.com>
38344 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38345 cores and change the partno/implementer to be correct.
38346 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38347 the 'B" as the implementer.
38348 * config/aarch64/aarch64-tune.md: Regenerate.
38350 2017-02-14 Carl Love <cel@us.ibm.com>
38352 * config/rs6000/rs6000.c: Add case statement entry to make the
38353 xvcvuxdsp built-in argument unsigned.
38354 * config/rs6000/vsx.md: Fix the source and return operand types so they
38355 match the instruction definitions from the ISA document. Fix typo
38356 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38359 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38362 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38363 member early_clobber_alts.
38364 * lra-lives.c (reg_early_clobber_p): New.
38365 (process_bb_lives): Use it.
38366 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38367 (debug_operand_data): Initialize early_clobber_alts.
38368 (setup_operand_alternative): Set up early_clobber_alts.
38369 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38370 alternatives to new_insn_reg.
38371 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38373 (lra_update_insn_regno_info): Pass the new arg.
38375 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38377 PR middle-end/79505
38378 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38379 (new_oacc_loop_raw): Don't clear already cleared fields.
38382 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38383 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38384 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38386 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38389 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38391 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38394 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38395 the extra instruction to the right place to store 128-bit constant
38398 2017-02-14 Martin Sebor <msebor@redhat.com>
38400 PR middle-end/79448
38401 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38402 warning for strings of unknown length.
38404 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38406 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38408 2017-02-14 Jeff Law <law@redhat.com>
38411 * ira-costs.c (scan_one_insn): Initialize register move costs
38412 for pseudos seen in USE/CLOBBER insns.
38414 PR tree-optimization/79095
38415 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38416 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38417 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38418 if the operands are known to be not equal, then the resulting range
38420 (intersect_ranges): If the new range is ~[0,0] and the old range is
38421 wide, then prefer ~[0,0].
38422 * tree-vrp.c (overflow_comparison_p_1): New function.
38423 (overflow_comparison_p): New function.
38424 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38425 if NAME is used in an overflow test.
38426 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38427 overflow check that can be expressed as an equality test, then adjust
38428 ops to be that equality test.
38430 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38432 * config/s390/s390-builtin-types.def: Remove flags argument.
38433 * config/s390/s390.c (s390_init_builtins): Likewise.
38435 2017-02-14 Martin Liska <mliska@suse.cz>
38437 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38438 vector. Fix trailing white spaces.
38440 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38442 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38445 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38447 PR rtl-optimization/68664
38448 * config/arm/arm.c (arm_sched_can_speculate_insn):
38449 New function. Declare prototype.
38450 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38452 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38454 PR rtl-optimization/68664
38455 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38457 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38459 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38461 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38462 max skip bytes for function, loop and jump.
38464 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38466 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38467 ABS_EXPR for gimple dump.
38469 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38472 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38474 PR tree-optimization/79408
38475 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38476 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38477 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38478 also if rhs1 is INTEGER_CST.
38480 2017-02-14 Richard Biener <rguenther@suse.de>
38482 PR middle-end/79432
38483 * tree-into-ssa.c (insert_phi_nodes): When the function can
38484 have abnormal edges rewrite SSA names with broken use-def
38485 dominance out of SSA and register them for PHI insertion.
38487 2017-02-13 Martin Sebor <msebor@redhat.com>
38489 PR middle-end/79496
38490 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38491 clearing info.nowrite flag when snprintf size argument is a range.
38493 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38495 * cprop.c (cprop_jump): Add missing space in string literal.
38496 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38497 (get_constraint_for_component_ref): Likewise.
38498 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38499 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38500 * lra-constraints.c (process_alt_operands): Likewise.
38501 * ipa-inline.c (inline_small_functions): Likewise.
38502 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38503 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38504 * trans-mem.c (diagnose_tm_1_op): Likewise.
38505 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38506 (grid_parallel_clauses_gridifiable): Likewise.
38508 * config/nvptx/mkoffload.c (process): Add space in between
38511 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38512 "MOD4_SSE_REGS" and "ALL_REGS".
38514 * spellcheck.c (test_data): Add , in between "foo" and "food".
38516 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38519 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38520 boundary crossing check and subsequent code generation agree.
38522 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38524 * config/aarch64/aarch64.c (has_memory_op): Delete.
38525 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38528 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38530 PR rtl-optimization/79388
38531 PR rtl-optimization/79450
38532 * combine.c (distribute_notes): When removing TEM_INSN for which
38533 corresponding dest has last value recorded, invalidate that last
38536 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38538 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38539 of explicit '@'. Add missing assembly comment marker on branch costs
38542 2017-02-13 Nathan Sidwell <nathan@acm.org>
38544 * gengtype-lex.l (<in_struct>): Add '/'.
38546 2017-02-13 Martin Liska <mliska@suse.cz>
38549 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38551 2017-02-13 Richard Biener <rguenther@suse.de>
38553 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38555 * configure: Re-generate.
38556 * config.in: Likewise.
38557 * graphite-dependences.c: Simplify as if
38558 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38559 * graphite-isl-ast-to-gimple.c: Likewise.
38560 * graphite-optimize-isl.c: Likewise.
38561 * graphite-poly.c: Likewise.
38562 * graphite-sese-to-poly.c: Likewise.
38563 * graphite.h: Likewise.
38564 * toplev.c: Include isl/version.h and use isl_version () for
38565 printing the ISL version.
38566 * doc/install.texi: Update ISL requirement.
38568 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38570 * doc/standards.texi (Standards): Update reference to
38573 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38575 * doc/extend.texi (Named Address Spaces): sourceware.org now
38577 * doc/install.texi (Binaries): Ditto.
38580 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38582 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38583 standard terminology "stringize"/"stringizing" throughout.
38584 * doc/cppinternals.texi: Likewise.
38586 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38588 * doc/extend.texi: Fix some spelling mistakes and typos.
38589 * doc/invoke.texi: Likewise.
38591 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38594 * params.def (inline-min-speedup) Change from 10 to 8.
38596 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38598 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38601 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38604 * ipa-inline-analysis.c (get_minimal_bb): New function.
38605 (record_modified): Use it.
38606 (remap_edge_change_prob): Handle also ancestor functions.
38608 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38610 * doc/contrib.texi (Contributors): Remove broken link into
38611 the Mauve CVS repository.
38613 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38615 PR middle-end/79454
38616 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38617 result computation whenever lhs doesn't have vector mode, not
38618 just when it has BLKmode.
38620 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38622 * doc/makefile.texi (profiledbootstrap): Refer to the
38623 installation instructions only in textual form.
38625 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38628 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38630 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38632 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38633 (Specific): Update mingw-w64 reference.
38635 (Specific): Remove broken link to Renesas RX processor.
38637 2017-02-10 Richard Biener <rguenther@suse.de>
38639 * toplev.c (process_options): Do not mention obsolete graphite
38640 options when printing sorry message about missing graphite support.
38641 Mention -floop-nest-optimize.
38643 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38645 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38646 (vtst_p16): Likewise.
38647 (vtstq_p8): Likewise.
38648 (vtstq_p16): Likewise.
38650 (vtstq_p64): Likewise.
38651 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38652 (vset_lane_p64): New.
38653 (vsetq_lane_p64): New.
38655 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38657 PR tree-optimization/79411
38658 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38659 stmt operands are SSA_NAMEs used in abnormal phis.
38660 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38663 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38666 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38669 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38671 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38673 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38675 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38679 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38680 not arbitrary TREE_CONSTANT.
38683 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38684 "omp declare target link" attribute unless is_global_var.
38685 * omp-offload.c (find_link_var_op): Likewise.
38687 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38688 Chung-Lin Tang <cltang@codesourcery.com>
38690 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38692 (gimplify_adjust_omp_clauses): Don't delete TILE.
38693 (gimplify_omp_for): Deal with TILE.
38694 * internal-fn.c (expand_GOACC_TILE): New function.
38695 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38697 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38698 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38700 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38701 avoid DIV for outermost collapse var.
38702 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38703 Remove out of date comments, fix whitespace.
38704 * omp-general.c (omp_extract_for_data): Deal with tiling.
38705 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38706 adjust OLF_DIM_BASE value.
38707 (struct omp_for_data): Add tiling field.
38708 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38709 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38710 for auto loops. Remove default auto determining, moved to
38711 oacc_loop_fixed_partitions.
38712 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38713 stmts, add e_mask field.
38714 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38715 (oacc_thread_numbers): Use oacc_dim_call.
38716 (oacc_xform_tile): New.
38717 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38718 (finish_oacc_loop): Adjust for ifns vector.
38719 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38720 add case for GOACC_TILE fns.
38721 (oacc_loop_xform_loop): Delete.
38722 (oacc_loop_process): Iterate over call list directly, and add
38723 handling for GOACC_TILE fns.
38724 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38726 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38727 vector partitioning to outer loops. Assign 2 partitions to loops
38728 when available. Add TILE handling.
38729 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38730 (execite_oacc_device_lower): Process GOACC_TILE fns,
38731 ignore unknown specs.
38732 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38733 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38734 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38736 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38738 * configure.ac (ACX_BUGURL): Update.
38739 * configure: Regenerate.
38741 2017-02-09 Richard Biener <rguenther@suse.de>
38743 PR tree-optimization/69823
38744 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38745 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38747 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38749 * config/arc/arc-c.def: Add __NPS400__ definition.
38750 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38751 (TARGET_NPS400): Define.
38753 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38755 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38757 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38758 pointer, arch_info.
38759 (arc_cpu_types): Fill the arch_info field with a pointer into the
38760 arc_arch_types table.
38761 (arc_selected_cpu): Declare.
38762 * config/arc/arc.c (arc_selected_cpu): Make global.
38763 (arc_selected_arch): Delete.
38764 (arc_base_cpu): Delete.
38765 (arc_override_options): Remove references to deleted variables,
38766 update access to arch information.
38767 (ARC_OPT): Update access to arch information.
38768 (ARC_OPTX): Likewise.
38769 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38770 (TARGET_ARC600): Update access to arch information.
38771 (TARGET_ARC601): Likewise.
38772 (TARGET_ARC700): Likewise.
38773 (TARGET_EM): Likewise.
38774 (TARGET_HS): Likewise.
38775 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38778 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38781 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38782 condition/operands for integer GE/LE/GEU/LEU operations.
38784 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38786 PR translation/79397
38787 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38790 2017-02-08 Martin Jambor <mjambor@suse.cz>
38793 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38794 whether allocation happened.
38795 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38796 nothing was allocated.
38798 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38800 PR tree-optimization/79408
38801 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38802 constant, but SSA_NAME with a known integer range, use the minimum
38803 of that range instead of op1 to determine if modulo can be replaced
38804 with its first operand.
38806 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38808 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38810 2017-02-08 Richard Biener <rguenther@suse.de>
38812 PR tree-optimization/71824
38813 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38814 Check all loops contained in the merged region.
38816 2017-02-07 Andrew Pinski <apinski@cavium.com>
38818 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38820 2017-02-07 Andrew Pinski <apinski@cavium.com>
38822 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38823 (thunderxt88): Likewise.
38824 (thunderxt81): Disable LSE and change v8.1 to v8.
38825 (thunderxt83): Likewise.
38827 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38828 Richard Biener <rguenther@suse.de>
38830 PR middle-end/79399
38831 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38832 type from int to size_t.
38833 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38835 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38837 PR rtl-optimization/79386
38838 * cprop.c (bypass_conditional_jumps): Initialize
38839 bypass_last_basic_block already before splitting bbs after
38840 unconditional traps...
38841 (bypass_conditional_jumps): ... rather than here.
38844 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38845 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38846 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38847 fix -masm=intel patterns.
38849 2017-02-07 Richard Biener <rguenther@suse.de>
38851 PR tree-optimization/79256
38852 PR middle-end/79278
38853 * builtins.c (get_object_alignment_2): Use min_align_of_type
38854 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38855 and ADJUST_FIELD_ALIGN.
38857 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38859 * doc/tm.texi: Regenerate.
38860 * stor-layout.c (layout_decl): Adjust.
38861 (update_alignment_for_field): Likewise.
38862 (place_field): Likewise.
38863 (min_align_of_type): Likewise.
38864 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38865 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38866 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38867 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38868 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38869 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38870 * config/i386/i386.c (x86_field_alignment): Likewise.
38871 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38872 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38873 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38874 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38875 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38876 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38880 2017-01-30 Richard Biener <rguenther@suse.de>
38882 PR tree-optimization/79256
38883 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38884 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38887 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38889 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38890 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38891 builtins to SImode and emit a zero-extend, if necessary.
38893 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38895 * docs/invoke.texi (RISC-V Options): Alphabetize.
38897 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38899 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38902 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38904 * config/riscv/riscv.c: New file.
38905 * common/config/riscv/riscv-common.c: Likewise.
38906 * config.gcc: Likewise.
38907 * config/riscv/constraints.md: Likewise.
38908 * config/riscv/elf.h: Likewise.
38909 * config/riscv/generic.md: Likewise.
38910 * config/riscv/linux.h: Likewise.
38911 * config/riscv/multilib-generator: Likewise.
38912 * config/riscv/peephole.md: Likewise.
38913 * config/riscv/pic.md: Likewise.
38914 * config/riscv/predicates.md: Likewise.
38915 * config/riscv/riscv-builtins.c: Likewise.
38916 * config/riscv/riscv-c.c: Likewise.
38917 * config/riscv/riscv-ftypes.def: Likewise.
38918 * config/riscv/riscv-modes.def: Likewise.
38919 * config/riscv/riscv-opts.h: Likewise.
38920 * config/riscv/riscv-protos.h: Likewise.
38921 * config/riscv/riscv.h: Likewise.
38922 * config/riscv/riscv.md: Likewise.
38923 * config/riscv/riscv.opt: Likewise.
38924 * config/riscv/sync.md: Likewise.
38925 * config/riscv/t-elf-multilib: Likewise.
38926 * config/riscv/t-linux: Likewise.
38927 * config/riscv/t-linux-multilib: Likewise.
38928 * config/riscv/t-riscv: Likewise.
38929 * configure.ac: Likewise.
38930 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38931 Waterman as RISC-V maintainers.
38932 * doc/install.texi: Add RISC-V entries.
38933 * doc/invoke.texi: Add RISC-V options section.
38934 * doc/md.texi: Add RISC-V constraints section.
38935 * configure: Regenerated.
38937 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38940 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38941 false values to be constant vectors with all 0 or all 1 bits set.
38942 (vcondu<mode><mode>): Likewise.
38943 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38945 (fpmask_comparison_operator): Update comment.
38946 (vecint_comparison_operator): New predicate.
38947 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38948 vector conditionals when the true and false values are constant
38949 vectors with all 0 bits or all 1 bits set.
38951 2017-02-06 Martin Sebor <msebor@redhat.com>
38953 PR tree-optimization/79376
38954 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38956 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38958 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38959 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38960 to simplify split condition.
38962 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38964 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38965 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38968 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38970 PR rtl-optimization/68664
38971 * target.def (can_speculate_insn): New hook.
38972 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38973 * doc/tm.texi: Regenerate.
38974 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38975 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38976 (rs6000_sched_can_speculate_insn): New function.
38978 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38980 PR tree-optimization/79284
38981 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38982 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38983 vectorizable_mask_load_store, vectorizable_operation,
38984 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38985 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38986 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38987 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38988 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38989 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38990 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38991 is_gimple_assign (stmt). Replace another such test with
38992 is_gimple_assign (stmt).
38994 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38997 * config/avr/avr.c (rtl-iter.h): Include it.
38998 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38999 (avr_legitimate_combined_insn): ...and implementation.
39001 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
39003 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
39004 * config/s390/s390.c (s390_const_operand_ok)
39005 (s390_canonicalize_comparison, s390_extract_part)
39006 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
39007 (s390_contiguous_bitmask_p, s390_rtx_costs)
39008 (legitimize_pic_address): Likewise.
39009 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
39010 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
39011 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
39012 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
39013 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
39015 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39017 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
39018 REGNO($0) == REGNO($1).
39020 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39022 * config/s390/linux.h(SIZE_TYPE): Add comment.
39024 2017-02-06 Julian Brown <julian@codesourcery.com>
39025 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39026 Virendra Pathak <virendra.pathak@broadcom.com>
39028 * config/aarch64/aarch64-cores.def: Change the scheduler
39030 * config/aarch64/aarch64.md: Include thunderx2t99.md.
39031 * config/aarch64/thunderx2t99.md: New file.
39033 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39035 * doc/standards.texi (Go Language): Update link to language
39038 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39040 * tree-eh.c (lower_resx): Sanitize profile.
39041 (cleanup_empty_eh_move_lp): Likewise.
39043 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39046 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
39048 * cfgloopmanip.h (loop_version): Update prototype.
39049 * modulo-sched.c (sms_schedule): Update call of loop_version.
39050 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
39051 * tree-parloops.c (gen_parallel_loop): Likewise.
39052 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
39053 * tree-ssa-loop-split.c (split_loop): Likewise.
39054 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
39055 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
39057 2017-02-05 Martin Liska <mliska@suse.cz>
39060 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39062 (print_operand_address): Initialize a struct to zero.
39064 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39066 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39067 garbage collector only in textual form.
39069 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39071 * doc/extend.texi (x86 specific memory model extensions for
39072 transactional memory): Simplify a phrase.
39074 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39077 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39078 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39079 (atomic_storedi_1): Likewise.
39081 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39083 PR tree-optimization/79338
39084 * tree-parloops.c (gather_scalar_reductions): Don't call
39085 vect_analyze_loop_form for loop->inner before destroying loop's
39088 2017-02-03 Martin Sebor <msebor@redhat.com>
39090 PR tree-optimization/79327
39091 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39092 when precision has resulted in leading zeros.
39093 (format_integer): Adjust the likely counter to assume an unknown
39094 argument that may be zero is non-zero.
39096 2017-02-03 Jason Merrill <jason@redhat.com>
39099 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39100 avoid copying non-taken branch.
39102 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39104 PR tree-optimization/79340
39105 * tree-vect-loop.c (vectorizable_reduction): Release
39106 vec_defs elements after safe_splicing them into other vectors.
39109 PR tree-optimization/79327
39110 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39111 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39113 (format_integer): Use wide_int_to_tree instead of build_int_cst
39114 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39115 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39116 of shortest and longest sequence.
39118 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39120 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39121 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39123 2017-02-03 Walter Lee <walt@tilera.com>
39126 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39127 after initial stackframe link reg save.
39128 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39130 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39133 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39134 wu for stxssp alternative.
39136 2017-02-03 Martin Sebor <msebor@redhat.com>
39138 PR tree-optimization/79352
39139 * gimple-fold.c (get_range_strlen): Add argument.
39140 (get_range_strlen): Change return type to bool.
39141 (get_maxval_strlen): Pass in a dummy argument.
39142 * gimple-fold.h (get_range_strlen): Change return type to bool.
39143 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39144 * tree.h (array_at_struct_end_p): Add argument.
39145 * tree.c (array_at_struct_end_p): Handle it.
39147 2017-02-03 Martin Liska <mliska@suse.cz>
39150 * multiple_target.c (create_dispatcher_calls): Redirect edge
39151 from a caller of a dispatcher.
39152 (expand_target_clones): Make the clones local.
39153 (ipa_target_clone): Do both target clones and resolvers.
39154 (ipa_dispatcher_calls): Remove the pass.
39155 (pass_dispatcher_calls::gate): Likewise.
39156 (make_pass_dispatcher_calls): Likewise.
39157 * passes.def (pass_target_clone): Put as very first IPA early
39160 2017-02-03 Martin Liska <mliska@suse.cz>
39162 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39163 in case of a function with ifunc attribute.
39165 2017-02-03 Martin Liska <mliska@suse.cz>
39167 * cgraph.c (cgraph_node::dump): Dump function version info.
39168 * symtab.c (symtab_node::dump_base): Add missing new line.
39170 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39172 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39173 (ifcombine_ifandif): Use it.
39175 2017-02-03 Martin Liska <mliska@suse.cz>
39177 * doc/invoke.texi: Document default value for
39178 use-after-scope-direct-emission-threshold.
39180 2017-02-03 Martin Liska <mliska@suse.cz>
39182 PR tree-optimization/79339
39183 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39184 (format_floating): Likewise.
39186 2017-02-03 Martin Liska <mliska@suse.cz>
39189 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39191 (ipa_node_params_t::remove): Likewise.
39192 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39193 initialization from removed ipa_node_params_t::insert.
39194 (ipa_node_params::~ipa_node_params): Move from removed
39195 ipa_node_params_t::release.
39196 * symbol-summary.h (symbol_summary::m_released): New member.
39197 Do not release a summary twice. Do not allow to call finalizer
39198 for types of a summary that live in GGC memory.
39200 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39202 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39205 2017-02-02 Martin Sebor <msebor@redhat.com>
39207 PR middle-end/79275
39208 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39209 (format_string): Tighten up the range of output for non-constant
39210 strings and correct the expected range for wide non-constant strings.
39212 2017-02-02 Martin Sebor <msebor@redhat.com>
39214 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39216 PR middle-end/32003
39217 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39219 (-fdump-tree-@var): Add to index and document how to come up
39220 with pass-specific option and dump file names.
39221 (-fdump-passes): Clarify where to look for output.
39223 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39225 PR middle-end/77445
39226 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39227 statistics of the analyzed path; allow threading for speed when
39228 any of BBs along the path are optimized for speed.
39230 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39232 PR middle-end/78468
39233 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39234 settings of the virtual registers.
39237 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39239 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39240 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39243 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39245 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39246 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39248 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39250 * config/s390/s390.md: Add missing comments with the expanded
39252 * config/s390/vector.md: Likewise.
39253 * config/s390/vx-builtins.md: Likewise.
39255 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39258 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39259 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39260 conditions on a single line.
39262 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39264 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39265 __S390_VX__ to __VX__.
39267 2017-02-01 Andrew Pinski <apinski@cavium.com>
39269 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39270 stmt_info to record_stmt_cost.
39271 (vect_get_known_peeling_cost): Pass stmt_info if known to
39273 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39274 cpu_vector_cost field into
39275 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39276 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39277 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39278 splitting of scalar_stmt_cost and vec_stmt_cost.
39279 (thunderx_vector_cost): Likewise.
39280 (cortexa57_vector_cost): LIkewise.
39281 (exynosm1_vector_cost): Likewise.
39282 (xgene1_vector_cost): Likewise.
39283 (thunderx2t99_vector_cost): Improve after the splitting of the two
39285 (aarch64_builtin_vectorization_cost): Update for the splitting of
39286 scalar_stmt_cost and vec_stmt_cost.
39288 2017-02-01 Torvald Riegel <triegel@redhat.com>
39289 Richard Henderson <rth@redhat.com>
39291 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39292 conditional on existance of a fast atomic load.
39293 * optabs-query.c (can_atomic_load_p): New function.
39294 * optabs-query.h (can_atomic_load_p): Declare it.
39295 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39296 no fast atomic load is available for the particular size of access.
39297 (expand_atomic_compare_and_swap): Likewise.
39298 (expand_atomic_load): Likewise.
39299 (expand_atomic_store): Likewise.
39300 (expand_atomic_fetch_op): Likewise.
39301 * testsuite/lib/target-supports.exp
39302 (check_effective_target_sync_int_128): Remove x86 because it provides
39303 no fast atomic load.
39304 (check_effective_target_sync_int_128_runtime): Likewise.
39306 2017-02-01 Richard Biener <rguenther@suse.de>
39308 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39309 (graphite_transform_loops): Provide opt-info for optimized nests.
39310 * tree-parloop.c (parallelize_loops): Provide opt-info for
39311 parallelized loops.
39313 2017-02-01 Richard Biener <rguenther@suse.de>
39315 PR middle-end/79315
39316 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39317 was not set before.
39319 2017-02-01 Richard Biener <rguenther@suse.de>
39321 PR tree-optimization/71824
39322 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39323 Verify the loops are valid in the merged SESE region.
39324 (scop_detection::can_represent_loop_1): Check analyzing the
39325 evolution of the number of iterations in the region succeeds.
39327 2017-01-31 Ian Lance Taylor <iant@golang.org>
39329 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39330 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39332 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39334 PR preprocessor/79210
39335 * input.c (get_substring_ranges_for_loc): Replace line_width
39336 assertion with error-handling.
39338 2017-01-31 Richard Biener <rguenther@suse.de>
39340 PR tree-optimization/77318
39341 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39342 (create_pw_aff_from_tree): Take loop parameter.
39343 (add_condition_to_pbb): Pass loop of the condition to
39344 create_pw_aff_from_tree.
39346 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39348 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39349 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39351 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39355 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39357 (convert_int_to_float128): Likewise.
39358 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39359 (convert_int_to_float128): Likewise.
39360 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39361 (UNSPEC_IEEE128_CONVERT): Likewise.
39362 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39363 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39364 Use local variables for IBM extended format.
39365 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39366 (fix_trunc<mode>si2_fprs): Likewise.
39367 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39368 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39369 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39370 to know that we can now have integers of all sizes in vector
39372 (fix<uns>_<mode>di2_hw): Likewise.
39373 (float<uns>_<mode>si2_hw): Likewise.
39374 (fix_<mode>si2_hw): Likewise.
39375 (fixuns_<mode>si2_hw): Likewise.
39376 (float<uns>_<mode>di2_hw): Likewise.
39377 (float_<mode>di2_hw): Likewise.
39378 (float_<mode>si2_hw): Likewise.
39379 (floatuns_<mode>di2_hw): Likewise.
39380 (floatuns_<mode>si2_hw): Likewise.
39381 (xscvqp<su>wz_<mode>): Delete, no longer used.
39382 (xscvqp<su>dz_<mode>): Likewise.
39383 (xscv<su>dqp_<mode>): Likewise.
39384 (ieee128_mfvsrd_64bit): Likewise.
39385 (ieee128_mfvsrd_32bit): Likewise.
39386 (ieee128_mfvsrwz): Likewise.
39387 (ieee128_mtvsrw): Likewise.
39388 (ieee128_mtvsrd_64bit): Likewise.
39389 (ieee128_mtvsrd_32bit): Likewise.
39391 2017-01-31 Martin Liska <mliska@suse.cz>
39394 * ipa-prop.c (ipa_free_all_node_params): Call release method
39395 instead of ~sumbol_summary to not to trigger double times
39398 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39400 PR tree-optimization/71691
39401 * bitmap.h (class auto_bitmap): New.
39402 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39403 is_maybe_undefined instead of ssa_undefined_value_p.
39405 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39407 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39408 __S390_ARCH_LEVEL__ to __ARCH__.
39410 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39412 PR tree-optimization/79267
39413 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39414 if should_remove_lhs_p is true.
39416 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39419 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39420 (add_alignment_attribute): New.
39421 (base_type_die): Add alignment attribute.
39422 (subrange_type_die): Likewise.
39423 (modified_type_die): Likewise.
39424 (gen_array_type_die): Likewise.
39425 (gen_descr_array_type_die: Likewise.
39426 (gen_enumeration_type_die): Likewise.
39427 (gen_subprogram_die): Likewise.
39428 (gen_variable_die): Likewise.
39429 (gen_field_die): Likewise.
39430 (gen_ptr_to_mbr_type_die): Likewise.
39431 (gen_struct_or_union_type_die): Likewise.
39432 (gen_subroutine_type_die): Likewise.
39433 (gen_typedef_die): Likewise.
39434 (base_type_cmp): Compare alignment attribute.
39436 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39439 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39440 (setb_unsigned) New pattern for setb with CCUNS.
39441 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39442 subfc./subfe sequence to avoid overflow problems. Generate a
39443 shorter sequence with cmpld/setb for power9.
39444 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39445 for generating subfc. instruction.
39446 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39447 now uses this instruction.
39449 2017-01-30 Ian Lance Taylor <iant@google.com>
39452 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39453 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39455 2017-01-30 Martin Sebor <msebor@redhat.com>
39457 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39458 Move constant to the right of a relational operator.
39459 (get_mpfr_format_length, format_character, format_string): Ditto.
39460 (should_warn_p, maybe_warn): Same.
39462 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39464 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39467 * asan.c (get_translation_unit_decl): Remove function.
39468 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39470 2017-01-30 Martin Liska <mliska@suse.cz>
39472 PR gcov-profile/79259
39473 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39474 -fprofile-generate.
39476 2017-01-30 Martin Liska <mliska@suse.cz>
39479 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39480 Initialize variables with NULL value.
39482 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39485 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39487 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39489 2017-01-30 Richard Biener <rguenther@suse.de>
39491 PR tree-optimization/79276
39492 * tree-vrp.c (process_assert_insertions): Properly adjust common
39493 when removing a duplicate.
39495 2017-01-30 Richard Biener <rguenther@suse.de>
39497 PR tree-optimization/79256
39498 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39499 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39501 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39503 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39506 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39507 ("*r<noxa>sbg_<mode>_sll_bitmask")
39508 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39509 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39510 Use contiguous_bitmask_nowrap_operand.
39512 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39515 * config/rs6000/altivec.h (vec_xl): Revise #define.
39516 (vec_xst): Likewise.
39518 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39520 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39522 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39524 PR rtl-optimization/79194
39525 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39526 traps before call to bypass_conditional_jumps.
39528 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39530 PR tree-optimization/71374
39531 * lra-constraints.c (check_conflict_input_operands): New.
39532 (match_reload): Use it.
39534 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39537 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39538 account to calculate conflict_set.
39540 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39542 PR rtl-optimization/78559
39543 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39544 other_insn in combine.
39546 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39548 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39549 uint16_type_node for BT_UINT16.
39551 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39553 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39554 "RTL Tests" to menu.
39555 (GIMPLE Tests): New node.
39556 (RTL Tests): New node.
39558 2017-01-27 Richard Biener <rguenther@suse.de>
39560 PR tree-optimization/79245
39561 * tree-loop-distribution.c (distribute_loop): Apply cost
39562 modeling also to detected patterns.
39564 2017-01-27 Richard Biener <rguenther@suse.de>
39566 PR tree-optimization/71433
39567 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39568 (compare_assert_loc): New function.
39569 (process_assert_insertions): Sort and optimize assert locations
39570 to remove duplicates and push down identical assertions on
39571 edges to their destination block.
39573 2017-01-27 Richard Biener <rguenther@suse.de>
39575 PR tree-optimization/79244
39576 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39577 out SSA names even if abnormal.
39579 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39581 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39582 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39583 instead of MPFR_RNDN.
39585 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39588 * arm.c (arm_option_override): Don't call build_target_option_node
39589 until after doing all option overrides.
39590 (arm_valid_target_attribute_tree): Likewise.
39592 2017-01-27 Martin Liska <mliska@suse.cz>
39594 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39595 for {cd}tors and C++ {cd}tors.
39597 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39599 * config/s390/s390.md ("*setmem_long_and")
39600 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39602 2017-01-26 Martin Sebor <msebor@redhat.com>
39604 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39607 2017-01-26 Martin Sebor <msebor@redhat.com>
39609 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39610 HAVE_DFmode before using XFmode or DFmode.
39611 (parse_directive): Avoid using the z length modifier to avoid
39612 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39614 PR middle-end/78703
39615 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39616 to accept adjustment as an array.
39617 (get_int_range): New function.
39618 (struct directive): Make width and prec arrays.
39619 (directive::set_width, directive::set_precision): Call get_int_range.
39620 (format_integer, format_floating): Handle width and precision ranges.
39621 (format_string, parse_directive): Same.
39623 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39626 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39627 comdat_type_p set, just clone them, but keep the children in the
39631 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39632 which have direct callers with -fvar-tracking-assignments enabled
39634 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39635 inside of type units.
39637 2017-01-26 Martin Sebor <msebor@redhat.com>
39639 PR middle-end/78703
39640 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39642 (struct format_result): Replace number_chars, number_chars_min,
39643 and number_chars_max with a single member of struct result_range.
39645 (format_result::operator+=): Adjust.
39646 (struct fmtresult): Remove bounded. Handle likely and unlikely
39648 (fmtresult::adjust_for_width_or_precision): New function.
39649 (fmtresult:type_max_digits): New function.
39650 (bytes_remaining): Handle likely and unlikely counters.
39651 (min_bytes_remaining): Remove.
39652 (format_percent): Simplify.
39653 (format_integer, format_floating): Set likely and unlikely counters.
39654 (get_string_length, format_character, format_string): Same.
39655 (format_plain, should_warn_p): New function.
39656 (maybe_warn): Call should_warn_p. Update diagnostic messages
39657 and handle those for all directives, including plain strings.
39658 (format_directive): Handle likely and unlikely counters.
39659 Remove unnecessary quoting from diagnostics. Add an informational
39661 (add_bytes): Remove.
39662 (pass_sprintf_length::compute_format_length): Simplify.
39663 (try_substitute_return_value): Handle likely and unlikely counters.
39665 2017-01-26 Carl Love <cel@us.ibm.com>
39667 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39668 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39670 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39673 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39674 endianess for subregs into account.
39675 * lra-constraints.c (lra_constraints): Do risky transformations
39676 always on the first iteration.
39677 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39678 last_call_used_reg_set.
39679 (process_bb_lives): Define and use last_call_used_reg_set.
39680 * lra.c (lra): Always continue after lra_constraints on the first
39683 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39685 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39687 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39689 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39691 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39692 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39693 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39694 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39695 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39696 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39697 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39698 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39699 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39701 2017-01-26 Marek Polacek <polacek@redhat.com>
39704 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39705 for the third operand.
39707 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39709 PR middle-end/79236
39710 * omp-low.c (struct omp_context): Add simt_stmt field.
39711 (scan_omp_for): Return omp_context *.
39712 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39713 context to the _simt_ SIMD stmt.
39714 (lower_omp_for): For combined SIMD with sibling _simt_
39715 SIMD, make sure to use the same decls in _looptemp_
39716 clauses as in the sibling.
39718 2017-01-26 David Sherwood <david.sherwood@arm.com>
39720 PR middle-end/79212
39721 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39724 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39727 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39728 emit fld b; fld a; if possible.
39730 * brig-builtins.def: Update copyright years.
39731 * config/arm/arm_acle_builtins.def: Update copyright years.
39733 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39736 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39737 constraint instead of o for the stxsd instruction.
39739 2017-01-25 Carl Love <cel@us.ibm.com>
39741 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39742 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39744 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39746 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39748 2017-01-25 Richard Biener <rguenther@suse.de>
39750 PR tree-optimization/69264
39751 * target.def (vector_alignment_reachable): Improve documentation.
39752 * doc/tm.texi: Regenerate.
39753 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39755 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39756 earlier changes with respect to TYPE_USER_ALIGN.
39757 (vector_alignment_reachable_p): Likewise. Improve dumping.
39759 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39762 * config/arm/arm.md (xordi3): Force constant operand into a register
39765 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39767 * doc/invoke.texi (-fstore-merging): Correct default optimization
39768 levels at which it is enabled.
39769 (-O): Move -fstore-merging from list to...
39772 2017-01-25 Richard Biener <rguenther@suse.de>
39775 * omp-expand.c: Include debug.h.
39776 (expand_omp_taskreg): Make sure to generate early debug before
39777 outlining anything from a function.
39778 (expand_omp_target): Likewise.
39779 (grid_expand_target_grid_body): Likewise.
39781 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39784 * asan.c (get_translation_unit_decl): New function.
39785 (asan_add_global): Extract modules file name from globals
39786 TRANSLATION_UNIT_DECL name.
39788 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39791 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39792 for long calls with APCS frame and VFP.
39794 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39796 * cfg.c (original_copy_tables_initialized_p): New function.
39797 * cfg.h (original_copy_tables_initialized_p): New decl.
39798 * cfgrtl.c (relink_block_chain): Guard the call to
39799 free_original_copy_tables with a call to
39800 original_copy_tables_initialized_p.
39801 * cgraph.h (symtab_node::native_rtl_p): New decl.
39802 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39803 (symtab_node::needed_p): Don't assert for early assembly output
39804 for __RTL functions.
39805 (cgraph_node::finalize_function): Set "force_output" for __RTL
39807 (cgraph_node::analyze): Bail out early for __RTL functions.
39808 (analyze_functions): Update assertion to support __RTL functions.
39809 (cgraph_node::expand): Bail out early for __RTL functions.
39810 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39812 * function.h (struct function): Update comment for field
39814 * gimple-expr.c: Include "tree-pass.h".
39815 (gimple_has_body_p): Return false for __RTL functions.
39816 * Makefile.in (OBJS): Add run-rtl-passes.o.
39817 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39819 (gcc::pass_manager::get_clean_slate): New accessor.
39820 * passes.c: Include "insn-addr.h".
39821 (should_skip_pass_p): Add logging. Update logic for running
39822 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39823 property-provider override so it is only done for gimple passes.
39825 (skip_pass): New function.
39826 (execute_one_pass): Call skip_pass when skipping passes.
39827 * read-md.c (md_reader::read_char): Support filtering
39828 the input to a subset of line numbers.
39829 (md_reader::md_reader): Initialize fields
39830 m_first_line and m_last_line.
39831 (md_reader::read_file_fragment): New function.
39832 * read-md.h (md_reader::read_file_fragment): New decl.
39833 (md_reader::m_first_line): New field.
39834 (md_reader::m_last_line): New field.
39835 * read-rtl-function.c (function_reader::create_function): Only
39836 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39837 curr_properties. Set DECL_INITIAL to a dummy block.
39838 (read_rtl_function_body_from_file_range): New function.
39839 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39841 * run-rtl-passes.c: New file.
39842 * run-rtl-passes.h: New file.
39844 2017-01-24 Jeff Law <law@redhat.com>
39846 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39849 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39851 PR tree-optimization/79159
39852 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39853 (record_nonwrapping_iv): Improve boundary using above function if no
39854 value range information.
39856 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39857 Martin Jambor <mjambor@suse.cz>
39859 * brig-builtins.def: New file.
39860 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39861 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39862 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39863 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39864 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39865 * builtin-types.def (BT_INT8): New.
39866 (BT_INT16): Likewise.
39867 (BT_UINT8): Likewise.
39868 (BT_UINT16): Likewise.
39869 (BT_FN_ULONG): Likewise.
39870 (BT_FN_UINT_INT): Likewise.
39871 (BT_FN_UINT_ULONG): Likewise.
39872 (BT_FN_UINT_LONG): Likewise.
39873 (BT_FN_UINT_PTR): Likewise.
39874 (BT_FN_ULONG_PTR): Likewise.
39875 (BT_FN_INT8_FLOAT): Likewise.
39876 (BT_FN_INT16_FLOAT): Likewise.
39877 (BT_FN_UINT32_FLOAT): Likewise.
39878 (BT_FN_UINT16_FLOAT): Likewise.
39879 (BT_FN_UINT8_FLOAT): Likewise.
39880 (BT_FN_UINT64_FLOAT): Likewise.
39881 (BT_FN_UINT16_UINT32): Likewise.
39882 (BT_FN_UINT32_UINT16): Likewise.
39883 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39884 (BT_FN_INT_PTR_INT): Likewise.
39885 (BT_FN_UINT_PTR_UINT): Likewise.
39886 (BT_FN_LONG_PTR_LONG): Likewise.
39887 (BT_FN_ULONG_PTR_ULONG): Likewise.
39888 (BT_FN_VOID_UINT64_UINT64): Likewise.
39889 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39890 (BT_FN_INT8_INT8_INT8): Likewise.
39891 (BT_FN_INT16_INT16_INT16): Likewise.
39892 (BT_FN_INT_INT_INT): Likewise.
39893 (BT_FN_UINT_FLOAT_UINT): Likewise.
39894 (BT_FN_FLOAT_UINT_UINT): Likewise.
39895 (BT_FN_ULONG_UINT_UINT): Likewise.
39896 (BT_FN_ULONG_UINT_PTR): Likewise.
39897 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39898 (BT_FN_UINT_UINT_UINT): Likewise.
39899 (BT_FN_VOID_UINT_PTR): Likewise.
39900 (BT_FN_UINT_UINT_PTR: Likewise.
39901 (BT_FN_UINT32_UINT64_PTR): Likewise.
39902 (BT_FN_INT_INT_UINT_UINT): Likewise.
39903 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39904 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39905 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39906 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39907 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39908 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39909 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39910 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39911 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39912 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39913 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39914 * doc/frontends.texi: List BRIG FE.
39915 * doc/install.texi (Testing): Add BRIG tesring requirements.
39916 * doc/invoke.texi (Overall Options): Mention BRIG.
39917 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39919 2017-01-24 Richard Biener <rguenther@suse.de>
39921 PR translation/79208
39922 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39924 2017-01-24 Martin Jambor <mjambor@suse.cz>
39927 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39928 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39929 and known_contexts.
39931 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39933 PR middle-end/79123
39934 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39935 casts from signed to unsigned really don't have a range.
39937 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39939 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39940 GMP_RNDx for compatiblity.
39942 2017-01-24 Martin Liska <mliska@suse.cz>
39945 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39946 that would prevent us to call alloca with -1 as argument.
39948 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39950 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39951 Avoid -Wformat-security warning.
39953 2017-01-23 Andrew Pinski <apinski@cavium.com>
39955 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39958 2017-01-23 Martin Sebor <msebor@redhat.com>
39960 PR middle-end/78703
39961 * gimple-ssa-sprintf.c (warn_level): New global.
39962 (format_integer): Use it here and throughout the rest of the file.
39963 Use the same switch to compute sign as base.
39964 (maybe_warn): New function.
39965 (format_directive): Factor out warnings into maybe_warn.
39966 Add debugging output. Use warn_level.
39967 (add_bytes): Use warn_level.
39968 (pass_sprintf_length::compute_format_length): Add debugging output.
39969 (try_substitute_return_value): Same.
39970 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39972 PR middle-end/78703
39973 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39974 (struct fmtresult, format_integer, format_floating): Adjust.
39975 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39976 (get_string_length, format_string,format_directive): Same.
39977 (pass_sprintf_length::compute_format_length): Same.
39978 (try_substitute_return_value): Simplify slightly.
39980 PR middle-end/78703
39981 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39982 (fmtresult::operator+=): Outlined.
39983 (struct fmtresult): Add ctors.
39984 (struct conversion_spec): Rename...
39985 (struct directive): ...to this. Add and remove data members.
39986 (directive::set_width, directive::set_precision): New functions.
39987 (format_percent): Use fmtresult ctor.
39988 (get_width_and_precision): Remove.
39989 (format_integer): Make naming changes. Avoid computing width and
39991 (format_floating): Same. Adjust indentation.
39992 (format_character, format_none): New functions.
39993 (format_string): Moved character handling to format_character.
39994 (format_directive): Remove arguments, change return type.
39995 (parse_directive): New function.
39996 (pass_sprintf_length::compute_format_length): Move directive
39997 parsing to parse_directive.
39999 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40001 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
40002 (assign_assembler_name_if_needed): ... this.
40003 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
40004 (assign_assembler_name_if_needed): ... this.
40005 (free_lang_data_in_cgraph): Adjust callers.
40006 * cgraphunit.c (cgraph_node::analyze): Likewise.
40007 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
40010 2017-01-23 Richard Biener <rguenther@suse.de>
40012 PR tree-optimization/79088
40013 PR tree-optimization/79188
40014 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
40015 resetting loop bounds after last path deletion. Reset loop
40016 bounds of the target loop, make code match the comments.
40017 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
40018 Make sure loops need no fixups.
40020 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
40022 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
40023 exponent support with double type for first argument.
40024 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
40025 type returned by __builtin_vec_extract_sig,
40026 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
40027 functions from "vector int" to "vector unsigned int" or from
40028 "vector long long int" to "vector unsigned long long int".
40029 Changed type returned by __builtin_vec_extract_exp,
40030 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
40031 functions from "vector int" to "vector unsigned int" or from
40032 "vector long long int" to "vector unsigned long long int".
40033 Changed return type of __builtin_vec_test_data_class,
40034 __builtin_vec_test_data_class_sp, and
40035 __builtin_vec_test_data_class_dp from "vector int" to
40036 "vector bool int" or from "vector long long int" to "vector bool
40037 long long int" and changed second argument type from "unsigned
40038 int" to "int". Added new overloaded function forms "vector float
40039 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
40040 "vector float __builtin_vec_insert_exp_sp (vector float, vector
40041 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
40042 double, vector unsigned long long int)" and "vector double
40043 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
40044 long int)". Changed return type of
40045 __builtin_scalar_test_data_class and
40046 __builtin_scalar_test_data_class_sp and
40047 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
40048 int" and changed second argument from "unsigned int" to "int".
40049 Changed type returned by __builtin_scalar_test_neg,
40050 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
40051 from "int" to "bool int". Added new overloaded function form
40052 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
40053 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
40054 exponent double-precision with floating point first argument.
40055 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
40056 documentation of scalar_test_data_class, scalar_test_neg,
40057 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
40058 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
40059 vec_test_data_class built-in functions to reflect refinements in
40060 their type signatures.
40062 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40064 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40066 (aarch64_elf_asm_destructor): Likewise.
40068 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40070 PR rtl-optimization/78634
40071 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40072 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40073 * ifcvt.c (noce_try_cmove): Add missing cost check.
40075 PR rtl-optimization/71724
40076 * combine.c (if_then_else_cond): Look for situations where it is
40077 beneficial to undo the work of one of the recursive calls.
40079 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40081 PR tree-optimization/70754
40082 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40083 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40084 combined stmt before it if not NULL.
40085 (combine_chains): Process refs reversely and compute dominance point
40088 2017-01-23 Martin Liska <mliska@suse.cz>
40090 PR tree-optimization/79196
40091 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40092 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40094 (strlen_optimize_stmt): Call the renamed function.
40096 2017-01-23 Michael Matz <matz@suse.de>
40098 PR tree-optimization/78384
40099 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40101 2017-01-23 Richard Biener <rguenther@suse.de>
40103 PR tree-optimization/79186
40104 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40105 both incoming edges before moving an assert.
40107 2017-01-23 Martin Jambor <mjambor@suse.cz>
40109 * ipa-prop.c (load_from_param_1): Removed.
40110 (load_from_unmodified_param): Bits from load_from_param_1 put back
40112 (load_from_param): Removed.
40113 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40114 with stmt. Reverted back to use of load_from_unmodified_param.
40116 2017-01-23 Martin Jambor <mjambor@suse.cz>
40119 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40120 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40121 field a pointer to garbage collected vector, mark lattices and
40122 ipcp_orig_node with GTY((skip)).
40123 (ipa_get_param_count): Adjust to descriptors being a pointer.
40124 (ipa_get_param): Likewise.
40125 (ipa_get_type): Likewise.
40126 (ipa_get_param_move_cost): Likewise.
40127 (ipa_set_param_used): Likewise.
40128 (ipa_get_controlled_uses): Likewise.
40129 (ipa_set_controlled_uses): Likewise.
40130 (ipa_is_param_used): Likewise.
40131 (ipa_node_params_t): Move into garbage collector. New methods insert
40133 (ipa_node_params_sum): Annotate wth GTY(()).
40134 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40136 (ipa_load_from_parm_agg): Adjust declaration.
40137 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40138 * ipa-profile.c (ipa_profile): Likewise.
40139 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40140 (ipa_populate_param_decls): Make descriptors parameter garbage
40142 (ipa_dump_param): Adjust to descriptors being a pointer.
40143 (ipa_alloc_node_params): Likewise.
40144 (ipa_initialize_node_params): Likewise.
40145 (load_from_param_1): Make descriptors parameter garbage collected.
40146 (load_from_unmodified_param): Likewise.
40147 (load_from_param): Likewise.
40148 (ipa_load_from_parm_agg): Likewise.
40149 (ipa_node_params::~ipa_node_params): Removed.
40150 (ipa_free_all_node_params): Remove call to delete operator.
40151 (ipa_node_params_t::insert): New.
40152 (ipa_node_params_t::remove): Likewise.
40153 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40154 copy known_csts and known_contexts vectors.
40155 (ipa_read_node_info): Adjust to descriptors being a pointer.
40156 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40158 (ipcp_transform_function): Make descriptors variable garbage
40161 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40163 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40164 * config/i386/avx512dqintrin.h: Ditto.
40165 * config/i386/avx512fintrin.h: Ditto.
40166 * config/i386/i386.c: Handle new builtins.
40167 * config/i386/i386-builtin.def: Add new builtins.
40168 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40169 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40171 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40172 Martin Liska <mliska@suse.cz>
40174 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40175 * asan.c (asan_expand_poison_ifn): Support stores and use
40176 appropriate ASAN report function.
40177 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40178 * internal-fn.def (ASAN_POISON_USE): Declare.
40179 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40180 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40181 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40182 ASAN_POISON calls w/o LHS.
40183 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40184 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40185 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40186 * gimplify.c (asan_poison_variables): Add attribute
40187 use_after_scope_memory to variables that really needs to live
40189 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40190 having the attribute.
40192 2017-01-23 Martin Liska <mliska@suse.cz>
40194 * asan.c (create_asan_shadow_var): New function.
40195 (asan_expand_poison_ifn): Likewise.
40196 * asan.h (asan_expand_poison_ifn): New declaration.
40197 * internal-fn.c (expand_ASAN_POISON): Likewise.
40198 * internal-fn.def (ASAN_POISON): New builtin.
40199 * sanopt.c (pass_sanopt::execute): Expand
40200 asan_expand_poison_ifn.
40201 * tree-inline.c (copy_decl_for_dup_finish): Make function
40203 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40204 * tree-ssa.c (is_asan_mark_p): New function.
40205 (execute_update_addresses_taken): Rewrite local variables
40206 (identified just by use-after-scope as addressable) into SSA.
40208 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40210 * doc/install.texi (Specific): opensource.apple.com uses https
40211 now. Remove trailing slash.
40213 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40215 * README.Portability: Remove note on an Irix compatibility issue.
40217 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40219 * gcov.c (INCLUDE_ALGORITHM): Define.
40220 (INCLUDE_VECTOR): Define.
40221 No longer include <vector> and <algorithm> directly.
40223 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40225 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40227 * doc/invoke.texi (Code Gen Options): Ditto.
40229 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40232 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40234 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40236 rtl-optimization/79125
40237 * cprop.c (local_cprop_pass): Handle cases where we make an
40238 unconditional trap.
40240 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40244 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40245 read from, for big endian.
40247 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40249 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40250 register pauth builtins for LP64 only.
40252 2017-01-20 Marek Polacek <polacek@redhat.com>
40255 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40258 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40260 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40262 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40263 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40264 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40266 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40269 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40270 in big-endian mode when they are not a single duplicated value.
40272 2017-01-20 Richard Biener <rguenther@suse.de>
40274 * BASE-VER: Bump to 7.0.1.
40276 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40278 * omp-low.c (omplow_simd_context): New struct. Use it...
40279 (lower_rec_simd_input_clauses): ...here and...
40280 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40281 references to idx, lane, max_vf, is_simt.
40283 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40285 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40288 2017-01-20 Martin Jambor <mjambor@suse.cz>
40290 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40291 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40293 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40294 (GTFILES): Rename hsa.c to hsa-common.c.
40295 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40296 * hsa-dump.c: Likewise.
40297 * hsa-gen.c: Likewise.
40298 * hsa-regalloc.c: Likewise.
40299 * ipa-hsa.c: Likewise.
40300 * omp-expand.c: Likewise.
40301 * omp-low.c: Likewise.
40302 * toplev.c: Likewise.
40304 2017-01-20 Marek Polacek <polacek@redhat.com>
40307 * doc/invoke.texi: Document -Wduplicated-branches.
40308 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40309 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40310 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40311 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40312 return 0 only when not OEP_LEXICOGRAPHIC.
40313 (fold_build_cleanup_point_expr): Use the expression
40314 location when building CLEANUP_POINT_EXPR.
40315 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40316 * tree.c (add_expr): Handle error_mark_node.
40318 2017-01-20 Martin Liska <mliska@suse.cz>
40321 * tree-profile.c (init_ic_make_global_vars): Do not call
40323 (gimple_init_gcov_profiler): Likewise.
40325 2017-01-20 Martin Liska <mliska@suse.cz>
40328 * cgraph.h (maybe_create_reference): Remove argument and
40330 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40332 * ipa-cp.c (create_specialized_node): Likewise.
40333 * symtab.c (symtab_node::maybe_create_reference): Handle
40334 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40336 2017-01-20 Martin Liska <mliska@suse.cz>
40338 * read-rtl-function.c (function_reader::create_function): Use
40339 build_decl instread of build_decl_stat.
40341 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40343 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40344 * config/i386/avx512dqintrin.h: Ditto.
40345 * config/i386/avx512fintrin.h: Ditto.
40346 * config/i386/i386-builtin-types.def: Add new types.
40347 * config/i386/i386.c: Handle new types.
40348 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40349 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40350 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40351 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40352 (__builtin_ia32_kshiftridi): New.
40353 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40355 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40359 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40360 define to rs6000_init_stack_protect_guard.
40361 (rs6000_init_stack_protect_guard): New function.
40363 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40364 Yunqiang Su <yunqiang.su@imgtec.com>
40366 * config.gcc (supported_defaults): Add madd4.
40367 (with_madd4): Add validation.
40368 (all_defaults): Add madd4.
40369 * config/mips/mips.opt (mmadd4): New option.
40370 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40372 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40374 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40375 (ISA_HAS_FUSED_MADD4): Likewise.
40376 * doc/invoke.texi (-mmadd4): Document the new option.
40377 * doc/install.texi (--with-madd4): Document the new option.
40379 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40381 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40382 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40383 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40384 (aarch64_init_pauth_hint_builtins): New.
40385 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40386 (aarch64_expand_builtin): Expand new builtins.
40388 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40390 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40391 * combine-stack-adj.c (no_unhandled_cfa): Handle
40392 REG_CFA_TOGGLE_RA_MANGLE.
40393 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40394 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40395 info for return address signing.
40396 (aarch64_expand_epilogue): Likewise.
40398 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40400 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40401 * config/aarch64/aarch64-protos.h
40402 (aarch64_return_address_signing_enabled): New declaration.
40403 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40405 (aarch64_expand_prologue): Sign return address before it's pushed onto
40407 (aarch64_expand_epilogue): Authenticate return address fetched from
40409 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40410 (aarch64_attributes): New function attributes for "sign-return-address".
40411 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40412 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40413 ("*do_return"): Generate combined instructions according to key index.
40414 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40415 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40417 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40418 * config/aarch64/aarch64.opt (msign-return-address=): New.
40419 * doc/extend.texi (AArch64 Function Attributes): Documents
40420 "sign-return-address=".
40421 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40423 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40425 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40426 overall option summary.
40428 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40430 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40431 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40432 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40433 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40435 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40437 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40438 -mpower9-minmax by default for -mcpu=power9.
40439 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40440 128-bit floating point.
40442 2017-01-20 Alan Modra <amodra@gmail.com>
40444 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40445 optimizing for size.
40447 2017-01-20 Alan Modra <amodra@gmail.com>
40450 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40451 for strcmp and strncmp from corresponding builtin decl.
40453 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40455 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40456 instead of i386/rtems-64.h.
40457 * config/i386/rtems-64.h: Remove.
40459 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40463 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40465 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40467 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40469 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40470 Change int to HOST_WIDE_INT.
40471 * config/aarch64/aarch64-protos.h
40472 (aarch64_simd_gen_const_vector_dup): Likewise.
40473 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40475 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40477 * langhooks-def.h (lhd_type_for_size): New decl.
40478 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40479 * langhooks.c (lhd_type_for_size): New function, taken from
40482 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40484 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40485 define_bypass for CR latency.
40486 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40487 (power9-alu2): Add define_bypass for CR latency.
40489 (power9-mul): Update insn latency.
40490 (power9-mul-compare): Update insn latency, bypass latency and remove
40493 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40495 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40497 * config/aarch64/aarch64.md
40498 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40499 aarch64_nopcrelative_literal_loads.
40500 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40502 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40504 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40505 TARGET_LOONGSON_3A.
40506 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40508 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40511 * config.gcc (supported_defaults): Add lxc1-sxc1.
40512 (with_lxc1_sxc1): Add validation.
40513 (all_defaults): Add lxc1-sxc1.
40514 * config/mips/mips.opt (mlxc1-sxc1): New option.
40515 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40517 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40518 __mips_no_lxc1_sxc1.
40519 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40520 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40521 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40523 2017-01-19 Richard Biener <rguenther@suse.de>
40525 PR tree-optimization/72488
40526 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40527 sure to restore SSA info.
40528 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40530 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40532 PR rtl-optimization/79121
40533 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40534 of the inner type when shifting an extended value.
40536 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40539 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40540 interposable aliases.
40542 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40545 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40546 Use the evmergelohi instruction.
40547 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40548 (mov_sitf_e500_subreg8_2_be): Likewise.
40549 (mov_sitf_e500_subreg12_2_le): Likewise.
40550 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40551 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40552 (mov_sitf_e500_subreg8_2_le): Likewise.
40553 (mov_sitf_e500_subreg12_2_be): Likewise.
40555 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40557 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40558 attribute from vecsimple to vecperm.
40559 (altivec_vbpermq2): Likewise.
40561 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40564 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40566 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40567 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40568 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40569 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40570 case where N arg is SIZE_MAX.
40571 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40572 (cmpstrsi): Add pattern.
40574 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40576 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40577 __builtin_vec_revb builtins.
40578 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40579 built-in functions to support generation of the ISA 3.0 XXBR<x>
40580 vector byte reverse instructions.
40581 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40582 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40583 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40584 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40585 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40586 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40587 (P9V_BUILTIN_VEC_REVB): Likewise.
40588 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40589 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40590 (p9_xxbrq_v16qi): Likewise.
40591 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40592 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40593 (p9_xxbrh_v8hi): Likewise.
40594 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40595 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40596 vec_revb built-in functions.
40598 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40600 PR rtl-optimization/78952
40601 * config/i386/i386.md (any_extract): New code iterator.
40602 (*insvqi_2): Use any_extract for source operand.
40603 (*insvqi_3): Use any_shiftrt for source operand.
40605 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40607 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40609 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40611 2017-01-18 Matthias Klose <doko@ubuntu.com>
40613 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40615 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40617 * config/rs6000/altivec.h (vec_bperm): Change #define.
40618 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40619 (altivec_vbpermq2): New define_insn.
40620 (altivec_vbpermd): Likewise.
40621 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40622 function interface.
40623 (VBPERMD): Likewise.
40624 (VBPERM): New polymorphic function interface.
40625 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40626 Add entries for P9V_BUILTIN_VEC_VBPERM.
40627 * doc/extend.texi: Add interfaces for vec_bperm.
40629 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40631 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40632 first letter of error messages.
40633 (s390_resolve_overloaded_builtin): Likewise.
40634 * config/s390/s390.c (s390_expand_builtin): Likewise.
40635 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40636 (s390_valid_target_attribute_inner_p): Likewise.
40637 * config/s390/s390.md ("tabort"): Likewise.
40639 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40641 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40642 (ISA_AVOID_DIV_HILO): New macro.
40643 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40644 (ISA_HAS_DDIV): Likewise.
40646 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40648 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40650 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40652 * doc/invoke.texi (fabi-version): Spelling fix.
40654 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40657 * doc/invoke.texi (fabi-version): Mention mangling fix for
40660 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40663 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40665 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40668 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40669 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40671 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40673 (stack_protect_test): Ditto.
40674 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40675 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40677 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40678 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40679 -mstack-protector-guard-offset=.
40680 (RS/6000 and PowerPC Options): Ditto.
40682 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40684 * config/i386/i386.h (MASK_CLASS_P): New define.
40685 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40686 there are no registers from different register sets also when
40687 mask registers are used. Update function comment.
40688 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40689 to (*k/*r) and (*k/*km) alternatives.
40691 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40693 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40694 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40695 (EH_RETURN_HANDLER_RTX): New define.
40696 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40697 Force frame pointer in EH return functions.
40698 (aarch64_expand_epilogue): Add barrier for eh_return.
40699 (aarch64_final_eh_return_addr): Remove.
40700 (aarch64_eh_return_handler_rtx): New function.
40701 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40703 (aarch64_eh_return_handler_rtx): New prototype.
40705 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40707 * config/rs6000/altivec.h (vec_rlmi): New #define.
40708 (vec_vrlnm): Likewise.
40709 (vec_rlnm): Likewise.
40710 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40711 (UNSPEC_VRLNM): Likewise.
40712 (VIlong): New mode iterator.
40713 (altivec_vrl<VI_char>mi): New define_insn.
40714 (altivec_vrl<VI_char>nm): Likewise.
40715 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40717 (VRLDNM): Likewise.
40718 (RLNM): New polymorphic function entry.
40719 (VRLWMI): New monomorphic function entry.
40720 (VRLDMI): Likewise.
40721 (RLMI): New polymorphic function entry.
40722 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40723 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40724 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40727 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40730 * dwarf2out.c (field_byte_offset): Restore the
40731 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40732 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40733 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40736 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40739 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40741 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40744 * dwarf2out.c (add_data_member_location_attribute): For constant
40745 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40746 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40747 DW_AT_byte_size attributes.
40749 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40751 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40752 after forcing to constant memory when the code model is medium.
40754 2017-01-17 Julia Koval <julia.koval@intel.com>
40757 * config/i386/avx512fintrin.h
40758 (_mm512_i32gather_ps): Change __addr type to void const*.
40759 (_mm512_mask_i32gather_ps): Ditto.
40760 (_mm512_i32gather_pd): Ditto.
40761 (_mm512_mask_i32gather_pd): Ditto.
40762 (_mm512_i64gather_ps): Ditto.
40763 (_mm512_mask_i64gather_ps): Ditto.
40764 (_mm512_i64gather_pd): Ditto.
40765 (_mm512_mask_i64gather_pd): Ditto.
40766 (_mm512_i32gather_epi32): Ditto.
40767 (_mm512_mask_i32gather_epi32): Ditto.
40768 (_mm512_i32gather_epi64): Ditto.
40769 (_mm512_mask_i32gather_epi64): Ditto.
40770 (_mm512_i64gather_epi32): Ditto.
40771 (_mm512_mask_i64gather_epi32): Ditto.
40772 (_mm512_i64gather_epi64): Ditto.
40773 (_mm512_mask_i64gather_epi64): Ditto.
40774 (_mm512_i32scatter_ps): Change __addr type to void*.
40775 (_mm512_mask_i32scatter_ps): Ditto.
40776 (_mm512_i32scatter_pd): Ditto.
40777 (_mm512_mask_i32scatter_pd): Ditto.
40778 (_mm512_i64scatter_ps): Ditto.
40779 (_mm512_mask_i64scatter_ps): Ditto.
40780 (_mm512_i64scatter_pd): Ditto.
40781 (_mm512_mask_i64scatter_pd): Ditto.
40782 (_mm512_i32scatter_epi32): Ditto.
40783 (_mm512_mask_i32scatter_epi32): Ditto.
40784 (_mm512_i32scatter_epi64): Ditto.
40785 (_mm512_mask_i32scatter_epi64): Ditto.
40786 (_mm512_i64scatter_epi32): Ditto.
40787 (_mm512_mask_i64scatter_epi32): Ditto.
40788 (_mm512_i64scatter_epi64): Ditto.
40789 (_mm512_mask_i64scatter_epi64): Ditto.
40790 * config/i386/avx512pfintrin.h
40791 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40792 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40793 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40794 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40795 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40796 (_mm512_prefetch_i32scatter_ps): Ditto.
40797 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40798 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40799 (_mm512_prefetch_i64scatter_pd): Ditto.
40800 (_mm512_prefetch_i64scatter_ps): Ditto.
40801 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40802 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40803 * config/i386/avx512vlintrin.h
40804 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40805 (_mm_mmask_i32gather_ps): Ditto.
40806 (_mm256_mmask_i32gather_pd): Ditto.
40807 (_mm_mmask_i32gather_pd): Ditto.
40808 (_mm256_mmask_i64gather_ps): Ditto.
40809 (_mm_mmask_i64gather_ps): Ditto.
40810 (_mm256_mmask_i64gather_pd): Ditto.
40811 (_mm_mmask_i64gather_pd): Ditto.
40812 (_mm256_mmask_i32gather_epi32): Ditto.
40813 (_mm_mmask_i32gather_epi32): Ditto.
40814 (_mm256_mmask_i32gather_epi64): Ditto.
40815 (_mm_mmask_i32gather_epi64): Ditto.
40816 (_mm256_mmask_i64gather_epi32): Ditto.
40817 (_mm_mmask_i64gather_epi32): Ditto.
40818 (_mm256_mmask_i64gather_epi64): Ditto.
40819 (_mm_mmask_i64gather_epi64): Ditto.
40820 (_mm256_i32scatter_ps): Change __addr type to void*.
40821 (_mm256_mask_i32scatter_ps): Ditto.
40822 (_mm_i32scatter_ps): Ditto.
40823 (_mm_mask_i32scatter_ps): Ditto.
40824 (_mm256_i32scatter_pd): Ditto.
40825 (_mm256_mask_i32scatter_pd): Ditto.
40826 (_mm_i32scatter_pd): Ditto.
40827 (_mm_mask_i32scatter_pd): Ditto.
40828 (_mm256_i64scatter_ps): Ditto.
40829 (_mm256_mask_i64scatter_ps): Ditto.
40830 (_mm_i64scatter_ps): Ditto.
40831 (_mm_mask_i64scatter_ps): Ditto.
40832 (_mm256_i64scatter_pd): Ditto.
40833 (_mm256_mask_i64scatter_pd): Ditto.
40834 (_mm_i64scatter_pd): Ditto.
40835 (_mm_mask_i64scatter_pd): Ditto.
40836 (_mm256_i32scatter_epi32): Ditto.
40837 (_mm256_mask_i32scatter_epi32): Ditto.
40838 (_mm_i32scatter_epi32): Ditto.
40839 (_mm_mask_i32scatter_epi32): Ditto.
40840 (_mm256_i32scatter_epi64): Ditto.
40841 (_mm256_mask_i32scatter_epi64): Ditto.
40842 (_mm_i32scatter_epi64): Ditto.
40843 (_mm_mask_i32scatter_epi64): Ditto.
40844 (_mm256_i64scatter_epi32): Ditto.
40845 (_mm256_mask_i64scatter_epi32): Ditto.
40846 (_mm_i64scatter_epi32): Ditto.
40847 (_mm_mask_i64scatter_epi32): Ditto.
40848 (_mm256_i64scatter_epi64): Ditto.
40849 (_mm256_mask_i64scatter_epi64): Ditto.
40850 (_mm_i64scatter_epi64): Ditto.
40851 (_mm_mask_i64scatter_epi64): Ditto.
40852 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40853 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40854 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40855 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40856 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40857 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40858 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40859 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40860 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40861 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40862 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40863 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40864 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40865 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40866 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40867 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40868 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40869 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40870 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40871 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40872 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40873 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40874 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40875 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40876 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40877 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40878 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40879 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40880 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40881 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40882 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40883 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40884 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40885 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40886 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40887 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40888 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40889 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40890 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40891 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40892 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40893 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40894 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40895 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40896 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40897 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40898 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40899 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40900 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40901 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40902 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40903 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40904 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40905 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40906 definitions accordingly.
40908 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40909 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40912 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40915 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40918 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40919 hard regs for inner regno.
40921 2017-01-17 Martin Liska <mliska@suse.cz>
40924 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40925 assumption and add comment.
40927 2017-01-17 Nathan Sidwell <nathan@acm.org>
40929 * ipa-visibility.c (localize_node): New function, broken out of ...
40930 (function_and_variable_visibility): ... here. Call it.
40932 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40934 PR middle-end/77445
40935 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40936 correctly set frequency of oudgoing edge.
40937 (duplicate_thread_path): Fix profile updating.
40939 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40942 * configure.ac: Add GCC_BASE_VER.
40943 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40944 version from BASE-VER file.
40945 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40946 (gcc.o): Depend on $(BASEVER).
40947 * common.opt (dumpfullversion): New option.
40948 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40949 * doc/invoke.texi: Document -dumpfullversion.
40950 * doc/install.texi: Document --with-gcc-major-version-only.
40951 * configure: Regenerated.
40953 2017-01-17 Richard Biener <rguenther@suse.de>
40955 PR tree-optimization/71433
40956 * tree-vrp.c (register_new_assert_for): Merge same asserts
40957 on all incoming edges.
40958 (process_assert_insertions_for): Handle insertions at the
40961 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40963 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40964 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40966 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40969 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40972 2017-01-17 Alan Modra <amodra@gmail.com>
40975 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40976 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40977 symbolic stack limit when pic.
40979 2017-01-16 Martin Sebor <msebor@redhat.com>
40981 PR tree-optimization/78608
40982 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40984 2017-01-16 Jeff Law <law@redhat.com>
40987 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40988 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40989 for several include directories that may be relative to sysroot.
40990 * config/i386/x-mingw32 (gplus_includedir): Define.
40991 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40992 (native_system_includedir): Likewise.
40993 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40994 override if TARGET_SYSTEM_ROOT is defined.
40995 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40997 PR tree-optimization/79090
40998 PR tree-optimization/33562
40999 PR tree-optimization/61912
41000 PR tree-optimization/77485
41001 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
41002 and computed trims into the dump file.
41004 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
41006 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
41008 2017-01-16 Jakub Jelinek <jakub@redhat.com>
41011 * gimplify.c (gimplify_init_constructor): If want_value and
41012 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
41016 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
41017 sequence. Formatting fixes.
41018 (doloop_optimize): Formatting fixes.
41021 * gcc.c (debug_level_greater_than_spec_func): New function.
41022 (static_spec_functions): Add debug-level-gt spec function.
41023 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
41025 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41026 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41027 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
41028 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
41029 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
41030 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
41032 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
41034 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
41035 QImode fixups to general and mask registers only.
41037 2017-01-16 Carl Love <cel@us.ibm.com>
41039 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
41040 for built-in functions
41041 vector signed char vec_nabs (vector signed char)
41042 vector signed short vec_nabs (vector signed short)
41043 vector signed int vec_nabs (vector signed int)
41044 vector signed long long vec_nabs (vector signed long long)
41045 vector float vec_nabs (vector float)
41046 vector double vec_nabs (vector double)
41047 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
41049 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
41050 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
41051 * doc/extend.texi: Update the documentation file for the new built-in
41054 2017-01-16 Martin Sebor <msebor@redhat.com>
41056 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
41059 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41061 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41062 UNSPEC_VSX__XXSPLTD to require special splat handling.
41064 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41067 * system.h: Poison strndup.
41069 2017-01-16 Alan Modra <amodra@gmail.com>
41072 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41075 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41077 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41079 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41081 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41082 call recog here. Assert that INSN_CODE (insn) is non-negative.
41084 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41087 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41089 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41090 in the currently scheduled RTL fragment.
41092 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41094 PR rtl-optimization/78751
41095 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41098 2017-01-14 Jeff Law <law@redhat.com>
41100 PR tree-optimization/79090
41101 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41102 variable length stores.
41103 (compute_trims): Delete dead assignment to *trim_tail.
41104 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41107 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41109 PR rtl-optimization/78626
41110 PR rtl-optimization/78727
41111 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41112 of a block, and split such blocks after everything else is finished.
41114 2017-01-14 Alan Modra <amodra@gmail.com>
41117 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41118 target legitimate_combined_insn.
41119 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41120 (rs6000_legitimate_combined_insn): New function.
41121 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41123 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41124 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41125 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41127 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41129 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41131 2017-01-13 Jeff Law <law@redhat.com>
41133 PR tree-optimization/33562
41134 PR tree-optimization/61912
41135 PR tree-optimization/77485
41136 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41138 (delete_dead_assignment): Likewise.
41139 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41140 statement to delete_dead_call and delete_dead_assignment.
41142 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41145 * substring-locations.c (format_warning_va): Strengthen case 1 so
41146 that both endpoints of the substring must be within the format
41147 range for just the substring to be printed.
41149 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41151 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41152 * config/i386/i386.c (ix86_target_string): Add missing options
41153 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41154 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41155 flags_other and ix86_target_other to flags2_other. Display unknown
41157 (ix86_valid_target_attribute_inner_p): Add missing options and
41158 reorder options by implied ISAs, as in ix86_target_string.
41160 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41162 * hash-table.h (hash_table::too_empty_p): New function.
41163 (hash_table::expand): Use it.
41164 (hash_table::traverse): Likewise.
41165 (hash_table::empty_slot): Use sizeof (value_type) instead of
41166 sizeof (PTR) to convert bytes to elements. Shrink the table
41167 if the current size is excessive for the current number of
41170 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41172 * ira-costs.c (record_reg_classes): Break from the inner loop
41173 early once alt_fail is known to be true. Update outer loop
41174 handling accordingly.
41176 2017-01-13 Jeff Law <law@redhat.com>
41178 * tree-ssa-dse.c (decrement_count): New function.
41179 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41180 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41181 when we know the partially dead statement is a mem* function.
41183 PR tree-optimization/61912
41184 PR tree-optimization/77485
41185 * tree-ssa-dse.c: Include expr.h.
41186 (maybe_trim_constructor_store): New function.
41187 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41189 PR tree-optimization/33562
41190 PR tree-optimization/61912
41191 PR tree-optimization/77485
41192 * doc/invoke.texi: Document new dse-max-object-size param.
41193 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41194 * tree-ssa-dse.c: Include params.h.
41195 (dse_store_status): New enum.
41196 (initialize_ao_ref_for_dse): New, partially extracted from
41198 (valid_ao_ref_for_dse, normalize_ref): New.
41199 (setup_live_bytes_from_ref, compute_trims): Likewise.
41200 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41201 (maybe_trim_partially_dead_store): Likewise.
41202 (maybe_trim_complex_store): Likewise.
41203 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41204 Track what bytes live from the original store. Return tri-state
41205 for dead, partially dead or live.
41206 (dse_dom_walker): Add constructor, destructor and new private members.
41207 (delete_dead_call, delete_dead_assignment): New extracted from
41209 (dse_optimize_stmt): Make a member of dse_dom_walker.
41210 Use initialize_ao_ref_for_dse.
41212 PR tree-optimization/33562
41213 PR tree-optimization/61912
41214 PR tree-optimization/77485
41215 * sbitmap.h (bitmap_count_bits): Prototype.
41216 (bitmap_clear_range, bitmap_set_range): Likewise.
41217 * sbitmap.c (bitmap_clear_range): New function.
41218 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41220 2017-01-13 Martin Liska <mliska@suse.cz>
41223 * function.c (set_cfun): Add new argument force.
41224 * function.h (set_cfun): Likewise.
41225 * ipa-inline-transform.c (inline_call): Use the function when
41226 strict alising from is dropped for function we inline to.
41228 2017-01-13 Richard Biener <rguenther@suse.de>
41230 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41231 for dumping GIMPLE INTEGER_CSTs.
41233 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41235 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41236 to 201112L since C++17.
41238 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41241 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41242 if -fsanitize=kernel-address is present.
41244 2017-01-13 Richard Biener <rguenther@suse.de>
41246 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41247 as _Literal ( type ) number in case usual suffixes do not
41248 preserve all information.
41250 2017-01-13 Richard Biener <rguenther@suse.de>
41252 PR tree-optimization/77283
41253 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41254 and ssa-iterators.h.
41255 (is_feasible_trace): Implement a cost model based on joiner
41258 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41261 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41262 char or short to __float128/_Float128 directly.
41264 2017-01-12 Martin Sebor <msebor@redhat.com>
41266 to -Wformat-overflow.
41267 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41268 (min_bytes_remaining): Same.
41269 (get_string_length): Same.
41270 (format_string): Same.
41271 (format_directive): Same.
41273 (pass_sprintf_length::handle_gimple_call): Same.
41275 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41277 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41278 info.nowrite calls with no lhs that can't throw. Return bool
41279 whether gsi_remove has been called or not.
41280 (pass_sprintf_length::handle_gimple_call): Return bool whether
41281 try_substitute_return_value called gsi_remove. Formatting fix.
41282 (pass_sprintf_length::execute): Don't use gsi_remove if
41283 handle_gimple_call returned true.
41286 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41287 be removed due to side-effects, don't remove following barrier nor
41288 turn the successor edge into fallthru edge.
41290 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41293 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41294 element-reversing loads and stores as not swappable.
41296 2017-01-12 Nathan Sidwell <nathan@acm.org>
41297 Nicolai Stange <nicstange@gmail.com>
41299 * combine.c (try_combine): Don't ignore result of overlap checking
41300 loop. Combine overlap & asm check into single loop.
41302 2017-01-12 Richard Biener <rguenther@suse.de>
41304 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41305 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41307 2017-01-12 Richard Biener <rguenther@suse.de>
41309 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41310 and TS_TARGET_OPTION directly derive from TS_BASE.
41311 * tree-core.h (tree_optimization_option): Derive from tree_base.
41312 (tree_target_option): Likewise.
41314 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41316 * config/i386/i386.c (memory_address_length): Increase len
41317 only when rip_relative_addr_p returns false.
41319 2017-01-11 Julia Koval <julia.koval@intel.com>
41321 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41322 (OPTION_MASK_ISA_SGX_SET): New.
41323 (ix86_handle_option): Handle OPT_msgx.
41324 * config.gcc: Added sgxintrin.h.
41325 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41326 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41327 * config/i386/i386.c (ix86_target_string): Add -msgx.
41329 (ix86_option_override_internal): Handle new options.
41330 (ix86_valid_target_attribute_inner_p): Add sgx.
41331 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41332 * config/i386/i386.opt: Add msgx.
41333 * config/i386/sgxintrin.h: New file.
41334 * config/i386/x86intrin.h: Add sgxintrin.h.
41336 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41339 * fold-const.c (maybe_nonzero_address): Return 1 for function
41341 (tree_single_nonzero_warnv_p): Don't handle function local objects
41345 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41348 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41351 * opts.c: Include "spellcheck.h"
41352 (struct string_fragment): New struct.
41353 (struct edit_distance_traits<const string_fragment &>): New
41355 (get_closest_sanitizer_option): New function.
41356 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41358 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41360 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41362 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41363 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41364 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41365 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41366 for initial die_offset if dwarf_split_debug_info.
41367 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41368 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41370 (output_skeleton_debug_sections): Formatting fix. Use
41371 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41372 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41374 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41376 * config/arm/cortex-a53.md: Add bypasses for
41377 cortex_a53_r2f_cvt.
41378 (cortex_a53_r2f): Only use for transfers.
41379 (cortex_a53_f2r): Likewise.
41380 (cortex_a53_r2f_cvt): Add reservation for conversions.
41381 (cortex_a53_f2r_cvt): Likewise.
41383 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41385 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41386 to all inlined functions, change static to extern.
41388 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41391 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41393 (arm_assemble_integer): Likewise.
41395 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41397 * config.gcc: Use new awk script to check CPU, FPU and architecture
41398 parameters for --with-... options.
41399 * config/arm/parsecpu.awk: New file
41400 * config/arm/arm-cpus.in: New file.
41401 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41403 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41405 * config/arm/t-arm: Update dependency rules.
41406 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41407 of processing .def files.
41408 * config/arm/genopt.sh: Deleted.
41409 * config/arm/gentune.sh: Deleted.
41410 * config/arm/arm-cores.def: Deleted.
41411 * config/arm/arm-arches.def: Deleted.
41412 * config/arm/arm-fpus.def: Deleted.
41413 * config/arm/arm-tune.md: Regenerated.
41414 * config/arm/arm-tables.opt: Regenerated.
41415 * config/arm/arm-cpu.h: New generated file.
41416 * config/arm/arm-cpu-data.h: New generated file.
41417 * config/arm/arm-cpu-cdata.h: New generated file.
41419 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41422 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41424 (input_varpool_node): Unpack dynamically_initialized bit.
41426 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41428 PR rtl-optimization/79032
41429 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41430 the alignment of the adjusted memory reference against that of MODE,
41431 instead of the alignment of the original memory reference.
41433 2017-01-11 Martin Jambor <mjambor@suse.cz>
41435 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41437 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41438 decorated functions.
41440 2017-01-11 Richard Biener <rguenther@suse.de>
41442 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41443 set range/nonnull info for PHI results. Do not set it on
41444 stmts marked for removal.
41446 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41448 * expr.c (store_field): In the bitfield case, fetch the return value
41449 from the registers before applying a single big-endian adjustment.
41450 Always do a final load for a BLKmode value not larger than a word.
41452 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41455 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41456 that we correctly handle column numbers greater than
41457 LINE_MAP_MAX_COLUMN_NUMBER.
41459 2017-01-10 Martin Sebor <msebor@redhat.com>
41461 PR middle-end/78245
41462 * gimple-ssa-sprintf.c (get_destination_size): Call
41463 {init,fini}object_sizes.
41464 * tree-object-size.c (addr_object_size): Adjust.
41465 (pass_through_call): Adjust.
41466 (pass_object_sizes::execute): Adjust.
41467 * tree-object-size.h (fini_object_sizes): Declare.
41469 2017-01-10 Martin Sebor <msebor@redhat.com>
41471 PR tree-optimization/78775
41472 * builtins.c (get_size_range): Move...
41473 * calls.c: ...to here.
41474 (alloc_max_size): Accept zero argument.
41475 (operand_signed_p): Remove.
41476 (maybe_warn_alloc_args_overflow): Call get_size_range.
41477 * calls.h (get_size_range): Declare.
41479 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41481 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41482 from TI's devices.csv file as of September 2016.
41483 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41485 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41487 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41488 * doc/invoke.texi: Likewise.
41489 * doc/md.texi: Likewise.
41490 * doc/objc.texi: Likewise.
41492 2017-01-10 Joshua Conner <joshconner@google.com>
41494 * config/arm/fuchsia-elf.h: New file.
41495 * config/fuchsia.h: New file.
41496 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41497 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41499 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41501 2016-01-10 Richard Biener <rguenther@suse.de>
41503 PR tree-optimization/79034
41504 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41505 Propagate out degenerate PHIs in the joiner.
41507 2017-01-10 Martin Liska <mliska@suse.cz>
41509 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41510 (sort_congruence_classes_by_decl_uid): Likewise.
41511 (sort_congruence_class_groups_by_decl_uid): Likewise.
41512 (sem_item_optimizer::merge_classes): Sort class, groups in these
41513 classes and members in the groups by DECL_UID of declarations.
41514 This would make merge operations stable.
41516 2017-01-10 Martin Liska <mliska@suse.cz>
41518 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41519 usage of m_classes_vec.
41520 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41521 (sem_item_optimizer::get_group_by_hash): Likewise.
41522 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41523 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41524 (sem_item_optimizer::verify_classes): Likewise.
41525 (sem_item_optimizer::process_cong_reduction): Likewise.
41526 (sem_item_optimizer::dump_cong_classes): Likewise.
41527 (sem_item_optimizer::merge_classes): Likewise.
41528 * ipa-icf.h (congruence_class_hash): Rename from
41529 congruence_class_group_hash. Remove declaration of m_classes_vec.
41531 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41533 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41534 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41535 * config.gcc: Add avx512vpopcntdqintrin.h.
41536 * config/i386/avx512vpopcntdqintrin.h: New.
41537 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41538 * config/i386/i386-builtin-types.def: Add new types.
41539 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41540 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41541 __builtin_ia32_vpopcountq_v8di_mask): New.
41542 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41543 __AVX512VPOPCNTDQ__.
41544 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41545 (PTA_AVX512VPOPCNTDQ): Define.
41546 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41547 TARGET_AVX512VPOPCNTDQ_P): Define.
41548 * config/i386/i386.opt: Add mavx512vpopcntdq.
41549 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41550 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41552 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41554 PR middle-end/77484
41555 * predict.def (PRED_CALL): Set to 67.
41557 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41559 * expr.c (store_field): In the bitfield case, if the value comes from
41560 a function call and is of an aggregate type returned in registers, do
41561 not modify the field mode; extract the value in all cases if the mode
41562 is BLKmode and the size is not larger than a word.
41564 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41567 * config/i386/cpuid.h: Fix undefined behavior.
41569 2017-01-04 Jeff Law <law@redhat.com>
41571 PR tree-optimization/79007
41572 PR tree-optimization/67955
41573 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41574 conservative for pt.null when flag_non_call_exceptions is on.
41576 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41578 PR translation/79019
41579 PR translation/79020
41580 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41581 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41582 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41584 * config/avr/avr.opt (maccumulate-args): Likewise.
41585 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41586 * common.opt (freport-bug): Likewise.
41587 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41588 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41589 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41590 translatable string.
41591 * config/i386/i386.c (function_value_32): Likewise.
41592 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41593 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41595 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41596 * common/config/msp430/msp430-common.c (msp430_handle_option):
41598 * symtab.c (symtab_node::verify_base): Likewise.
41599 * opts.c (set_debug_level): Likewise.
41600 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41601 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41602 missing whitespace to translatable strings.
41603 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41604 * config/sh/superh.h: Likewise.
41605 * config/i386/xopintrin.h: Likewise.
41606 * config/i386/znver1.md: Likewise.
41607 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41608 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41609 * double-int.h (struct double_int): Likewise.
41610 * double-int.c (div_and_round_double): Likewise.
41611 * wide-int.cc: Likewise.
41612 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41613 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41614 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41615 (crossjumps_occurred): ... this.
41616 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41619 PR tree-optimization/78899
41620 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41621 returning bool return struct loop *, NULL for failure and the new
41623 (versionable_outer_loop_p): Don't version outer loop if it has
41624 dont_vectorized bit set.
41625 (tree_if_conversion): When versioning outer loop, ensure
41626 tree_if_conversion is performed also on the inner loop of the
41627 non-vectorizable outer loop copy.
41628 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41629 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41630 prevent vectorization of it.
41631 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41632 the outer loop vectorization of the non-scalar version is attempted
41633 before vectorization of the inner loop in scalar version. If
41634 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41635 vectorization of its inner loop.
41636 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41637 has 2 inner loops, rename also on edges from bb whose single pred
41638 is outer_loop->header. Fix typo in function comment.
41640 2017-01-09 Martin Sebor <msebor@redhat.com>
41643 * asan.c (asan_emit_stack_protection): Increase local buffer size
41644 to avoid snprintf truncation warning.
41646 2017-01-09 Andrew Pinski <apinski@cavium.com>
41648 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41649 to reference thunderx2t99 for the tuning structure
41650 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41652 (thunderx2t99_extra_costs): This.
41653 * config/aarch64/aarch64-tune.md: Regenerate.
41654 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41655 (vulcan_addrcost_table): This.
41656 (vulcan_regmove_cost): Rename to ...
41657 (thunderx2t99_regmove_cost): This.
41658 (vulcan_vector_cost): Rename to ...
41659 (thunderx2t99_vector_cost): this.
41660 (vulcan_branch_cost): Rename to ...
41661 (thunderx2t99_branch_cost): This.
41662 (vulcan_tunings): Rename to ...
41663 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41664 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41666 2017-01-09 Martin Jambor <mjambor@suse.cz>
41670 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41671 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41672 (propagate_vr_accross_jump_function): Use the above function for all
41673 value range computations for pass-through jump functions and type
41674 converasion from explicit value range values.
41675 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41676 parameters from TYPE_ARG_TYPES.
41677 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41678 (ipa_write_node_info): Stream type of the actual argument.
41679 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41681 2017-01-09 Martin Liska <mliska@suse.cz>
41684 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41685 (lookup_compiler): Do not show error message with have_E.
41687 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41689 PR tree-optimization/78938
41690 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41691 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41692 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41693 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41696 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41698 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41701 2017-01-09 Richard Biener <rguenther@suse.de>
41703 PR tree-optimization/78997
41704 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41705 name condition properly.
41707 2017-01-09 Richard Biener <rguenther@suse.de>
41710 * dwarf2out.c (is_cxx): New overload with context.
41711 (is_naming_typedef_decl): Use it.
41713 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41715 * invoke.texi (Option Summary): Correct spacing in option lists
41716 and add line breaks to fix over-long lines.
41718 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41720 PR middle-end/17660
41722 * extend.texi (Common Variable Attributes): Add xref to GCC
41723 Internals manual to explain mode attribute keywords.
41725 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41728 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41729 and Preprocessor Options.
41730 (Options for Linking): Document -pthread here....
41731 (RS/6000 and PowerPC Options): ...not here.
41732 (Solaris 2 Options): ...or here.
41733 * doc/cppopts.texi: Document -pthread.
41735 2017-01-08 Martin Sebor <msebor@redhat.com>
41737 PR middle-end/77708
41738 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41739 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41740 New member functions.
41741 (format_directive): Used them.
41743 (pass_sprintf_length::handle_gimple_call): Same.
41744 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41745 to avoid truncation for any argument.
41746 (extract_affine_mul): Same.
41747 * tree.c (get_file_function_name): Same.
41749 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41751 PR middle-end/77484
41752 * predict.def (PRED_INDIR_CALL): Set to 86.
41754 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41756 PR preprocessor/54124
41757 * doc/cppopts.texi: Reformat -d subtable to list the full name
41758 of the options. Add cross-reference to the docs for the general
41759 compiler -d options.
41760 * doc/invoke.texi (Developer Options): Add cross-reference to the
41761 preprocessor-specific -d option documentation.
41763 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41765 PR preprocessor/13498
41766 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41767 redudant material, and reflect new command-line options.
41768 (System Headers): Likewise.
41770 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41772 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41773 -isystem, and -idirafter. Copy-edit.
41774 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41775 default for -ftrack-macro-expansion. Delete obsolete and
41776 badly-formatted implementation details about -fdebug-cpp output.
41777 * doc/cppwarnopts.texi: Copy-edit.
41779 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41782 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41783 that the transition from a max line width >= 1<<10 to narrower
41784 lines works correctly.
41786 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41788 * doc/options.texi (PerFunction): New.
41789 * opt-functions.awk (switch_flags): Map both Optimization and
41790 PerFunction to CL_OPTIMIZATION.
41791 * opth-gen.awk: Test for PerFunction flag along with
41793 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41794 it only when the latter is present. Skip those that don't in
41795 the hash function generator.
41796 * common.opt (fvar-tracking): Mark as PerFunction instead of
41798 (fvar-tracking-assignments): Likewise.
41799 (fvar-tracking-assignments-toggle): Likewise.
41800 (fvar-tracking-uninit): Likewise.
41802 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41804 PR translation/79018
41805 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41808 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41811 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41812 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41813 TARGET_LONG_JUMP_TABLE_OFFSETS.
41814 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41815 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41816 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41817 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41818 * config/m68k/m68k.md (tablejump expander): Likewise.
41819 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41820 TARGET_LONG_JUMP_TABLE_OFFSETS.
41821 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41822 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41824 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41825 David Holsgrove <david.holsgrove@xilinx.com>
41827 * common/config/microblaze/microblaze-common.c
41828 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41829 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41831 * config/microblaze/microblaze.c (microblaze_must_save_register)
41832 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41834 (microblaze_eh_return): New function.
41835 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41836 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41837 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41838 * config/microblaze/microblaze.md (eh_return): New pattern.
41840 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41842 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41843 GCC_DIAGNOSTIC_STRINGIFY): Define.
41845 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41847 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41849 * config/arm/arm.md (<mcrr>): New.
41851 * config/arm/arm.c (arm_arch5te): New.
41852 (arm_option_override): Set arm_arch5te.
41853 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41855 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41856 (arm_mcrr_qualifiers): ... this. New.
41857 (MRRC_QUALIFIERS): Define to...
41858 (arm_mrrc_qualifiers): ... this. New.
41859 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41861 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41862 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41863 (MRRCI, mrrc, MRRC): New.
41864 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41865 VUNSPEC_MRRC2): New.
41867 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41869 * config/arm/arm.md (<mcr>): New.
41871 * config/arm/arm.c (arm_coproc_builtin_available): Add
41872 support for mcr, mrc, mcr2 and mrc2.
41873 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41874 (arm_mcr_qualifiers): ... this. New.
41875 (MRC_QUALIFIERS): Define to ...
41876 (arm_mrc_qualifiers): ... this. New.
41877 (MCR_QUALIFIERS): Define to ...
41878 (arm_mcr_qualifiers): ... this. New.
41879 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41881 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41882 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41883 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41884 VUNSPEC_MRC2): New.
41886 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41888 * config/arm/arm.md (*ldc): New.
41892 * config/arm/arm.c (arm_coproc_builtin_available): Add
41893 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41894 (arm_coproc_ldc_stc_legitimate_address): New.
41895 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41896 'qualifier_const_pointer'.
41897 (LDC_QUALIFIERS): Define to...
41898 (arm_ldc_qualifiers): ... this. New.
41899 (STC_QUALIFIERS): Define to...
41900 (arm_stc_qualifiers): ... this. New.
41901 * config/arm/arm-protos.h
41902 (arm_coproc_ldc_stc_legitimate_address): New.
41903 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41904 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41905 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41906 stc2, stcl, stc2l): New.
41907 * config/arm/constraints.md (Uz): New.
41908 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41909 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41910 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41911 VUNSPEC_STC2L): New.
41913 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41915 * config/arm/arm.md (<cdp>): New.
41916 * config/arm/arm.c (neon_const_bounds): Rename this ...
41917 (arm_const_bounds): ... this.
41918 (arm_coproc_builtin_available): New.
41919 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41920 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41921 (CDP_QUALIFIERS): Define to...
41922 (arm_cdp_qualifiers): ... this. New.
41924 (arm_expand_builtin_args): Add case for 6 arguments.
41925 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41926 (arm_const_bounds): ... this.
41927 (arm_coproc_builtin_available): New.
41928 * config/arm/arm_acle.h (__arm_cdp): New.
41930 * config/arm/arm_acle_builtins.def (cdp): New.
41932 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41933 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41934 'arm_const_bounds'.
41935 * config/arm/types.md (coproc): New.
41936 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41937 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41938 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41939 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41941 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41943 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41944 (UBINOP_QUALIFIERS): New.
41946 (acle_builtin_data): New. Change comment.
41947 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41948 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41949 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41950 arm_acle_builtins.def.
41951 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41952 (arm_init_acle_builtins): New.
41953 (CRC32_BUILTIN): Remove.
41954 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41955 crc32cb, crc32ch and crc32cw.
41956 (arm_init_crc32_builtins): Remove.
41957 (arm_init_builtins): Use arm_init_acle_builtins rather
41958 than arm_init_crc32_builtins.
41959 (arm_expand_acle_builtin): New.
41960 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41961 * config/arm/arm_acle_builtins.def: New.
41963 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41965 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41966 (arm_builtin_datum): ... this.
41967 (arm_init_neon_builtin): Rename to ...
41968 (arm_init_builtin): ... this. Add a new parameters PREFIX
41969 and USE_SIG_IN_NAME.
41970 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41971 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41972 'arm_builtin_datum'.
41973 (arm_init_vfp_builtins): Likewise.
41974 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41975 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41976 (arm_expand_neon_args): Rename to ...
41977 (arm_expand_builtin_args): ... this. Rename builtin_arg
41978 enum values and differentiate between ARG_BUILTIN_MEMORY
41979 and ARG_BUILTIN_NEON_MEMORY.
41980 (arm_expand_neon_builtin_1): Rename to ...
41981 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41982 values, arm_expand_builtin_args and add bool parameter NEON.
41983 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41984 (arm_expand_vfp_builtin): Likewise.
41985 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41987 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41989 PR middle-end/77484
41990 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41991 * predict.c (tree_estimate_probability_bb): Reverse direction of
41992 polymorphic call predictor.
41994 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41996 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41997 (determine_pass_name_match): ...this new function and...
41998 (should_skip_pass_p): ...this new function.
42000 2017-01-06 Nathan Sidwell <nathan@acm.org>
42002 * ipa-visibility.c (function_and_variable_visibility): Reformat
42003 comments and long lines. Remove extrneous if.
42004 * symtab.c (symtab_node::make_decl_local): Fix code format.
42005 (symtab_node::set_section_for_node): Fix comment typo.
42007 2017-01-06 Martin Liska <mliska@suse.cz>
42010 * lra-constraints.c: Rename invariant to lra_invariant.
42011 * predict.c (set_even_probabilities): Initialize e to NULL.
42013 2017-01-05 Martin Sebor <msebor@redhat.com>
42015 PR tree-optimization/78910
42016 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
42017 (format_integer): Correct off-by-one error in the handling
42018 of precision with negative numbers in signed conversions..
42020 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
42022 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
42024 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42026 PR tree-optimization/71016
42027 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
42028 factor_out_conditional_conversion. Formatting fix.
42029 (factor_out_conditional_conversion): Add cond_stmt argument.
42030 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
42031 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
42034 2017-01-05 David Malcolm <dmalcolm@redhat.com>
42036 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
42037 read-rtl-function.o, and selftest-rtl.o.
42038 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
42039 (selftest::aarch64_test_loading_full_dump): New function.
42040 (selftest::aarch64_run_selftests): New function.
42041 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
42042 selftest::aarch64_run_selftests.
42043 * config/i386/i386.c
42044 (selftest::ix86_test_loading_dump_fragment_1): New function.
42045 (selftest::ix86_test_loading_call_insn): New function.
42046 (selftest::ix86_test_loading_full_dump): New function.
42047 (selftest::ix86_test_loading_unspec): New function.
42048 (selftest::ix86_run_selftests): Call the new functions.
42049 * emit-rtl.c (maybe_set_max_label_num): New function.
42050 * emit-rtl.h (maybe_set_max_label_num): New decl.
42051 * function.c (instantiate_decls): Guard call to
42052 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
42053 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
42055 * gensupport.c (gen_reader::gen_reader): Pass "false"
42056 for new "compact" param of rtx_reader.
42057 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
42058 rather than an empty string for NULL strings.
42059 * read-md.c: Potentially include config.h rather than bconfig.h.
42060 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42061 (have_error): New global, copied from errors.c.
42062 (md_reader::read_name): Rename to...
42063 (md_reader::read_name_1): ...this, adding "out_loc" param,
42064 and converting "missing name or number" to returning false, rather
42066 (md_reader::read_name): Reimplement in terms of read_name_1.
42067 (md_reader::read_name_or_nil): New function.
42068 (md_reader::read_string): Handle "(nil)" by returning NULL.
42069 (md_reader::md_reader): Add new param "compact".
42070 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42071 (md_reader::read_file): New method.
42072 * read-md.h (md_reader::md_reader): Add new param "compact".
42073 (md_reader::read_file): New method.
42074 (md_reader::is_compact): New accessor.
42075 (md_reader::read_name): Convert return type from void to file_location.
42076 (md_reader::read_name_or_nil): New decl.
42077 (md_reader::read_name_1): New decl.
42078 (md_reader::m_compact): New field.
42079 (noop_reader::noop_reader): Pass "false" for new "compact" param
42081 (rtx_reader::rtx_reader): Add new "compact" param.
42082 (rtx_reader::read_rtx_operand): Make virtual and convert return
42083 type from void to rtx.
42084 (rtx_reader::read_until): New decl.
42085 (rtx_reader::handle_any_trailing_information): New virtual function.
42086 (rtx_reader::postprocess): New virtual function.
42087 (rtx_reader::finalize_string): New virtual function.
42088 (rtx_reader::m_in_call_function_usage): New field.
42089 (rtx_reader::m_reuse_rtx_by_id): New field.
42090 * read-rtl-function.c: New file.
42091 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42092 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42093 (selftest::verify_three_block_rtl_cfg): New decl.
42094 * read-rtl-function.h: New file.
42095 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42096 For host, include function.h, memmodel.h, and emit-rtl.h.
42097 (one_time_initialization): New function.
42098 (struct compact_insn_name): New struct.
42099 (compact_insn_names): New array.
42100 (find_code): Handle insn codes in compact dumps.
42101 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42102 (bind_subst_iter_and_attr): Likewise.
42103 (add_condition_to_string): Likewise.
42104 (add_condition_to_rtx): Likewise.
42105 (apply_attribute_uses): Likewise.
42106 (add_current_iterators): Likewise.
42107 (apply_iterators): Likewise.
42108 (initialize_iterators): Guard usage of apply_subst_iterator with
42109 #ifdef GENERATOR_FILE.
42110 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42111 (md_reader::read_mapping): Likewise.
42112 (add_define_attr_for_define_subst): Likewise.
42113 (add_define_subst_attr): Likewise.
42114 (read_subst_mapping): Likewise.
42115 (check_code_iterator): Likewise.
42116 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42118 (one_time_initialization): New function.
42119 (rtx_reader::read_until): New method.
42120 (read_flags): New function.
42121 (parse_reg_note_name): New function.
42122 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42123 Handle reuse_rtx ids.
42124 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42125 Add parsing support for RTL dumps, mirroring the special-cases in
42126 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42127 values, and calling handle_any_trailing_information.
42128 (rtx_reader::read_rtx_operand): Convert return type from void
42129 to rtx, returning return_rtx. Handle case 'e'. Call
42130 finalize_string on XSTR and XTMPL fields.
42131 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42132 "(nil)" values were omitted. Call the postprocess vfunc on the
42134 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42135 class ctor. Initialize m_in_call_function_usage. Call
42136 one_time_initialization.
42137 * rtl-tests.c (selftest::test_uncond_jump): Call
42138 set_new_first_and_last_insn.
42139 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42140 * selftest-rtl.c: New file.
42141 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42142 (selftest::get_insn_by_uid): New decl.
42143 * selftest-run-tests.c (selftest::run_tests): Call
42144 read_rtl_function_c_tests.
42145 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42146 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42149 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42151 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42152 operands in a special way. Assert that pos+len <= mode precision.
42154 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42156 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42157 3 argument Alias with unlimited for the negative form.
42158 (fno-vect-cost-model): Removed.
42160 2017-01-05 Martin Liska <mliska@suse.cz>
42162 * hsa-gen.c (gen_hsa_divmod): New function.
42163 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42165 2017-01-05 Martin Liska <mliska@suse.cz>
42168 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42171 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42173 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42174 small constant length operands.
42176 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42178 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42179 between loop iterations.
42181 2017-01-05 Martin Liska <mliska@suse.cz>
42184 * gimplify.c (gimplify_decl_expr): Compare to
42185 asan_poisoned_variables instread of checking flags.
42186 (gimplify_target_expr): Likewise.
42187 (gimplify_expr): Likewise.
42188 (gimplify_function_tree): Conditionally initialize
42189 asan_poisoned_variables.
42191 2017-01-04 Jeff Law <law@redhat.com>
42193 PR tree-optimizatin/78812
42194 * rtl.h (contains_mem_rtx_p): Prototype.
42195 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42196 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42197 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42198 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42199 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42201 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42203 * input.c (assert_char_at_range): Default-initialize actual_range.
42205 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42207 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42208 to match the caller.
42210 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42212 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42213 insns after final jump in test to emit dummy move.
42215 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42217 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42218 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42220 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42222 * multiple_target.c (create_dispatcher_calls): Init e_next.
42223 * tree-ssa-loop-split.c (split_loop): Init border.
42224 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42227 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42232 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42233 (altivec_register_operand): Do not return true if the operand
42234 contains a SUBREG mixing SImode and SFmode.
42235 (vsx_register_operand): Likewise.
42236 (vsx_reg_sfsubreg_ok): New predicate.
42237 (vfloat_operand): Do not return true if the operand contains a
42238 SUBREG mixing SImode and SFmode.
42239 (vint_operand): Likewise.
42240 (vlogical_operand): Likewise.
42241 (gpc_reg_operand): Likewise.
42242 (int_reg_operand): Likewise.
42243 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42244 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42245 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42247 (rs6000_emit_move_si_sf_subreg): New helper function.
42248 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42249 fixup SUBREGs involving SImode and SFmode.
42250 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42251 numbers for the new peephole2 optimization.
42252 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42253 the GLIBC math library that do AND/IOR/XOR operations on single
42254 precision floating point.
42255 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42256 target macros to say whether we need to avoid SUBREGs mixing
42258 (TARGET_ALLOW_SF_SUBREG): Likewise.
42259 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42260 (UNSPEC_SI_FROM_SF): Likewise.
42261 (iorxor): Change spacing.
42262 (and_ior_xor): New iterator for AND, IOR, and XOR.
42263 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42264 (movdi_from_sf_zero_ext): Likewise.
42265 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42266 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42267 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42268 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42269 (fms<mode>4): Likewise.
42270 (fnma<mode>4): Likewise.
42271 (fnms<mode>4): Likewise.
42272 (nfma<mode>4): Likewise.
42273 (nfms<mode>4): Likewise.
42275 2017-01-04 Marek Polacek <polacek@redhat.com>
42278 * doc/invoke.texi: Document -Wpointer-compare.
42280 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42282 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42285 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42286 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42289 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42292 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42293 documentation of the powerpc_popcntb_ok attribute.
42294 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42295 code to issue warning messages if a requested CPU configuration is
42296 not supported by the binary (assembler and loader) toolchain.
42297 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42298 made to define a built-in function that has been disabled.
42299 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42300 made to define a built-in function that has been disabled.
42301 (altivec_init_builtins): Add comment explaining why definition
42302 of the DST built-in functions is not preceded by an assertion
42303 check. Add assertions to prevent ICE if attempts are made to
42304 define an altivec predicate or an abs* built-in function that has
42306 (htm_init_builtins): Add comment explaining why definition of the
42307 htm built-in functions is not preceded by an assertion check.
42309 2017-01-04 Jeff Law <law@redhat.com>
42311 PR tree-optimizatin/67955
42312 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42313 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42314 the points-to solution does not include pt_null. Use DECL_PT_UID
42317 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42319 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42320 Use gen_int_mode instead of gen_lopwart for const_int operands.
42322 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42324 PR tree-optimization/71563
42325 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42326 out of range value - has low bits known to be zero.
42328 2017-01-04 Alan Modra <amodra@gmail.com>
42330 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42331 * configure: Regenerate.
42332 * config.in: Regenerate.
42334 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42337 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42338 a substring of the message, but strcmp with the whole message. Ifdef
42339 ENABLE_NLS, translate the message first using dgettext.
42341 2017-01-03 Jeff Law <law@redhat.com>
42343 PR tree-optimizatin/78856
42344 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42345 (mark_threaded_blocks): Remove code to truncate thread paths that
42346 cross multiple loop headers. Instead invalidate the cached loop
42347 iteration information and handle case of a thread path walking
42348 into an irreducible region.
42350 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42353 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42354 assertions. Add support for doing the signbit if the IEEE 128-bit
42355 floating point value is in a GPR.
42356 * config/rs6000/rs6000.md (Fsignbit): Delete.
42357 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42358 Update the length attribute if the value is in a GPR.
42359 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42360 the sign or zero extension instruction, since the value is always 0/1.
42361 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42364 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42365 extracting SImode to a GPR register so that we can generate a
42366 store, limit the vector to be in a traditional Altivec register
42367 for the vextuwrx instruction.
42369 2017-01-03 Ian Lance Taylor <iant@google.com>
42371 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42373 2017-01-03 Martin Sebor <msebor@redhat.com>
42375 PR tree-optimization/78696
42376 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42377 precision. Use MPFR for %f for greater fidelity. Correct handling
42379 (pass_sprintf_length::compute_format_length): Set width and precision
42380 specified by asrerisk to void_node for vararg functions.
42381 (try_substitute_return_value): Adjust dump output.
42383 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42385 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42387 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42389 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42390 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42391 -mlra/-mno-lra was passed to the compiler.
42393 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42395 PR rtl-optimization/65618
42396 * emit-rtl.c (try_split): Move initialization of "before" and
42397 "after" to just before the call to emit_insn_after_setloc.
42399 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42401 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42403 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42405 * dwarf2out.c (gen_enumeration_type_die): When
42406 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42408 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42410 PR tree-optimization/78965
42411 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42412 Change first argument from const call_info & to call_info &. For %n
42413 set info.nowrite to false.
42415 PR middle-end/78901
42416 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42417 possibly throwing calls.
42419 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42420 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42421 and fns handling, rather than in a separate case SSA_NAME.
42423 2017-01-02 Jeff Law <law@redhat.com>
42425 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42426 fixes for first_period and second_period variables.
42428 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42431 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42432 (*insvqi_1): New insn pattern.
42433 (*insvqi_1_mem_rex64): Ditto.
42434 (*insvqi_2): Ditto.
42435 (*insvqi_3): Rename from *insvqi.
42437 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42439 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42441 * doc/cfg.texi (Edges): Remove reference to Java.
42442 (Maintaining the CFG): Ditto.
42444 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42446 PR middle-end/77674
42447 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42448 transparent aliases.
42450 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42452 PR middle-end/77484
42453 * predict.def (PRED_CALL): Update hitrate.
42454 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42455 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42456 into direct/indirect/polymorphic variants.
42458 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42460 Update copyright years.
42462 * gcc.c (process_command): Update copyright notice dates.
42463 * gcov-dump.c (print_version): Ditto.
42464 * gcov.c (print_version): Ditto.
42465 * gcov-tool.c (print_version): Ditto.
42466 * gengtype.c (create_file): Ditto.
42467 * doc/cpp.texi: Bump @copying's copyright year.
42468 * doc/cppinternals.texi: Ditto.
42469 * doc/gcc.texi: Ditto.
42470 * doc/gccint.texi: Ditto.
42471 * doc/gcov.texi: Ditto.
42472 * doc/install.texi: Ditto.
42473 * doc/invoke.texi: Ditto.
42475 Copyright (C) 2017 Free Software Foundation, Inc.
42477 Copying and distribution of this file, with or without modification,
42478 are permitted in any medium without royalty provided the copyright
42479 notice and this notice are preserved.